CINXE.COM

<!DOCTYPE html> <html lang="en"> <head> <meta charset="UTF-8">  <meta name="viewport" content="width=device-width, initial-scale=1.0"> <meta name="generator" content="Asciidoctor 1.5.3"> <meta name="author" content="Apache HBase Team"> <title>Apache HBase ™ Reference Guide</title> <link rel="stylesheet" href="./hbase.css"> <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/4.4.0/css/font-awesome.min.css"> <link rel="stylesheet" href="./coderay-asciidoctor.css"> </head> <body class="book toc2 toc-left"> <div id="header"> <h1>Apache HBase ™ Reference Guide</h1> <div class="details"> <span id="author" class="author">Apache HBase Team</span><br> <span id="email" class="email"><<a href="mailto:hbase-dev@lists.apache.org">hbase-dev@lists.apache.org</a>></span><br> <span id="revnumber">version 1.4.11</span> </div> <div id="toc" class="toc2"> <div id="toctitle">Contents</div> <ul class="sectlevel1"> <li><a href="#_preface">Preface</a></li> <li><a href="#_getting_started">Getting Started</a> <ul class="sectlevel1"> <li><a href="#_introduction">1. Introduction</a></li> <li><a href="#quickstart">2. Quick Start - Standalone HBase</a></li> </ul> </li> <li><a href="#configuration">Apache HBase Configuration</a> <ul class="sectlevel1"> <li><a href="#_configuration_files">3. Configuration Files</a></li> <li><a href="#basic.prerequisites">4. Basic Prerequisites</a></li> <li><a href="#standalone_dist">5. HBase run modes: Standalone and Distributed</a></li> <li><a href="#confirm">6. Running and Confirming Your Installation</a></li> <li><a href="#config.files">7. Default Configuration</a></li> <li><a href="#example_config">8. Example Configurations</a></li> <li><a href="#important_configurations">9. The Important Configurations</a></li> <li><a href="#dyn_config">10. Dynamic Configuration</a></li> </ul> </li> <li><a href="#upgrading">Upgrading</a> <ul class="sectlevel1"> <li><a href="#hbase.versioning">11. HBase version number and compatibility</a></li> <li><a href="#_upgrade_paths">12. Upgrade Paths</a></li> </ul> </li> <li><a href="#shell">The Apache HBase Shell</a> <ul class="sectlevel1"> <li><a href="#scripting">13. Scripting with Ruby</a></li> <li><a href="#_running_the_shell_in_non_interactive_mode">14. Running the Shell in Non-Interactive Mode</a></li> <li><a href="#hbase.shell.noninteractive">15. HBase Shell in OS Scripts</a></li> <li><a href="#_read_hbase_shell_commands_from_a_command_file">16. Read HBase Shell Commands from a Command File</a></li> <li><a href="#_passing_vm_options_to_the_shell">17. Passing VM Options to the Shell</a></li> <li><a href="#_overriding_configuration_starting_the_hbase_shell">18. Overriding configuration starting the HBase Shell</a></li> <li><a href="#_shell_tricks">19. Shell Tricks</a></li> </ul> </li> <li><a href="#datamodel">Data Model</a> <ul class="sectlevel1"> <li><a href="#conceptual.view">20. Conceptual View</a></li> <li><a href="#physical.view">21. Physical View</a></li> <li><a href="#_namespace">22. Namespace</a></li> <li><a href="#_table">23. Table</a></li> <li><a href="#_row">24. Row</a></li> <li><a href="#columnfamily">25. Column Family</a></li> <li><a href="#_cells">26. Cells</a></li> <li><a href="#_data_model_operations">27. Data Model Operations</a></li> <li><a href="#versions">28. Versions</a></li> <li><a href="#dm.sort">29. Sort Order</a></li> <li><a href="#dm.column.metadata">30. Column Metadata</a></li> <li><a href="#_joins">31. Joins</a></li> <li><a href="#_acid">32. ACID</a></li> </ul> </li> <li><a href="#schema">HBase and Schema Design</a> <ul class="sectlevel1"> <li><a href="#schema.creation">33. Schema Creation</a></li> <li><a href="#number.of.cfs">34. On the number of column families</a></li> <li><a href="#rowkey.design">35. Rowkey Design</a></li> <li><a href="#schema.versions">36. Number of Versions</a></li> <li><a href="#supported.datatypes">37. Supported Datatypes</a></li> <li><a href="#schema.joins">38. Joins</a></li> <li><a href="#ttl">39. Time To Live (TTL)</a></li> <li><a href="#cf.keep.deleted">40. Keeping Deleted Cells</a></li> <li><a href="#secondary.indexes">41. Secondary Indexes and Alternate Query Paths</a></li> <li><a href="#_constraints">42. Constraints</a></li> <li><a href="#schema.casestudies">43. Schema Design Case Studies</a></li> <li><a href="#schema.ops">44. Operational and Performance Configuration Options</a></li> </ul> </li> <li><a href="#mapreduce">HBase and MapReduce</a> <ul class="sectlevel1"> <li><a href="#hbase.mapreduce.classpath">45. HBase, MapReduce, and the CLASSPATH</a></li> <li><a href="#_mapreduce_scan_caching">46. MapReduce Scan Caching</a></li> <li><a href="#_bundled_hbase_mapreduce_jobs">47. Bundled HBase MapReduce Jobs</a></li> <li><a href="#_hbase_as_a_mapreduce_job_data_source_and_data_sink">48. HBase as a MapReduce Job Data Source and Data Sink</a></li> <li><a href="#_writing_hfiles_directly_during_bulk_import">49. Writing HFiles Directly During Bulk Import</a></li> <li><a href="#_rowcounter_example">50. RowCounter Example</a></li> <li><a href="#splitter">51. Map-Task Splitting</a></li> <li><a href="#mapreduce.example">52. HBase MapReduce Examples</a></li> <li><a href="#mapreduce.htable.access">53. Accessing Other HBase Tables in a MapReduce Job</a></li> <li><a href="#mapreduce.specex">54. Speculative Execution</a></li> </ul> </li> <li><a href="#security">Securing Apache HBase</a> <ul class="sectlevel1"> <li><a href="#_using_secure_http_https_for_the_web_ui">55. Using Secure HTTP (HTTPS) for the Web UI</a></li> <li><a href="#hbase.secure.spnego.ui">56. Using SPNEGO for Kerberos authentication with Web UIs</a></li> <li><a href="#hbase.secure.configuration">57. Secure Client Access to Apache HBase</a></li> <li><a href="#hbase.secure.simpleconfiguration">58. Simple User Access to Apache HBase</a></li> <li><a href="#_securing_access_to_hdfs_and_zookeeper">59. Securing Access to HDFS and ZooKeeper</a></li> <li><a href="#_securing_access_to_your_data">60. Securing Access To Your Data</a></li> <li><a href="#security.example.config">61. Security Configuration Example</a></li> </ul> </li> <li><a href="#_architecture">Architecture</a> <ul class="sectlevel1"> <li><a href="#arch.overview">62. Overview</a></li> <li><a href="#arch.catalog">63. Catalog Tables</a></li> <li><a href="#architecture.client">64. Client</a></li> <li><a href="#client.filter">65. Client Request Filters</a></li> <li><a href="#_master">66. Master</a></li> <li><a href="#regionserver.arch">67. RegionServer</a></li> <li><a href="#regions.arch">68. Regions</a></li> <li><a href="#arch.bulk.load">69. Bulk Loading</a></li> <li><a href="#arch.hdfs">70. HDFS</a></li> <li><a href="#arch.timelineconsistent.reads">71. Timeline-consistent High Available Reads</a></li> </ul> </li> <li><a href="#hbase_apis">Apache HBase APIs</a> <ul class="sectlevel1"> <li><a href="#_examples">72. Examples</a></li> </ul> </li> <li><a href="#external_apis">Apache HBase External APIs</a> <ul class="sectlevel1"> <li><a href="#nonjava.jvm">73. Non-Java Languages Talking to the JVM</a></li> <li><a href="#_rest">74. REST</a></li> <li><a href="#_thrift">75. Thrift</a></li> <li><a href="#c">76. C/C++ Apache HBase Client</a></li> </ul> </li> <li><a href="#thrift">Thrift API and Filter Language</a> <ul class="sectlevel1"> <li><a href="#thrift.filter_language">77. Filter Language</a></li> </ul> </li> <li><a href="#cp">Apache HBase Coprocessors</a> <ul class="sectlevel1"> <li><a href="#_coprocessor_framework">78. Coprocessor Framework</a></li> <li><a href="#_examples_2">79. Examples</a></li> <li><a href="#_building_a_coprocessor">80. Building A Coprocessor</a></li> <li><a href="#_check_the_status_of_a_coprocessor">81. Check the Status of a Coprocessor</a></li> <li><a href="#_monitor_time_spent_in_coprocessors">82. Monitor Time Spent in Coprocessors</a></li> </ul> </li> <li><a href="#performance">Apache HBase Performance Tuning</a> <ul class="sectlevel1"> <li><a href="#perf.os">83. Operating System</a></li> <li><a href="#perf.network">84. Network</a></li> <li><a href="#jvm">85. Java</a></li> <li><a href="#perf.configurations">86. HBase Configurations</a></li> <li><a href="#perf.zookeeper">87. ZooKeeper</a></li> <li><a href="#perf.schema">88. Schema Design</a></li> <li><a href="#perf.general">89. HBase General Patterns</a></li> <li><a href="#perf.writing">90. Writing to HBase</a></li> <li><a href="#perf.reading">91. Reading from HBase</a></li> <li><a href="#perf.deleting">92. Deleting from HBase</a></li> <li><a href="#perf.hdfs">93. HDFS</a></li> <li><a href="#perf.ec2">94. Amazon EC2</a></li> <li><a href="#perf.hbase.mr.cluster">95. Collocating HBase and MapReduce</a></li> <li><a href="#perf.casestudy">96. Case Studies</a></li> </ul> </li> <li><a href="#profiler">Profiler Servlet</a> <ul class="sectlevel1"> <li><a href="#_background">97. Background</a></li> <li><a href="#_prerequisites_2">98. Prerequisites</a></li> <li><a href="#_usage">99. Usage</a></li> <li><a href="#_ui">100. UI</a></li> <li><a href="#_notes">101. Notes</a></li> </ul> </li> <li><a href="#trouble">Troubleshooting and Debugging Apache HBase</a> <ul class="sectlevel1"> <li><a href="#trouble.general">102. General Guidelines</a></li> <li><a href="#trouble.log">103. Logs</a></li> <li><a href="#trouble.resources">104. Resources</a></li> <li><a href="#trouble.tools">105. Tools</a></li> <li><a href="#trouble.client">106. Client</a></li> <li><a href="#trouble.mapreduce">107. MapReduce</a></li> <li><a href="#trouble.namenode">108. NameNode</a></li> <li><a href="#trouble.network">109. Network</a></li> <li><a href="#trouble.rs">110. RegionServer</a></li> <li><a href="#trouble.master">111. Master</a></li> <li><a href="#trouble.zookeeper">112. ZooKeeper</a></li> <li><a href="#trouble.ec2">113. Amazon EC2</a></li> <li><a href="#trouble.versions">114. HBase and Hadoop version issues</a></li> <li><a href="#_ipc_configuration_conflicts_with_hadoop">115. IPC Configuration Conflicts with Hadoop</a></li> <li><a href="#_hbase_and_hdfs">116. HBase and HDFS</a></li> <li><a href="#trouble.tests">117. Running unit or integration tests</a></li> <li><a href="#trouble.casestudy">118. Case Studies</a></li> <li><a href="#trouble.crypto">119. Cryptographic Features</a></li> <li><a href="#_operating_system_specific_issues">120. Operating System Specific Issues</a></li> <li><a href="#_jdk_issues">121. JDK Issues</a></li> </ul> </li> <li><a href="#casestudies">Apache HBase Case Studies</a> <ul class="sectlevel1"> <li><a href="#casestudies.overview">122. Overview</a></li> <li><a href="#casestudies.schema">123. Schema Design</a></li> <li><a href="#casestudies.perftroub">124. Performance/Troubleshooting</a></li> </ul> </li> <li><a href="#ops_mgt">Apache HBase Operational Management</a> <ul class="sectlevel1"> <li><a href="#tools">125. HBase Tools and Utilities</a></li> <li><a href="#ops.regionmgt">126. Region Management</a></li> <li><a href="#node.management">127. Node Management</a></li> <li><a href="#_hbase_metrics">128. HBase Metrics</a></li> <li><a href="#ops.monitoring">129. HBase Monitoring</a></li> <li><a href="#_cluster_replication">130. Cluster Replication</a></li> <li><a href="#ops.backup">131. HBase Backup</a></li> <li><a href="#ops.snapshots">132. HBase Snapshots</a></li> <li><a href="#ops.capacity">133. Capacity Planning and Region Sizing</a></li> <li><a href="#table.rename">134. Table Rename</a></li> </ul> </li> <li><a href="#developer">Building and Developing Apache HBase</a> <ul class="sectlevel1"> <li><a href="#getting.involved">135. Getting Involved</a></li> <li><a href="#repos">136. Apache HBase Repositories</a></li> <li><a href="#_ides">137. IDEs</a></li> <li><a href="#build">138. Building Apache HBase</a></li> <li><a href="#releasing">139. Releasing Apache HBase</a></li> <li><a href="#hbase.rc.voting">140. Voting on Release Candidates</a></li> <li><a href="#documentation">141. Generating the HBase Reference Guide</a></li> <li><a href="#hbase.org">142. Updating <a href="https://hbase.apache.org">hbase.apache.org</a></a></li> <li><a href="#hbase.tests">143. Tests</a></li> <li><a href="#developing">144. Developer Guidelines</a></li> </ul> </li> <li><a href="#unit.tests">Unit Testing HBase Applications</a> <ul class="sectlevel1"> <li><a href="#_junit">145. JUnit</a></li> <li><a href="#_mockito">146. Mockito</a></li> <li><a href="#_mrunit">147. MRUnit</a></li> <li><a href="#_integration_testing_with_a_hbase_mini_cluster">148. Integration Testing with a HBase Mini-Cluster</a></li> </ul> </li> <li><a href="#zookeeper">ZooKeeper</a> <ul class="sectlevel1"> <li><a href="#_using_existing_zookeeper_ensemble">149. Using existing ZooKeeper ensemble</a></li> <li><a href="#zk.sasl.auth">150. SASL Authentication with ZooKeeper</a></li> </ul> </li> <li><a href="#community">Community</a> <ul class="sectlevel1"> <li><a href="#_decisions">151. Decisions</a></li> <li><a href="#community.roles">152. Community Roles</a></li> <li><a href="#hbase.commit.msg.format">153. Commit Message format</a></li> </ul> </li> <li><a href="#_appendix">Appendix</a> <ul class="sectlevel1"> <li><a href="#appendix_contributing_to_documentation">Appendix A: Contributing to Documentation</a></li> <li><a href="#faq">Appendix B: FAQ</a></li> <li><a href="#hbck.in.depth">Appendix C: hbck In Depth</a></li> <li><a href="#appendix_acl_matrix">Appendix D: Access Control Matrix</a></li> <li><a href="#compression">Appendix E: Compression and Data Block Encoding In HBase</a></li> <li><a href="#data.block.encoding.enable">154. Enable Data Block Encoding</a></li> <li><a href="#sql">Appendix F: SQL over HBase</a></li> <li><a href="#_ycsb">Appendix G: YCSB</a></li> <li><a href="#_hfile_format_2">Appendix H: HFile format</a></li> <li><a href="#other.info">Appendix I: Other Information About HBase</a></li> <li><a href="#hbase.history">Appendix J: HBase History</a></li> <li><a href="#asf">Appendix K: HBase and the Apache Software Foundation</a></li> <li><a href="#orca">Appendix L: Apache HBase Orca</a></li> <li><a href="#tracing">Appendix M: Enabling Dapper-like Tracing in HBase</a></li> <li><a href="#tracing.client.modifications">155. Client Modifications</a></li> <li><a href="#tracing.client.shell">156. Tracing from HBase Shell</a></li> <li><a href="#hbase.rpc">Appendix N: 0.95 RPC Specification</a></li> </ul> </li> </ul> </div> </div> <div id="content"> <div id="preamble"> <div class="sectionbody"> <div> <a href="http://hbase.apache.org"><img src="images/hbase_logo_with_orca.png" alt="Apache HBase Logo" /></a> </div> </div> </div> <div class="sect1"> <h2 id="_preface"><a class="anchor" href="#_preface"></a>Preface</h2> <div class="sectionbody"> <div class="paragraph"> <p>This is the official reference guide for the <a href="http://hbase.apache.org/">HBase</a> version it ships with.</p> </div> <div class="paragraph"> <p>Herein you will find either the definitive documentation on an HBase topic as of its standing when the referenced HBase version shipped, or it will point to the location in <a href="http://hbase.apache.org/apidocs/index.html">Javadoc</a>, <a href="https://issues.apache.org/jira/browse/HBASE">JIRA</a> or <a href="http://wiki.apache.org/hadoop/Hbase">wiki</a> where the pertinent information can be found.</p> </div> <div class="paragraph"> <div class="title">About This Guide</div> <p>This reference guide is a work in progress. The source for this guide can be found in the _src/main/asciidoc directory of the HBase source. This reference guide is marked up using <a href="http://asciidoc.org/">AsciiDoc</a> from which the finished guide is generated as part of the 'site' build target. Run</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn site</code></pre> </div> </div> <div class="paragraph"> <p>to generate this documentation. Amendments and improvements to the documentation are welcomed. Click <a href="https://issues.apache.org/jira/secure/CreateIssueDetails!init.jspa?pid=12310753&issuetype=1&components=12312132&summary=SHORT+DESCRIPTION">this link</a> to file a new documentation bug against Apache HBase with some values pre-selected.</p> </div> <div class="paragraph"> <div class="title">Contributing to the Documentation</div> <p>For an overview of AsciiDoc and suggestions to get started contributing to the documentation, see the <a href="#appendix_contributing_to_documentation">relevant section later in this documentation</a>.</p> </div> <div class="paragraph"> <div class="title">Heads-up if this is your first foray into the world of distributed computing…</div> <p>If this is your first foray into the wonderful world of Distributed Computing, then you are in for some interesting times. First off, distributed systems are hard; making a distributed system hum requires a disparate skillset that spans systems (hardware and software) and networking.</p> </div> <div class="paragraph"> <p>Your cluster’s operation can hiccup because of any of a myriad set of reasons from bugs in HBase itself through misconfigurations — misconfiguration of HBase but also operating system misconfigurations — through to hardware problems whether it be a bug in your network card drivers or an underprovisioned RAM bus (to mention two recent examples of hardware issues that manifested as "HBase is slow"). You will also need to do a recalibration if up to this your computing has been bound to a single box. Here is one good starting point: <a href="http://en.wikipedia.org/wiki/Fallacies_of_Distributed_Computing">Fallacies of Distributed Computing</a>.</p> </div> <div class="paragraph"> <p>That said, you are welcome.<br> It’s a fun place to be.<br> Yours, the HBase Community.</p> </div> <div class="paragraph"> <div class="title">Reporting Bugs</div> <p>Please use <a href="https://issues.apache.org/jira/browse/hbase">JIRA</a> to report non-security-related bugs.</p> </div> <div class="paragraph"> <p>To protect existing HBase installations from new vulnerabilities, please <strong>do not</strong> use JIRA to report security-related bugs. Instead, send your report to the mailing list <a href="mailto:private@apache.org">private@apache.org</a>, which allows anyone to send messages, but restricts who can read them. Someone on that list will contact you to follow up on your report.</p> </div> </div> </div> <h1 id="_getting_started" class="sect0"><a class="anchor" href="#_getting_started"></a>Getting Started</h1> <div class="sect1"> <h2 id="_introduction"><a class="anchor" href="#_introduction"></a>1. Introduction</h2> <div class="sectionbody"> <div class="paragraph"> <p><a href="#quickstart">Quickstart</a> will get you up and running on a single-node, standalone instance of HBase, followed by a pseudo-distributed single-machine instance, and finally a fully-distributed cluster.</p> </div> </div> </div> <div class="sect1"> <h2 id="quickstart"><a class="anchor" href="#quickstart"></a>2. Quick Start - Standalone HBase</h2> <div class="sectionbody"> <div class="paragraph"> <p>This guide describes the setup of a standalone HBase instance running against the local filesystem. This is not an appropriate configuration for a production instance of HBase, but will allow you to experiment with HBase. This section shows you how to create a table in HBase using the <code>hbase shell</code> CLI, insert rows into the table, perform put and scan operations against the table, enable or disable the table, and start and stop HBase. Apart from downloading HBase, this procedure should take less than 10 minutes.</p> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="title">Local Filesystem and Durability</div> <em>The following is fixed in HBase 0.98.3 and beyond. See <a href="https://issues.apache.org/jira/browse/HBASE-11272">HBASE-11272</a> and <a href="https://issues.apache.org/jira/browse/HBASE-11218">HBASE-11218</a>.</em> </td> </tr> </table> </div> <div class="paragraph"> <p>Using HBase with a local filesystem does not guarantee durability. The HDFS local filesystem implementation will lose edits if files are not properly closed. This is very likely to happen when you are experimenting with new software, starting and stopping the daemons often and not always cleanly. You need to run HBase on HDFS to ensure all writes are preserved. Running against the local filesystem is intended as a shortcut to get you familiar with how the general system works, as the very first phase of evaluation. See <a href="https://issues.apache.org/jira/browse/HBASE-3696">HBASE-3696</a> and its associated issues for more details about the issues of running on the local filesystem.</p> </div> <div id="loopback.ip" class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Loopback IP - HBase 0.94.x and earlier</div> <em>The below advice is for hbase-0.94.x and older versions only. This is fixed in hbase-0.96.0 and beyond.</em> </td> </tr> </table> </div> <div class="paragraph"> <p>Prior to HBase 0.94.x, HBase expected the loopback IP address to be 127.0.0.1. Ubuntu and some other distributions default to 127.0.1.1 and this will cause problems for you. See <a href="http://devving.com/?p=414">Why does HBase care about /etc/hosts?</a> for detail</p> </div> <div class="exampleblock"> <div class="title">Example 1. Example /etc/hosts File for Ubuntu</div> <div class="content"> <div class="paragraph"> <p>The following <em>/etc/hosts</em> file works correctly for HBase 0.94.x and earlier, on Ubuntu. Use this as a template if you run into trouble.</p> </div> <div class="listingblock"> <div class="content"> <pre>127.0.0.1 localhost 127.0.0.1 ubuntu.ubuntu-domain ubuntu</pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="_jdk_version_requirements"><a class="anchor" href="#_jdk_version_requirements"></a>2.1. JDK Version Requirements</h3> <div class="paragraph"> <p>HBase requires that a JDK be installed. See <a href="#java">Java</a> for information about supported JDK versions.</p> </div> </div> <div class="sect2"> <h3 id="_get_started_with_hbase"><a class="anchor" href="#_get_started_with_hbase"></a>2.2. Get Started with HBase</h3> <div class="olist arabic"> <div class="title">Procedure: Download, Configure, and Start HBase</div> <ol class="arabic"> <li> <p>Choose a download site from this list of <a href="http://www.apache.org/dyn/closer.cgi/hbase/">Apache Download Mirrors</a>. Click on the suggested top link. This will take you to a mirror of <em>HBase Releases</em>. Click on the folder named <em>stable</em> and then download the binary file that ends in <em>.tar.gz</em> to your local filesystem. Be sure to choose the version that corresponds with the version of Hadoop you are likely to use later. In most cases, you should choose the file for Hadoop 2, which will be called something like <em>hbase-0.98.3-hadoop2-bin.tar.gz</em>. Do not download the file ending in <em>src.tar.gz</em> for now.</p> </li> <li> <p>Extract the downloaded file, and change to the newly-created directory.</p> <div class="listingblock"> <div class="content"> <pre>$ tar xzvf hbase-<?eval ${project.version}?>-hadoop2-bin.tar.gz $ cd hbase-<?eval ${project.version}?>-hadoop2/</pre> </div> </div> </li> <li> <p>For HBase 0.98.5 and later, you are required to set the <code>JAVA_HOME</code> environment variable before starting HBase. Prior to 0.98.5, HBase attempted to detect the location of Java if the variables was not set. You can set the variable via your operating system’s usual mechanism, but HBase provides a central mechanism, <em>conf/hbase-env.sh</em>. Edit this file, uncomment the line starting with <code>JAVA_HOME</code>, and set it to the appropriate location for your operating system. The <code>JAVA_HOME</code> variable should be set to a directory which contains the executable file <em>bin/java</em>. Most modern Linux operating systems provide a mechanism, such as /usr/bin/alternatives on RHEL or CentOS, for transparently switching between versions of executables such as Java. In this case, you can set <code>JAVA_HOME</code> to the directory containing the symbolic link to <em>bin/java</em>, which is usually <em>/usr</em>.</p> <div class="listingblock"> <div class="content"> <pre>JAVA_HOME=/usr</pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> These instructions assume that each node of your cluster uses the same configuration. If this is not the case, you may need to set <code>JAVA_HOME</code> separately for each node. </td> </tr> </table> </div> </li> <li> <p>Edit <em>conf/hbase-site.xml</em>, which is the main HBase configuration file. At this time, you only need to specify the directory on the local filesystem where HBase and ZooKeeper write data. By default, a new directory is created under /tmp. Many servers are configured to delete the contents of <em>/tmp</em> upon reboot, so you should store the data elsewhere. The following configuration will store HBase’s data in the <em>hbase</em> directory, in the home directory of the user called <code>testuser</code>. Paste the <code><property></code> tags beneath the <code><configuration></code> tags, which should be empty in a new HBase install.</p> <div class="exampleblock"> <div class="title">Example 2. Example <em>hbase-site.xml</em> for Standalone HBase</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><configuration></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rootdir<span class="tag"></name></span> <span class="tag"><value></span>file:///home/testuser/hbase<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.property.dataDir<span class="tag"></name></span> <span class="tag"><value></span>/home/testuser/zookeeper<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"></configuration></span></code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>You do not need to create the HBase data directory. HBase will do this for you. If you create the directory, HBase will attempt to do a migration, which is not what you want.</p> </div> </li> <li> <p>The <em>bin/start-hbase.sh</em> script is provided as a convenient way to start HBase. Issue the command, and if all goes well, a message is logged to standard output showing that HBase started successfully. You can use the <code>jps</code> command to verify that you have one running process called <code>HMaster</code>. In standalone mode HBase runs all daemons within this single JVM, i.e. the HMaster, a single HRegionServer, and the ZooKeeper daemon.</p> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Java needs to be installed and available. If you get an error indicating that Java is not installed, but it is on your system, perhaps in a non-standard location, edit the <em>conf/hbase-env.sh</em> file and modify the <code>JAVA_HOME</code> setting to point to the directory that contains <em>bin/java</em> your system. </td> </tr> </table> </div> </li> </ol> </div> <div id="shell_exercises" class="olist arabic"> <div class="title">Procedure: Use HBase For the First Time</div> <ol class="arabic"> <li> <p>Connect to HBase.</p> <div class="paragraph"> <p>Connect to your running instance of HBase using the <code>hbase shell</code> command, located in the <em class="path">bin/</em> directory of your HBase install. In this example, some usage and version information that is printed when you start HBase Shell has been omitted. The HBase Shell prompt ends with a <code>></code> character.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase shell hbase(main):001:0></pre> </div> </div> </li> <li> <p>Display HBase Shell Help Text.</p> <div class="paragraph"> <p>Type <code>help</code> and press Enter, to display some basic usage information for HBase Shell, as well as several example commands. Notice that table names, rows, columns all must be enclosed in quote characters.</p> </div> </li> <li> <p>Create a table.</p> <div class="paragraph"> <p>Use the <code>create</code> command to create a new table. You must specify the table name and the ColumnFamily name.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):001:0> create 'test', 'cf' 0 row(s) in 0.4170 seconds => Hbase::Table - test</pre> </div> </div> </li> <li> <p>List Information About your Table</p> <div class="paragraph"> <p>Use the <code>list</code> command to</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):002:0> list 'test' TABLE test 1 row(s) in 0.0180 seconds => ["test"]</pre> </div> </div> </li> <li> <p>Put data into your table.</p> <div class="paragraph"> <p>To put data into your table, use the <code>put</code> command.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):003:0> put 'test', 'row1', 'cf:a', 'value1' 0 row(s) in 0.0850 seconds hbase(main):004:0> put 'test', 'row2', 'cf:b', 'value2' 0 row(s) in 0.0110 seconds hbase(main):005:0> put 'test', 'row3', 'cf:c', 'value3' 0 row(s) in 0.0100 seconds</pre> </div> </div> <div class="paragraph"> <p>Here, we insert three values, one at a time. The first insert is at <code>row1</code>, column <code>cf:a</code>, with a value of <code>value1</code>. Columns in HBase are comprised of a column family prefix, <code>cf</code> in this example, followed by a colon and then a column qualifier suffix, <code>a</code> in this case.</p> </div> </li> <li> <p>Scan the table for all data at once.</p> <div class="paragraph"> <p>One of the ways to get data from HBase is to scan. Use the <code>scan</code> command to scan the table for data. You can limit your scan, but for now, all data is fetched.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):006:0> scan 'test' ROW COLUMN+CELL row1 column=cf:a, timestamp=1421762485768, value=value1 row2 column=cf:b, timestamp=1421762491785, value=value2 row3 column=cf:c, timestamp=1421762496210, value=value3 3 row(s) in 0.0230 seconds</pre> </div> </div> </li> <li> <p>Get a single row of data.</p> <div class="paragraph"> <p>To get a single row of data at a time, use the <code>get</code> command.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):007:0> get 'test', 'row1' COLUMN CELL cf:a timestamp=1421762485768, value=value1 1 row(s) in 0.0350 seconds</pre> </div> </div> </li> <li> <p>Disable a table.</p> <div class="paragraph"> <p>If you want to delete a table or change its settings, as well as in some other situations, you need to disable the table first, using the <code>disable</code> command. You can re-enable it using the <code>enable</code> command.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):008:0> disable 'test' 0 row(s) in 1.1820 seconds hbase(main):009:0> enable 'test' 0 row(s) in 0.1770 seconds</pre> </div> </div> <div class="paragraph"> <p>Disable the table again if you tested the <code>enable</code> command above:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):010:0> disable 'test' 0 row(s) in 1.1820 seconds</pre> </div> </div> </li> <li> <p>Drop the table.</p> <div class="paragraph"> <p>To drop (delete) a table, use the <code>drop</code> command.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):011:0> drop 'test' 0 row(s) in 0.1370 seconds</pre> </div> </div> </li> <li> <p>Exit the HBase Shell.</p> <div class="paragraph"> <p>To exit the HBase Shell and disconnect from your cluster, use the <code>quit</code> command. HBase is still running in the background.</p> </div> </li> </ol> </div> <div class="olist arabic"> <div class="title">Procedure: Stop HBase</div> <ol class="arabic"> <li> <p>In the same way that the <em>bin/start-hbase.sh</em> script is provided to conveniently start all HBase daemons, the <em>bin/stop-hbase.sh</em> script stops them.</p> <div class="listingblock"> <div class="content"> <pre>$ ./bin/stop-hbase.sh stopping hbase.................... $</pre> </div> </div> </li> <li> <p>After issuing the command, it can take several minutes for the processes to shut down. Use the <code>jps</code> to be sure that the HMaster and HRegionServer processes are shut down.</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="quickstart_pseudo"><a class="anchor" href="#quickstart_pseudo"></a>2.3. Intermediate - Pseudo-Distributed Local Install</h3> <div class="paragraph"> <p>After working your way through <a href="#quickstart">quickstart</a>, you can re-configure HBase to run in pseudo-distributed mode. Pseudo-distributed mode means that HBase still runs completely on a single host, but each HBase daemon (HMaster, HRegionServer, and Zookeeper) runs as a separate process. By default, unless you configure the <code>hbase.rootdir</code> property as described in <a href="#quickstart">quickstart</a>, your data is still stored in <em>/tmp/</em>. In this walk-through, we store your data in HDFS instead, assuming you have HDFS available. You can skip the HDFS configuration to continue storing your data in the local filesystem.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Hadoop Configuration</div> <div class="paragraph"> <p>This procedure assumes that you have configured Hadoop and HDFS on your local system and or a remote system, and that they are running and available. It also assumes you are using Hadoop 2. Currently, the documentation on the Hadoop website does not include a quick start for Hadoop 2, but the guide at link:http://www.alexjf.net/blog/distributed-systems/hadoop-yarn-installation-definitive-guide is a good starting point.</p> </div> </td> </tr> </table> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Stop HBase if it is running.</p> <div class="paragraph"> <p>If you have just finished <a href="#quickstart">quickstart</a> and HBase is still running, stop it. This procedure will create a totally new directory where HBase will store its data, so any databases you created before will be lost.</p> </div> </li> <li> <p>Configure HBase.</p> <div class="paragraph"> <p>Edit the <em>hbase-site.xml</em> configuration. First, add the following property. which directs HBase to run in distributed mode, with one JVM instance per daemon.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.cluster.distributed<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Next, change the <code>hbase.rootdir</code> from the local filesystem to the address of your HDFS instance, using the <code>hdfs:////</code> URI syntax. In this example, HDFS is running on the localhost at port 8020.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rootdir<span class="tag"></name></span> <span class="tag"><value></span>hdfs://localhost:8020/hbase<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>You do not need to create the directory in HDFS. HBase will do this for you. If you create the directory, HBase will attempt to do a migration, which is not what you want.</p> </div> </li> <li> <p>Start HBase.</p> <div class="paragraph"> <p>Use the <em>bin/start-hbase.sh</em> command to start HBase. If your system is configured correctly, the <code>jps</code> command should show the HMaster and HRegionServer processes running.</p> </div> </li> <li> <p>Check the HBase directory in HDFS.</p> <div class="paragraph"> <p>If everything worked correctly, HBase created its directory in HDFS. In the configuration above, it is stored in <em>/hbase/</em> on HDFS. You can use the <code>hadoop fs</code> command in Hadoop’s <em>bin/</em> directory to list this directory.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hadoop fs -ls /hbase Found 7 items drwxr-xr-x - hbase users 0 2014-06-25 18:58 /hbase/.tmp drwxr-xr-x - hbase users 0 2014-06-25 21:49 /hbase/WALs drwxr-xr-x - hbase users 0 2014-06-25 18:48 /hbase/corrupt drwxr-xr-x - hbase users 0 2014-06-25 18:58 /hbase/data -rw-r--r-- 3 hbase users 42 2014-06-25 18:41 /hbase/hbase.id -rw-r--r-- 3 hbase users 7 2014-06-25 18:41 /hbase/hbase.version drwxr-xr-x - hbase users 0 2014-06-25 21:49 /hbase/oldWALs</pre> </div> </div> </li> <li> <p>Create a table and populate it with data.</p> <div class="paragraph"> <p>You can use the HBase Shell to create a table, populate it with data, scan and get values from it, using the same procedure as in <a href="#shell_exercises">shell exercises</a>.</p> </div> </li> <li> <p>Start and stop a backup HBase Master (HMaster) server.</p> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Running multiple HMaster instances on the same hardware does not make sense in a production environment, in the same way that running a pseudo-distributed cluster does not make sense for production. This step is offered for testing and learning purposes only. </td> </tr> </table> </div> <div class="paragraph"> <p>The HMaster server controls the HBase cluster. You can start up to 9 backup HMaster servers, which makes 10 total HMasters, counting the primary. To start a backup HMaster, use the <code>local-master-backup.sh</code>. For each backup master you want to start, add a parameter representing the port offset for that master. Each HMaster uses three ports (16010, 16020, and 16030 by default). The port offset is added to these ports, so using an offset of 2, the backup HMaster would use ports 16012, 16022, and 16032. The following command starts 3 backup servers using ports 16012/16022/16032, 16013/16023/16033, and 16015/16025/16035.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/local-master-backup.sh 2 3 5</pre> </div> </div> <div class="paragraph"> <p>To kill a backup master without killing the entire cluster, you need to find its process ID (PID). The PID is stored in a file with a name like <em>/tmp/hbase-USER-X-master.pid</em>. The only contents of the file is the PID. You can use the <code>kill -9</code> command to kill that PID. The following command will kill the master with port offset 1, but leave the cluster running:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ cat /tmp/hbase-testuser-1-master.pid |xargs kill -9</pre> </div> </div> </li> <li> <p>Start and stop additional RegionServers</p> <div class="paragraph"> <p>The HRegionServer manages the data in its StoreFiles as directed by the HMaster. Generally, one HRegionServer runs per node in the cluster. Running multiple HRegionServers on the same system can be useful for testing in pseudo-distributed mode. The <code>local-regionservers.sh</code> command allows you to run multiple RegionServers. It works in a similar way to the <code>local-master-backup.sh</code> command, in that each parameter you provide represents the port offset for an instance. Each RegionServer requires two ports, and the default ports are 16020 and 16030. However, the base ports for additional RegionServers are not the default ports since the default ports are used by the HMaster, which is also a RegionServer since HBase version 1.0.0. The base ports are 16200 and 16300 instead. You can run 99 additional RegionServers that are not a HMaster or backup HMaster, on a server. The following command starts four additional RegionServers, running on sequential ports starting at 16202/16302 (base ports 16200/16300 plus 2).</p> </div> <div class="listingblock"> <div class="content"> <pre>$ .bin/local-regionservers.sh start 2 3 4 5</pre> </div> </div> <div class="paragraph"> <p>To stop a RegionServer manually, use the <code>local-regionservers.sh</code> command with the <code>stop</code> parameter and the offset of the server to stop.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ .bin/local-regionservers.sh stop 3</pre> </div> </div> </li> <li> <p>Stop HBase.</p> <div class="paragraph"> <p>You can stop HBase the same way as in the <a href="#quickstart">quickstart</a> procedure, using the <em>bin/stop-hbase.sh</em> command.</p> </div> </li> </ol> </div> </div> <div class="sect2"> <h3 id="quickstart_fully_distributed"><a class="anchor" href="#quickstart_fully_distributed"></a>2.4. Advanced - Fully Distributed</h3> <div class="paragraph"> <p>In reality, you need a fully-distributed configuration to fully test HBase and to use it in real-world scenarios. In a distributed configuration, the cluster contains multiple nodes, each of which runs one or more HBase daemon. These include primary and backup Master instances, multiple Zookeeper nodes, and multiple RegionServer nodes.</p> </div> <div class="paragraph"> <p>This advanced quickstart adds two more nodes to your cluster. The architecture will be as follows:</p> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 1. Distributed Cluster Demo Architecture</caption> <colgroup> <col style="width: 25%;"> <col style="width: 25%;"> <col style="width: 25%;"> <col style="width: 25%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Node Name</th> <th class="tableblock halign-left valign-top">Master</th> <th class="tableblock halign-left valign-top">ZooKeeper</th> <th class="tableblock halign-left valign-top">RegionServer</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">node-a.example.com</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">no</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">node-b.example.com</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">backup</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">node-c.example.com</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">no</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> </tr> </tbody> </table> <div class="paragraph"> <p>This quickstart assumes that each node is a virtual machine and that they are all on the same network. It builds upon the previous quickstart, <a href="#quickstart_pseudo">Intermediate - Pseudo-Distributed Local Install</a>, assuming that the system you configured in that procedure is now <code>node-a</code>. Stop HBase on <code>node-a</code> before continuing.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Be sure that all the nodes have full access to communicate, and that no firewall rules are in place which could prevent them from talking to each other. If you see any errors like <code>no route to host</code>, check your firewall. </td> </tr> </table> </div> <div id="passwordless.ssh.quickstart" class="paragraph"> <div class="title">Procedure: Configure Passwordless SSH Access</div> <p><code>node-a</code> needs to be able to log into <code>node-b</code> and <code>node-c</code> (and to itself) in order to start the daemons. The easiest way to accomplish this is to use the same username on all hosts, and configure password-less SSH login from <code>node-a</code> to each of the others.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>On <code>node-a</code>, generate a key pair.</p> <div class="paragraph"> <p>While logged in as the user who will run HBase, generate a SSH key pair, using the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ssh-keygen -t rsa</code></pre> </div> </div> <div class="paragraph"> <p>If the command succeeds, the location of the key pair is printed to standard output. The default name of the public key is <em>id_rsa.pub</em>.</p> </div> </li> <li> <p>Create the directory that will hold the shared keys on the other nodes.</p> <div class="paragraph"> <p>On <code>node-b</code> and <code>node-c</code>, log in as the HBase user and create a <em>.ssh/</em> directory in the user’s home directory, if it does not already exist. If it already exists, be aware that it may already contain other keys.</p> </div> </li> <li> <p>Copy the public key to the other nodes.</p> <div class="paragraph"> <p>Securely copy the public key from <code>node-a</code> to each of the nodes, by using the <code>scp</code> or some other secure means. On each of the other nodes, create a new file called <em>.ssh/authorized_keys</em> <em>if it does not already exist</em>, and append the contents of the <em>id_rsa.pub</em> file to the end of it. Note that you also need to do this for <code>node-a</code> itself.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ cat id_rsa.pub >> ~/.ssh/authorized_keys</pre> </div> </div> </li> <li> <p>Test password-less login.</p> <div class="paragraph"> <p>If you performed the procedure correctly, if you SSH from <code>node-a</code> to either of the other nodes, using the same username, you should not be prompted for a password.</p> </div> </li> <li> <p>Since <code>node-b</code> will run a backup Master, repeat the procedure above, substituting <code>node-b</code> everywhere you see <code>node-a</code>. Be sure not to overwrite your existing <em>.ssh/authorized_keys</em> files, but concatenate the new key onto the existing file using the <code>>></code> operator rather than the <code>></code> operator.</p> </li> </ol> </div> <div class="paragraph"> <div class="title">Procedure: Prepare <code>node-a</code></div> <p><code>node-a</code> will run your primary master and ZooKeeper processes, but no RegionServers. . Stop the RegionServer from starting on <code>node-a</code>.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Edit <em>conf/regionservers</em> and remove the line which contains <code>localhost</code>. Add lines with the hostnames or IP addresses for <code>node-b</code> and <code>node-c</code>.</p> <div class="paragraph"> <p>Even if you did want to run a RegionServer on <code>node-a</code>, you should refer to it by the hostname the other servers would use to communicate with it. In this case, that would be <code>node-a.example.com</code>. This enables you to distribute the configuration to each node of your cluster any hostname conflicts. Save the file.</p> </div> </li> <li> <p>Configure HBase to use <code>node-b</code> as a backup master.</p> <div class="paragraph"> <p>Create a new file in <em>conf/</em> called <em>backup-masters</em>, and add a new line to it with the hostname for <code>node-b</code>. In this demonstration, the hostname is <code>node-b.example.com</code>.</p> </div> </li> <li> <p>Configure ZooKeeper</p> <div class="paragraph"> <p>In reality, you should carefully consider your ZooKeeper configuration. You can find out more about configuring ZooKeeper in <a href="#zookeeper">zookeeper</a>. This configuration will direct HBase to start and manage a ZooKeeper instance on each node of the cluster.</p> </div> <div class="paragraph"> <p>On <code>node-a</code>, edit <em>conf/hbase-site.xml</em> and add the following properties.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.quorum<span class="tag"></name></span> <span class="tag"><value></span>node-a.example.com,node-b.example.com,node-c.example.com<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.property.dataDir<span class="tag"></name></span> <span class="tag"><value></span>/usr/local/zookeeper<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </li> <li> <p>Everywhere in your configuration that you have referred to <code>node-a</code> as <code>localhost</code>, change the reference to point to the hostname that the other nodes will use to refer to <code>node-a</code>. In these examples, the hostname is <code>node-a.example.com</code>.</p> </li> </ol> </div> <div class="paragraph"> <div class="title">Procedure: Prepare <code>node-b</code> and <code>node-c</code></div> <p><code>node-b</code> will run a backup master server and a ZooKeeper instance.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Download and unpack HBase.</p> <div class="paragraph"> <p>Download and unpack HBase to <code>node-b</code>, just as you did for the standalone and pseudo-distributed quickstarts.</p> </div> </li> <li> <p>Copy the configuration files from <code>node-a</code> to <code>node-b</code>.and <code>node-c</code>.</p> <div class="paragraph"> <p>Each node of your cluster needs to have the same configuration information. Copy the contents of the <em>conf/</em> directory to the <em>conf/</em> directory on <code>node-b</code> and <code>node-c</code>.</p> </div> </li> </ol> </div> <div class="olist arabic"> <div class="title">Procedure: Start and Test Your Cluster</div> <ol class="arabic"> <li> <p>Be sure HBase is not running on any node.</p> <div class="paragraph"> <p>If you forgot to stop HBase from previous testing, you will have errors. Check to see whether HBase is running on any of your nodes by using the <code>jps</code> command. Look for the processes <code>HMaster</code>, <code>HRegionServer</code>, and <code>HQuorumPeer</code>. If they exist, kill them.</p> </div> </li> <li> <p>Start the cluster.</p> <div class="paragraph"> <p>On <code>node-a</code>, issue the <code>start-hbase.sh</code> command. Your output will be similar to that below.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/start-hbase.sh node-c.example.com: starting zookeeper, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-zookeeper-node-c.example.com.out node-a.example.com: starting zookeeper, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-zookeeper-node-a.example.com.out node-b.example.com: starting zookeeper, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-zookeeper-node-b.example.com.out starting master, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-master-node-a.example.com.out node-c.example.com: starting regionserver, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-regionserver-node-c.example.com.out node-b.example.com: starting regionserver, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-regionserver-node-b.example.com.out node-b.example.com: starting master, logging to /home/hbuser/hbase-0.98.3-hadoop2/bin/../logs/hbase-hbuser-master-nodeb.example.com.out</pre> </div> </div> <div class="paragraph"> <p>ZooKeeper starts first, followed by the master, then the RegionServers, and finally the backup masters.</p> </div> </li> <li> <p>Verify that the processes are running.</p> <div class="paragraph"> <p>On each node of the cluster, run the <code>jps</code> command and verify that the correct processes are running on each server. You may see additional Java processes running on your servers as well, if they are used for other purposes.</p> </div> <div class="exampleblock"> <div class="title">Example 3. <code>node-a</code> <code>jps</code> Output</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>$ jps 20355 Jps 20071 HQuorumPeer 20137 HMaster</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 4. <code>node-b</code> <code>jps</code> Output</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>$ jps 15930 HRegionServer 16194 Jps 15838 HQuorumPeer 16010 HMaster</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 5. <code>node-a</code> <code>jps</code> Output</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>$ jps 13901 Jps 13639 HQuorumPeer 13737 HRegionServer</pre> </div> </div> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">ZooKeeper Process Name</div> <div class="paragraph"> <p>The <code>HQuorumPeer</code> process is a ZooKeeper instance which is controlled and started by HBase. If you use ZooKeeper this way, it is limited to one instance per cluster node, , and is appropriate for testing only. If ZooKeeper is run outside of HBase, the process is called <code>QuorumPeer</code>. For more about ZooKeeper configuration, including using an external ZooKeeper instance with HBase, see <a href="#zookeeper">zookeeper</a>.</p> </div> </td> </tr> </table> </div> </li> <li> <p>Browse to the Web UI.</p> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Web UI Port Changes</div> Web UI Port Changes </td> </tr> </table> </div> <div class="paragraph"> <p>In HBase newer than 0.98.x, the HTTP ports used by the HBase Web UI changed from 60010 for the Master and 60030 for each RegionServer to 16010 for the Master and 16030 for the RegionServer.</p> </div> <div class="paragraph"> <p>If everything is set up correctly, you should be able to connect to the UI for the Master <code><a href="http://node-a.example.com:16010/" class="bare">http://node-a.example.com:16010/</a></code> or the secondary master at <code><a href="http://node-b.example.com:16010/" class="bare">http://node-b.example.com:16010/</a></code> for the secondary master, using a web browser. If you can connect via <code>localhost</code> but not from another host, check your firewall rules. You can see the web UI for each of the RegionServers at port 16030 of their IP addresses, or by clicking their links in the web UI for the Master.</p> </div> </li> <li> <p>Test what happens when nodes or services disappear.</p> <div class="paragraph"> <p>With a three-node cluster like you have configured, things will not be very resilient. Still, you can test what happens when the primary Master or a RegionServer disappears, by killing the processes and watching the logs.</p> </div> </li> </ol> </div> </div> <div class="sect2"> <h3 id="_where_to_go_next"><a class="anchor" href="#_where_to_go_next"></a>2.5. Where to go next</h3> <div class="paragraph"> <p>The next chapter, <a href="#configuration">configuration</a>, gives more information about the different HBase run modes, system requirements for running HBase, and critical configuration areas for setting up a distributed HBase cluster.</p> </div> </div> </div> </div> <h1 id="configuration" class="sect0"><a class="anchor" href="#configuration"></a>Apache HBase Configuration</h1> <div class="openblock partintro"> <div class="content"> This chapter expands upon the <a href="#getting_started">[getting_started]</a> chapter to further explain configuration of Apache HBase. Please read this chapter carefully, especially the <a href="#basic.prerequisites">Basic Prerequisites</a> to ensure that your HBase testing and deployment goes smoothly, and prevent data loss. </div> </div> <div class="sect1"> <h2 id="_configuration_files"><a class="anchor" href="#_configuration_files"></a>3. Configuration Files</h2> <div class="sectionbody"> <div class="paragraph"> <p>Apache HBase uses the same configuration system as Apache Hadoop. All configuration files are located in the <em>conf/</em> directory, which needs to be kept in sync for each node on your cluster.</p> </div> <div class="dlist"> <div class="title">HBase Configuration File Descriptions</div> <dl> <dt class="hdlist1"><em>backup-masters</em></dt> <dd> <p>Not present by default. A plain-text file which lists hosts on which the Master should start a backup Master process, one host per line.</p> </dd> <dt class="hdlist1"><em>hadoop-metrics2-hbase.properties</em></dt> <dd> <p>Used to connect HBase Hadoop’s Metrics2 framework. See the <a href="http://wiki.apache.org/hadoop/HADOOP-6728-MetricsV2">Hadoop Wiki entry</a> for more information on Metrics2. Contains only commented-out examples by default.</p> </dd> <dt class="hdlist1"><em>hbase-env.cmd</em> and <em>hbase-env.sh</em></dt> <dd> <p>Script for Windows and Linux / Unix environments to set up the working environment for HBase, including the location of Java, Java options, and other environment variables. The file contains many commented-out examples to provide guidance.</p> </dd> <dt class="hdlist1"><em>hbase-policy.xml</em></dt> <dd> <p>The default policy configuration file used by RPC servers to make authorization decisions on client requests. Only used if HBase <a href="#security">security</a> is enabled.</p> </dd> <dt class="hdlist1"><em>hbase-site.xml</em></dt> <dd> <p>The main HBase configuration file. This file specifies configuration options which override HBase’s default configuration. You can view (but do not edit) the default configuration file at <em>docs/hbase-default.xml</em>. You can also view the entire effective configuration for your cluster (defaults and overrides) in the <span class="label">HBase Configuration</span> tab of the HBase Web UI.</p> </dd> <dt class="hdlist1"><em>log4j.properties</em></dt> <dd> <p>Configuration file for HBase logging via <code>log4j</code>.</p> </dd> <dt class="hdlist1"><em>regionservers</em></dt> <dd> <p>A plain-text file containing a list of hosts which should run a RegionServer in your HBase cluster. By default this file contains the single entry <code>localhost</code>. It should contain a list of hostnames or IP addresses, one per line, and should only contain <code>localhost</code> if each node in your cluster will run a RegionServer on its <code>localhost</code> interface.</p> </dd> </dl> </div> <div class="admonitionblock tip"> <table> <tr> <td class="icon"> <i class="fa icon-tip" title="Tip"></i> </td> <td class="content"> <div class="title">Checking XML Validity</div> <div class="paragraph"> <p>When you edit XML, it is a good idea to use an XML-aware editor to be sure that your syntax is correct and your XML is well-formed. You can also use the <code>xmllint</code> utility to check that your XML is well-formed. By default, <code>xmllint</code> re-flows and prints the XML to standard output. To check for well-formedness and only print output if errors exist, use the command <code>xmllint -noout filename.xml</code>.</p> </div> </td> </tr> </table> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="title">Keep Configuration In Sync Across the Cluster</div> <div class="paragraph"> <p>When running in distributed mode, after you make an edit to an HBase configuration, make sure you copy the content of the <em>conf/</em> directory to all nodes of the cluster. HBase will not do this for you. Use <code>rsync</code>, <code>scp</code>, or another secure mechanism for copying the configuration files to your nodes. For most configuration, a restart is needed for servers to pick up changes An exception is dynamic configuration. to be described later below.</p> </div> </td> </tr> </table> </div> </div> </div> <div class="sect1"> <h2 id="basic.prerequisites"><a class="anchor" href="#basic.prerequisites"></a>4. Basic Prerequisites</h2> <div class="sectionbody"> <div class="paragraph"> <p>This section lists required services and some required system configuration.</p> </div> <table id="java" class="tableblock frame-all grid-all spread"> <caption class="title">Table 2. Java</caption> <colgroup> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 57%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">HBase Version</th> <th class="tableblock halign-left valign-top">JDK 6</th> <th class="tableblock halign-left valign-top">JDK 7</th> <th class="tableblock halign-left valign-top">JDK 8</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">1.1</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><a href="http://search-hadoop.com/m/DHED4Zlz0R1">Not Supported</a></p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Running with JDK 8 will work but is not well tested.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">1.0</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><a href="http://search-hadoop.com/m/DHED4Zlz0R1">Not Supported</a></p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Running with JDK 8 will work but is not well tested.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">0.98</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Running with JDK 8 works but is not well tested. Building with JDK 8 would require removal of the deprecated <code>remove()</code> method of the <code>PoolMap</code> class and is under consideration. See <a href="https://issues.apache.org/jira/browse/HBASE-7608">HBASE-7608</a> for more information about JDK 8 support.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">0.96</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N/A</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">0.94</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N/A</p></td> </tr> </tbody> </table> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> In HBase 0.98.5 and newer, you must set <code>JAVA_HOME</code> on each node of your cluster. <em>hbase-env.sh</em> provides a handy mechanism to do this. </td> </tr> </table> </div> <div class="dlist"> <div class="title">Operating System Utilities</div> <dl> <dt class="hdlist1">ssh</dt> <dd> <p>HBase uses the Secure Shell (ssh) command and utilities extensively to communicate between cluster nodes. Each server in the cluster must be running <code>ssh</code> so that the Hadoop and HBase daemons can be managed. You must be able to connect to all nodes via SSH, including the local node, from the Master as well as any backup Master, using a shared key rather than a password. You can see the basic methodology for such a set-up in Linux or Unix systems at "<a href="#passwordless.ssh.quickstart">Procedure: Configure Passwordless SSH Access</a>". If your cluster nodes use OS X, see the section, <a href="http://wiki.apache.org/hadoop/Running_Hadoop_On_OS_X_10.5_64-bit_%28Single-Node_Cluster%29">SSH: Setting up Remote Desktop and Enabling Self-Login</a> on the Hadoop wiki.</p> </dd> <dt class="hdlist1">DNS</dt> <dd> <p>HBase uses the local hostname to self-report its IP address. Both forward and reverse DNS resolving must work in versions of HBase previous to 0.92.0. The <a href="https://github.com/sujee/hadoop-dns-checker">hadoop-dns-checker</a> tool can be used to verify DNS is working correctly on the cluster. The project <code>README</code> file provides detailed instructions on usage.</p> </dd> <dt class="hdlist1">Loopback IP</dt> <dd> <p>Prior to hbase-0.96.0, HBase only used the IP address <code>127.0.0.1</code> to refer to <code>localhost</code>, and this could not be configured. See <a href="#loopback.ip">Loopback IP</a> for more details.</p> </dd> <dt class="hdlist1">NTP</dt> <dd> <p>The clocks on cluster nodes should be synchronized. A small amount of variation is acceptable, but larger amounts of skew can cause erratic and unexpected behavior. Time synchronization is one of the first things to check if you see unexplained problems in your cluster. It is recommended that you run a Network Time Protocol (NTP) service, or another time-synchronization mechanism, on your cluster, and that all nodes look to the same service for time synchronization. See the <a href="http://www.tldp.org/LDP/sag/html/basic-ntp-config.html">Basic NTP Configuration</a> at <em class="citetitle">The Linux Documentation Project (TLDP)</em> to set up NTP.</p> </dd> <dt class="hdlist1">Limits on Number of Files and Processes (ulimit)</dt> <dd> <p>Apache HBase is a database. It requires the ability to open a large number of files at once. Many Linux distributions limit the number of files a single user is allowed to open to <code>1024</code> (or <code>256</code> on older versions of OS X). You can check this limit on your servers by running the command <code>ulimit -n</code> when logged in as the user which runs HBase. See <a href="#trouble.rs.runtime.filehandles">the Troubleshooting section</a> for some of the problems you may experience if the limit is too low. You may also notice errors such as the following:</p> <div class="listingblock"> <div class="content"> <pre>2010-04-06 03:04:37,542 INFO org.apache.hadoop.hdfs.DFSClient: Exception increateBlockOutputStream java.io.EOFException 2010-04-06 03:04:37,542 INFO org.apache.hadoop.hdfs.DFSClient: Abandoning block blk_-6935524980745310745_1391901</pre> </div> </div> <div class="paragraph"> <p>It is recommended to raise the ulimit to at least 10,000, but more likely 10,240, because the value is usually expressed in multiples of 1024. Each ColumnFamily has at least one StoreFile, and possibly more than six StoreFiles if the region is under load. The number of open files required depends upon the number of ColumnFamilies and the number of regions. The following is a rough formula for calculating the potential number of open files on a RegionServer.</p> </div> <div class="listingblock"> <div class="title">Calculate the Potential Number of Open Files</div> <div class="content"> <pre>(StoreFiles per ColumnFamily) x (regions per RegionServer)</pre> </div> </div> <div class="paragraph"> <p>For example, assuming that a schema had 3 ColumnFamilies per region with an average of 3 StoreFiles per ColumnFamily, and there are 100 regions per RegionServer, the JVM will open <code>3 * 3 * 100 = 900</code> file descriptors, not counting open JAR files, configuration files, and others. Opening a file does not take many resources, and the risk of allowing a user to open too many files is minimal.</p> </div> <div class="paragraph"> <p>Another related setting is the number of processes a user is allowed to run at once. In Linux and Unix, the number of processes is set using the <code>ulimit -u</code> command. This should not be confused with the <code>nproc</code> command, which controls the number of CPUs available to a given user. Under load, a <code>ulimit -u</code> that is too low can cause OutOfMemoryError exceptions. See Jack Levin’s major HDFS issues thread on the hbase-users mailing list, from 2011.</p> </div> <div class="paragraph"> <p>Configuring the maximum number of file descriptors and processes for the user who is running the HBase process is an operating system configuration, rather than an HBase configuration. It is also important to be sure that the settings are changed for the user that actually runs HBase. To see which user started HBase, and that user’s ulimit configuration, look at the first line of the HBase log for that instance. A useful read setting config on you hadoop cluster is Aaron Kimballs' Configuration Parameters: What can you just ignore?</p> </div> <div class="exampleblock"> <div class="title">Example 6. <code>ulimit</code> Settings on Ubuntu</div> <div class="content"> <div class="paragraph"> <p>To configure ulimit settings on Ubuntu, edit <em>/etc/security/limits.conf</em>, which is a space-delimited file with four columns. Refer to the man page for <em>limits.conf</em> for details about the format of this file. In the following example, the first line sets both soft and hard limits for the number of open files (nofile) to 32768 for the operating system user with the username hadoop. The second line sets the number of processes to 32000 for the same user.</p> </div> <div class="listingblock"> <div class="content"> <pre>hadoop - nofile 32768 hadoop - nproc 32000</pre> </div> </div> <div class="paragraph"> <p>The settings are only applied if the Pluggable Authentication Module (PAM) environment is directed to use them. To configure PAM to use these limits, be sure that the <em>/etc/pam.d/common-session</em> file contains the following line:</p> </div> <div class="listingblock"> <div class="content"> <pre>session required pam_limits.so</pre> </div> </div> </div> </div> </dd> <dt class="hdlist1">Linux Shell</dt> <dd> <p>All of the shell scripts that come with HBase rely on the <a href="http://www.gnu.org/software/bash">GNU Bash</a> shell.</p> </dd> <dt class="hdlist1">Windows</dt> <dd> <p>Prior to HBase 0.96, testing for running HBase on Microsoft Windows was limited. Running a on Windows nodes is not recommended for production systems.</p> </dd> </dl> </div> <div class="sect2"> <h3 id="hadoop"><a class="anchor" href="#hadoop"></a>4.1. <a href="http://hadoop.apache.org">Hadoop</a></h3> <div class="paragraph"> <p>The following table summarizes the versions of Hadoop supported with each version of HBase. Based on the version of HBase, you should select the most appropriate version of Hadoop. You can use Apache Hadoop, or a vendor’s distribution of Hadoop. No distinction is made here. See <a href="http://wiki.apache.org/hadoop/Distributions%20and%20Commercial%20Support">the Hadoop wiki</a> for information about vendors of Hadoop.</p> </div> <div class="admonitionblock tip"> <table> <tr> <td class="icon"> <i class="fa icon-tip" title="Tip"></i> </td> <td class="content"> <div class="title">Hadoop 2.x is recommended.</div> <div class="paragraph"> <p>Hadoop 2.x is faster and includes features, such as short-circuit reads, which will help improve your HBase random read profile. Hadoop 2.x also includes important bug fixes that will improve your overall HBase experience. HBase 0.98 drops support for Hadoop 1.0, deprecates use of Hadoop 1.1+, and HBase 1.0 will not support Hadoop 1.x.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>Use the following legend to interpret this table:</p> </div> <div class="ulist"> <div class="title">Hadoop version support matrix</div> <ul> <li> <p>"S" = supported</p> </li> <li> <p>"X" = not supported</p> </li> <li> <p>"NT" = Not tested</p> </li> </ul> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 14%;"> <col style="width: 14%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top"></th> <th class="tableblock halign-left valign-top">HBase-0.92.x</th> <th class="tableblock halign-left valign-top">HBase-0.94.x</th> <th class="tableblock halign-left valign-top">HBase-0.96.x</th> <th class="tableblock halign-left valign-top">HBase-0.98.x (Support for Hadoop 1.1+ is deprecated.)</th> <th class="tableblock halign-left valign-top">HBase-1.0.x (Hadoop 1.x is NOT supported)</th> <th class="tableblock halign-left valign-top">HBase-1.1.x</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-0.20.205</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-0.22.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-1.0.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-1.1.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-0.23.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.0.x-alpha</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.1.0-beta</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.2.0</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.3.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.4.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.5.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.6.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">S</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Hadoop-2.7.x</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">X</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">NT</p></td> </tr> </tbody> </table> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Replace the Hadoop Bundled With HBase!</div> <div class="paragraph"> <p>Because HBase depends on Hadoop, it bundles an instance of the Hadoop jar under its <em>lib</em> directory. The bundled jar is ONLY for use in standalone mode. In distributed mode, it is <em>critical</em> that the version of Hadoop that is out on your cluster match what is under HBase. Replace the hadoop jar found in the HBase lib directory with the hadoop jar you are running on your cluster to avoid version mismatch issues. Make sure you replace the jar in HBase everywhere on your cluster. Hadoop version mismatch issues have various manifestations but often all looks like its hung up.</p> </div> </td> </tr> </table> </div> <div class="sect3"> <h4 id="hadoop2.hbase_0.94"><a class="anchor" href="#hadoop2.hbase_0.94"></a>4.1.1. Apache HBase 0.94 with Hadoop 2</h4> <div class="paragraph"> <p>To get 0.94.x to run on Hadoop 2.2.0, you need to change the hadoop 2 and protobuf versions in the <em>pom.xml</em>: Here is a diff with pom.xml changes:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="error">$</span> svn diff pom.xml Index: pom.xml =================================================================== --- pom.xml (revision <span class="integer">1545157</span>) +++ pom.xml (working copy) <span class="error">@</span><span class="error">@</span> -<span class="integer">1034</span>,<span class="integer">7</span> +<span class="integer">1034</span>,<span class="integer">7</span> <span class="error">@</span><span class="error">@</span> <slf4j.version><span class="float">1.4</span><span class="float">.3</span></slf4j.version> <log4j.version><span class="float">1.2</span><span class="float">.16</span></log4j.version> <mockito-all.version><span class="float">1.8</span><span class="float">.5</span></mockito-all.version> - <protobuf.version><span class="float">2.4</span><span class="float">.0</span>a</protobuf.version> + <protobuf.version><span class="float">2.5</span><span class="float">.0</span></protobuf.version> <stax-api.version><span class="float">1.0</span><span class="float">.1</span></stax-api.version> <thrift.version><span class="float">0.8</span><span class="float">.0</span></thrift.version> <zookeeper.version><span class="float">3.4</span><span class="float">.5</span></zookeeper.version> <span class="error">@</span><span class="error">@</span> -<span class="integer">2241</span>,<span class="integer">7</span> +<span class="integer">2241</span>,<span class="integer">7</span> <span class="error">@</span><span class="error">@</span> </property> </activation> <properties> - <hadoop.version><span class="float">2.0</span><span class="float">.0</span>-alpha</hadoop.version> + <hadoop.version><span class="float">2.2</span><span class="float">.0</span></hadoop.version> <slf4j.version><span class="float">1.6</span><span class="float">.1</span></slf4j.version> </properties> <dependencies></code></pre> </div> </div> <div class="paragraph"> <p>The next step is to regenerate Protobuf files and assuming that the Protobuf has been installed:</p> </div> <div class="ulist"> <ul> <li> <p>Go to the HBase root folder, using the command line;</p> </li> <li> <p>Type the following commands:</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ protoc -Isrc/main/protobuf --java_out=src/main/java src/main/protobuf/hbase.proto</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ protoc -Isrc/main/protobuf --java_out=src/main/java src/main/protobuf/ErrorHandling.proto</code></pre> </div> </div> </li> </ul> </div> <div class="paragraph"> <p>Building against the hadoop 2 profile by running something like the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ mvn clean install assembly:single -Dhadoop.profile=2.0 -DskipTests</pre> </div> </div> </div> <div class="sect3"> <h4 id="hadoop.hbase_0.94"><a class="anchor" href="#hadoop.hbase_0.94"></a>4.1.2. Apache HBase 0.92 and 0.94</h4> <div class="paragraph"> <p>HBase 0.92 and 0.94 versions can work with Hadoop versions, 0.20.205, 0.22.x, 1.0.x, and 1.1.x. HBase-0.94 can additionally work with Hadoop-0.23.x and 2.x, but you may have to recompile the code using the specific maven profile (see top level pom.xml)</p> </div> </div> <div class="sect3"> <h4 id="hadoop.hbase_0.96"><a class="anchor" href="#hadoop.hbase_0.96"></a>4.1.3. Apache HBase 0.96</h4> <div class="paragraph"> <p>As of Apache HBase 0.96.x, Apache Hadoop 1.0.x at least is required. Hadoop 2 is strongly encouraged (faster but also has fixes that help MTTR). We will no longer run properly on older Hadoops such as 0.20.205 or branch-0.20-append. Do not move to Apache HBase 0.96.x if you cannot upgrade your Hadoop. See <a href="http://search-hadoop.com/m/7vFVx4EsUb2">HBase, mail # dev - DISCUSS: Have hbase require at least hadoop 1.0.0 in hbase 0.96.0?</a></p> </div> </div> <div class="sect3"> <h4 id="hadoop.older.versions"><a class="anchor" href="#hadoop.older.versions"></a>4.1.4. Hadoop versions 0.20.x - 1.x</h4> <div class="paragraph"> <p>HBase will lose data unless it is running on an HDFS that has a durable <code>sync</code> implementation. DO NOT use Hadoop 0.20.2, Hadoop 0.20.203.0, and Hadoop 0.20.204.0 which DO NOT have this attribute. Currently only Hadoop versions 0.20.205.x or any release in excess of this version — this includes hadoop-1.0.0 — have a working, durable sync. The Cloudera blog post <a href="http://www.cloudera.com/blog/2012/01/an-update-on-apache-hadoop-1-0/">An update on Apache Hadoop 1.0</a> by Charles Zedlweski has a nice exposition on how all the Hadoop versions relate. It’s worth checking out if you are having trouble making sense of the Hadoop version morass.</p> </div> <div class="paragraph"> <p>Sync has to be explicitly enabled by setting <code>dfs.support.append</code> equal to true on both the client side — in <em>hbase-site.xml</em> — and on the serverside in <em>hdfs-site.xml</em> (The sync facility HBase needs is a subset of the append code path).</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>dfs.support.append<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>You will have to restart your cluster after making this edit. Ignore the chicken-little comment you’ll find in the <em>hdfs-default.xml</em> in the description for the <code>dfs.support.append</code> configuration.</p> </div> </div> <div class="sect3"> <h4 id="hadoop.security"><a class="anchor" href="#hadoop.security"></a>4.1.5. Apache HBase on Secure Hadoop</h4> <div class="paragraph"> <p>Apache HBase will run on any Hadoop 0.20.x that incorporates Hadoop security features as long as you do as suggested above and replace the Hadoop jar that ships with HBase with the secure version. If you want to read more about how to setup Secure HBase, see <a href="#hbase.secure.configuration">hbase.secure.configuration</a>.</p> </div> </div> <div class="sect3"> <h4 id="dfs.datanode.max.transfer.threads"><a class="anchor" href="#dfs.datanode.max.transfer.threads"></a>4.1.6. <code>dfs.datanode.max.transfer.threads</code> </h4> <div class="paragraph"> <p>An HDFS DataNode has an upper bound on the number of files that it will serve at any one time. Before doing any loading, make sure you have configured Hadoop’s <em>conf/hdfs-site.xml</em>, setting the <code>dfs.datanode.max.transfer.threads</code> value to at least the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>dfs.datanode.max.transfer.threads<span class="tag"></name></span> <span class="tag"><value></span>4096<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Be sure to restart your HDFS after making the above configuration.</p> </div> <div class="paragraph"> <p>Not having this configuration in place makes for strange-looking failures. One manifestation is a complaint about missing blocks. For example:</p> </div> <div class="listingblock"> <div class="content"> <pre>10/12/08 20:10:31 INFO hdfs.DFSClient: Could not obtain block blk_XXXXXXXXXXXXXXXXXXXXXX_YYYYYYYY from any node: java.io.IOException: No live nodes contain current block. Will get new block locations from namenode and retry...</pre> </div> </div> <div class="paragraph"> <p>See also <a href="#casestudies.max.transfer.threads">casestudies.max.transfer.threads</a> and note that this property was previously known as <code>dfs.datanode.max.xcievers</code> (e.g. <a href="http://ccgtech.blogspot.com/2010/02/hadoop-hdfs-deceived-by-xciever.html">Hadoop HDFS: Deceived by Xciever</a>).</p> </div> </div> </div> <div class="sect2"> <h3 id="zookeeper.requirements"><a class="anchor" href="#zookeeper.requirements"></a>4.2. ZooKeeper Requirements</h3> <div class="paragraph"> <p>ZooKeeper 3.4.x is required as of HBase 1.0.0. HBase makes use of the <code>multi</code> functionality that is only available since 3.4.0 (The <code>useMulti</code> configuration option defaults to <code>true</code> in HBase 1.0.0). See <a href="https://issues.apache.org/jira/browse/HBASE-12241">HBASE-12241 (The crash of regionServer when taking deadserver’s replication queue breaks replication)</a> and <a href="https://issues.apache.org/jira/browse/HBASE-6775">HBASE-6775 (Use ZK.multi when available for HBASE-6710 0.92/0.94 compatibility fix)</a> for background.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="standalone_dist"><a class="anchor" href="#standalone_dist"></a>5. HBase run modes: Standalone and Distributed</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase has two run modes: <a href="#standalone">standalone</a> and <a href="#distributed">distributed</a>. Out of the box, HBase runs in standalone mode. Whatever your mode, you will need to configure HBase by editing files in the HBase <em>conf</em> directory. At a minimum, you must edit <span class="code">conf/hbase-env.sh</span> to tell HBase which java to use. In this file you set HBase environment variables such as the heapsize and other options for the <code>JVM</code>, the preferred location for log files, etc. Set <span class="var">JAVA_HOME</span> to point at the root of your java install.</p> </div> <div class="sect2"> <h3 id="standalone"><a class="anchor" href="#standalone"></a>5.1. Standalone HBase</h3> <div class="paragraph"> <p>This is the default mode. Standalone mode is what is described in the <a href="#quickstart">quickstart</a> section. In standalone mode, HBase does not use HDFS — it uses the local filesystem instead — and it runs all HBase daemons and a local ZooKeeper all up in the same JVM. Zookeeper binds to a well known port so clients may talk to HBase.</p> </div> </div> <div class="sect2"> <h3 id="_distributed"><a class="anchor" href="#_distributed"></a>5.2. Distributed</h3> <div class="paragraph"> <p>Distributed mode can be subdivided into distributed but all daemons run on a single node — a.k.a <em>pseudo-distributed</em> — and <em>fully-distributed</em> where the daemons are spread across all nodes in the cluster. The <em>pseudo-distributed</em> vs. <em>fully-distributed</em> nomenclature comes from Hadoop.</p> </div> <div class="paragraph"> <p>Pseudo-distributed mode can run against the local filesystem or it can run against an instance of the <em>Hadoop Distributed File System</em> (HDFS). Fully-distributed mode can ONLY run on HDFS. See the Hadoop <a href="http://hadoop.apache.org/docs/current/">documentation</a> for how to set up HDFS. A good walk-through for setting up HDFS on Hadoop 2 can be found at <a href="http://www.alexjf.net/blog/distributed-systems/hadoop-yarn-installation-definitive-guide" class="bare">http://www.alexjf.net/blog/distributed-systems/hadoop-yarn-installation-definitive-guide</a>.</p> </div> <div class="sect3"> <h4 id="pseudo"><a class="anchor" href="#pseudo"></a>5.2.1. Pseudo-distributed</h4> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Pseudo-Distributed Quickstart</div> <div class="paragraph"> <p>A quickstart has been added to the <a href="#quickstart">quickstart</a> chapter. See <a href="#quickstart_pseudo">quickstart-pseudo</a>. Some of the information that was originally in this section has been moved there.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>A pseudo-distributed mode is simply a fully-distributed mode run on a single host. Use this configuration testing and prototyping on HBase. Do not use this configuration for production nor for evaluating HBase performance.</p> </div> </div> </div> <div class="sect2"> <h3 id="fully_dist"><a class="anchor" href="#fully_dist"></a>5.3. Fully-distributed</h3> <div class="paragraph"> <p>By default, HBase runs in standalone mode. Both standalone mode and pseudo-distributed mode are provided for the purposes of small-scale testing. For a production environment, distributed mode is appropriate. In distributed mode, multiple instances of HBase daemons run on multiple servers in the cluster.</p> </div> <div class="paragraph"> <p>Just as in pseudo-distributed mode, a fully distributed configuration requires that you set the <code>hbase-cluster.distributed</code> property to <code>true</code>. Typically, the <code>hbase.rootdir</code> is configured to point to a highly-available HDFS filesystem.</p> </div> <div class="paragraph"> <p>In addition, the cluster is configured so that multiple cluster nodes enlist as RegionServers, ZooKeeper QuorumPeers, and backup HMaster servers. These configuration basics are all demonstrated in <a href="#quickstart_fully_distributed">quickstart-fully-distributed</a>.</p> </div> <div class="paragraph"> <div class="title">Distributed RegionServers</div> <p>Typically, your cluster will contain multiple RegionServers all running on different servers, as well as primary and backup Master and Zookeeper daemons. The <em>conf/regionservers</em> file on the master server contains a list of hosts whose RegionServers are associated with this cluster. Each host is on a separate line. All hosts listed in this file will have their RegionServer processes started and stopped when the master server starts or stops.</p> </div> <div class="paragraph"> <div class="title">ZooKeeper and HBase</div> <p>See the <a href="#zookeeper">ZooKeeper</a> section for ZooKeeper setup instructions for HBase.</p> </div> <div class="exampleblock"> <div class="title">Example 7. Example Distributed HBase Cluster</div> <div class="content"> <div class="paragraph"> <p>This is a bare-bones <em>conf/hbase-site.xml</em> for a distributed HBase cluster. A cluster that is used for real-world work would contain more custom configuration parameters. Most HBase configuration directives have default values, which are used unless the value is overridden in the <em>hbase-site.xml</em>. See "<a href="#config.files">Configuration Files</a>" for more information.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><configuration></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rootdir<span class="tag"></name></span> <span class="tag"><value></span>hdfs://namenode.example.org:8020/hbase<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.cluster.distributed<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.quorum<span class="tag"></name></span> <span class="tag"><value></span>node-a.example.com,node-b.example.com,node-c.example.com<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"></configuration></span></code></pre> </div> </div> <div class="paragraph"> <p>This is an example <em>conf/regionservers</em> file, which contains a list of nodes that should run a RegionServer in the cluster. These nodes need HBase installed and they need to use the same contents of the <em>conf/</em> directory as the Master server</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">node-a.example.com node-b.example.com node-c.example.com</code></pre> </div> </div> <div class="paragraph"> <p>This is an example <em>conf/backup-masters</em> file, which contains a list of each node that should run a backup Master instance. The backup Master instances will sit idle unless the main Master becomes unavailable.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">node-b.example.com node-c.example.com</code></pre> </div> </div> </div> </div> <div class="paragraph"> <div class="title">Distributed HBase Quickstart</div> <p>See <a href="#quickstart_fully_distributed">quickstart-fully-distributed</a> for a walk-through of a simple three-node cluster configuration with multiple ZooKeeper, backup HMaster, and RegionServer instances.</p> </div> <div class="olist arabic"> <div class="title">Procedure: HDFS Client Configuration</div> <ol class="arabic"> <li> <p>Of note, if you have made HDFS client configuration changes on your Hadoop cluster, such as configuration directives for HDFS clients, as opposed to server-side configurations, you must use one of the following methods to enable HBase to see and use these configuration changes:</p> <div class="olist loweralpha"> <ol class="loweralpha" type="a"> <li> <p>Add a pointer to your <code>HADOOP_CONF_DIR</code> to the <code>HBASE_CLASSPATH</code> environment variable in <em>hbase-env.sh</em>.</p> </li> <li> <p>Add a copy of <em>hdfs-site.xml</em> (or <em>hadoop-site.xml</em>) or, better, symlinks, under <em>${HBASE_HOME}/conf</em>, or</p> </li> <li> <p>if only a small set of HDFS client configurations, add them to <em>hbase-site.xml</em>.</p> </li> </ol> </div> </li> </ol> </div> <div class="paragraph"> <p>An example of such an HDFS client configuration is <code>dfs.replication</code>. If for example, you want to run with a replication factor of 5, HBase will create files with the default of 3 unless you do the above to make the configuration available to HBase.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="confirm"><a class="anchor" href="#confirm"></a>6. Running and Confirming Your Installation</h2> <div class="sectionbody"> <div class="paragraph"> <p>Make sure HDFS is running first. Start and stop the Hadoop HDFS daemons by running <em>bin/start-hdfs.sh</em> over in the <code>HADOOP_HOME</code> directory. You can ensure it started properly by testing the <code>put</code> and <code>get</code> of files into the Hadoop filesystem. HBase does not normally use the MapReduce or YARN daemons. These do not need to be started.</p> </div> <div class="paragraph"> <p><em>If</em> you are managing your own ZooKeeper, start it and confirm it’s running, else HBase will start up ZooKeeper for you as part of its start process.</p> </div> <div class="paragraph"> <p>Start HBase with the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre>bin/start-hbase.sh</pre> </div> </div> <div class="paragraph"> <p>Run the above from the <code>HBASE_HOME</code> directory.</p> </div> <div class="paragraph"> <p>You should now have a running HBase instance. HBase logs can be found in the <em>logs</em> subdirectory. Check them out especially if HBase had trouble starting.</p> </div> <div class="paragraph"> <p>HBase also puts up a UI listing vital attributes. By default it’s deployed on the Master host at port 16010 (HBase RegionServers listen on port 16020 by default and put up an informational HTTP server at port 16030). If the Master is running on a host named <code>master.example.org</code> on the default port, point your browser at <em><a href="http://master.example.org:16010" class="bare">http://master.example.org:16010</a></em> to see the web interface.</p> </div> <div class="paragraph"> <p>Prior to HBase 0.98 the master UI was deployed on port 60010, and the HBase RegionServers UI on port 60030.</p> </div> <div class="paragraph"> <p>Once HBase has started, see the <a href="#shell_exercises">shell exercises</a> section for how to create tables, add data, scan your insertions, and finally disable and drop your tables.</p> </div> <div class="paragraph"> <p>To stop HBase after exiting the HBase shell enter</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/stop-hbase.sh stopping hbase...............</pre> </div> </div> <div class="paragraph"> <p>Shutdown can take a moment to complete. It can take longer if your cluster is comprised of many machines. If you are running a distributed operation, be sure to wait until HBase has shut down completely before stopping the Hadoop daemons.</p> </div> </div> </div> <div class="sect1"> <h2 id="config.files"><a class="anchor" href="#config.files"></a>7. Default Configuration</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="hbase.site"><a class="anchor" href="#hbase.site"></a>7.1. <em>hbase-site.xml</em> and <em>hbase-default.xml</em></h3> <div class="paragraph"> <p>Just as in Hadoop where you add site-specific HDFS configuration to the <em>hdfs-site.xml</em> file, for HBase, site specific customizations go into the file <em>conf/hbase-site.xml</em>. For the list of configurable properties, see <a href="#hbase_default_configurations">hbase default configurations</a> below or view the raw <em>hbase-default.xml</em> source file in the HBase source code at <em>src/main/resources</em>.</p> </div> <div class="paragraph"> <p>Not all configuration options make it out to <em>hbase-default.xml</em>. Configuration that it is thought rare anyone would change can exist only in code; the only way to turn up such configurations is via a reading of the source code itself.</p> </div> <div class="paragraph"> <p>Currently, changes here will require a cluster restart for HBase to notice the change.</p> </div> </div> <div class="sect2"> <h3 id="hbase_default_configurations"><a class="anchor" href="#hbase_default_configurations"></a>7.2. HBase Default Configuration</h3> <div class="paragraph"> <p>The documentation below is generated using the default hbase configuration file, <em>hbase-default.xml</em>, as source.</p> </div> <div id="hbase.tmp.dir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.tmp.dir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Temporary directory on the local filesystem. Change this setting to point to a location more permanent than '/tmp', the usual resolve for java.io.tmpdir, as the '/tmp' directory is cleared on machine restart.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>${java.io.tmpdir}/hbase-${user.name}</code></p> </div> </dd> </dl> </div> <div id="hbase.rootdir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rootdir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The directory shared by region servers and into which HBase persists. The URL should be 'fully-qualified' to include the filesystem scheme. For example, to specify the HDFS directory '/hbase' where the HDFS instance’s namenode is running at namenode.example.org on port 9000, set this value to: hdfs://namenode.example.org:9000/hbase. By default, we write to whatever ${hbase.tmp.dir} is set too — usually /tmp — so change this configuration or else all data will be lost on machine restart.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>${hbase.tmp.dir}/hbase</code></p> </div> </dd> </dl> </div> <div id="hbase.fs.tmp.dir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.fs.tmp.dir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A staging directory in default file system (HDFS) for keeping temporary data.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>/user/${user.name}/hbase-staging</code></p> </div> </dd> </dl> </div> <div id="hbase.bulkload.staging.dir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.bulkload.staging.dir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A staging directory in default file system (HDFS) for bulk loading.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>${hbase.fs.tmp.dir}</code></p> </div> </dd> </dl> </div> <div id="hbase.cluster.distributed" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.cluster.distributed</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The mode the cluster will be in. Possible values are false for standalone mode and true for distributed mode. If false, startup will run all HBase and ZooKeeper daemons together in the one JVM.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.quorum" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.quorum</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Comma separated list of servers in the ZooKeeper ensemble (This config. should have been named hbase.zookeeper.ensemble). For example, "host1.mydomain.com,host2.mydomain.com,host3.mydomain.com". By default this is set to localhost for local and pseudo-distributed modes of operation. For a fully-distributed setup, this should be set to a full list of ZooKeeper ensemble servers. If HBASE_MANAGES_ZK is set in hbase-env.sh this is the list of servers which hbase will start/stop ZooKeeper on as part of cluster start/stop. Client-side, we will take this list of ensemble members and put it together with the hbase.zookeeper.clientPort config. and pass it into zookeeper constructor as the connectString parameter.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>localhost</code></p> </div> </dd> </dl> </div> <div id="zookeeper.recovery.retry.maxsleeptime" class="dlist"> <dl> <dt class="hdlist1"><code>zookeeper.recovery.retry.maxsleeptime</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Max sleep time before retry zookeeper operations in milliseconds, a max time is needed here so that sleep time won’t grow unboundedly</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>60000</code></p> </div> </dd> </dl> </div> <div id="hbase.local.dir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.local.dir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Directory on the local filesystem to be used as a local storage.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>${hbase.tmp.dir}/local/</code></p> </div> </dd> </dl> </div> <div id="hbase.master.port" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.port</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The port the HBase Master should bind to.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16000</code></p> </div> </dd> </dl> </div> <div id="hbase.master.info.port" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.info.port</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The port for the HBase Master web UI. Set to -1 if you do not want a UI instance run.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16010</code></p> </div> </dd> </dl> </div> <div id="hbase.master.info.bindAddress" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.info.bindAddress</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The bind address for the HBase Master web UI</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.0.0.0</code></p> </div> </dd> </dl> </div> <div id="hbase.master.logcleaner.plugins" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.logcleaner.plugins</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of BaseLogCleanerDelegate invoked by the LogsCleaner service. These WAL cleaners are called in order, so put the cleaner that prunes the most files in front. To implement your own BaseLogCleanerDelegate, just put it in HBase’s classpath and add the fully qualified class name here. Always add the above default log cleaners in the list.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.master.cleaner.TimeToLiveLogCleaner</code></p> </div> </dd> </dl> </div> <div id="hbase.master.logcleaner.ttl" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.logcleaner.ttl</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum time a WAL can stay in the .oldlogdir directory, after which it will be cleaned by a Master thread.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>600000</code></p> </div> </dd> </dl> </div> <div id="hbase.master.hfilecleaner.plugins" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.hfilecleaner.plugins</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of BaseHFileCleanerDelegate invoked by the HFileCleaner service. These HFiles cleaners are called in order, so put the cleaner that prunes the most files in front. To implement your own BaseHFileCleanerDelegate, just put it in HBase’s classpath and add the fully qualified class name here. Always add the above default log cleaners in the list as they will be overwritten in hbase-site.xml.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.master.cleaner.TimeToLiveHFileCleaner</code></p> </div> </dd> </dl> </div> <div id="hbase.master.infoserver.redirect" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.infoserver.redirect</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Whether or not the Master listens to the Master web UI port (hbase.master.info.port) and redirects requests to the web UI server shared by the Master and RegionServer.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.port" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.port</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The port the HBase RegionServer binds to.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16020</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.info.port" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.info.port</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The port for the HBase RegionServer web UI Set to -1 if you do not want the RegionServer UI to run.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16030</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.info.bindAddress" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.info.bindAddress</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The address for the HBase RegionServer web UI</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.0.0.0</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.info.port.auto" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.info.port.auto</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Whether or not the Master or RegionServer UI should search for a port to bind to. Enables automatic port search if hbase.regionserver.info.port is already in use. Useful for testing, turned off by default.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.handler.count" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.handler.count</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Count of RPC Listener instances spun up on RegionServers. Same property is used by the Master for count of master handlers.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>30</code></p> </div> </dd> </dl> </div> <div id="hbase.ipc.server.callqueue.handler.factor" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.ipc.server.callqueue.handler.factor</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Factor to determine the number of call queues. A value of 0 means a single queue shared between all the handlers. A value of 1 means that each handler has its own queue.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.1</code></p> </div> </dd> </dl> </div> <div id="hbase.ipc.server.callqueue.read.ratio" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.ipc.server.callqueue.read.ratio</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Split the call queues into read and write queues. The specified interval (which should be between 0.0 and 1.0) will be multiplied by the number of call queues. A value of 0 indicate to not split the call queues, meaning that both read and write requests will be pushed to the same set of queues. A value lower than 0.5 means that there will be less read queues than write queues. A value of 0.5 means there will be the same number of read and write queues. A value greater than 0.5 means that there will be more read queues than write queues. A value of 1.0 means that all the queues except one are used to dispatch read requests. Example: Given the total number of call queues being 10 a read.ratio of 0 means that: the 10 queues will contain both read/write requests. a read.ratio of 0.3 means that: 3 queues will contain only read requests and 7 queues will contain only write requests. a read.ratio of 0.5 means that: 5 queues will contain only read requests and 5 queues will contain only write requests. a read.ratio of 0.8 means that: 8 queues will contain only read requests and 2 queues will contain only write requests. a read.ratio of 1 means that: 9 queues will contain only read requests and 1 queues will contain only write requests.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0</code></p> </div> </dd> </dl> </div> <div id="hbase.ipc.server.callqueue.scan.ratio" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.ipc.server.callqueue.scan.ratio</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Given the number of read call queues, calculated from the total number of call queues multiplied by the callqueue.read.ratio, the scan.ratio property will split the read call queues into small-read and long-read queues. A value lower than 0.5 means that there will be less long-read queues than short-read queues. A value of 0.5 means that there will be the same number of short-read and long-read queues. A value greater than 0.5 means that there will be more long-read queues than short-read queues A value of 0 or 1 indicate to use the same set of queues for gets and scans. Example: Given the total number of read call queues being 8 a scan.ratio of 0 or 1 means that: 8 queues will contain both long and short read requests. a scan.ratio of 0.3 means that: 2 queues will contain only long-read requests and 6 queues will contain only short-read requests. a scan.ratio of 0.5 means that: 4 queues will contain only long-read requests and 4 queues will contain only short-read requests. a scan.ratio of 0.8 means that: 6 queues will contain only long-read requests and 2 queues will contain only short-read requests.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.msginterval" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.msginterval</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Interval between messages from the RegionServer to Master in milliseconds.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3000</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.logroll.period" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.logroll.period</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Period at which we will roll the commit log regardless of how many edits it has.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3600000</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.logroll.errors.tolerated" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.logroll.errors.tolerated</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The number of consecutive WAL close errors we will allow before triggering a server abort. A setting of 0 will cause the region server to abort if closing the current WAL writer fails during log rolling. Even a small value (2 or 3) will allow a region server to ride over transient HDFS errors.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.hlog.reader.impl" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.hlog.reader.impl</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The WAL file reader implementation.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.regionserver.wal.ProtobufLogReader</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.hlog.writer.impl" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.hlog.writer.impl</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The WAL file writer implementation.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.regionserver.wal.ProtobufLogWriter</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.global.memstore.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.global.memstore.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum size of all memstores in a region server before new updates are blocked and flushes are forced. Defaults to 40% of heap (0.4). Updates are blocked and flushes are forced until size of all memstores in a region server hits hbase.regionserver.global.memstore.size.lower.limit. The default value in this configuration has been intentionally left emtpy in order to honor the old hbase.regionserver.global.memstore.upperLimit property if present.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.regionserver.global.memstore.size.lower.limit" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.global.memstore.size.lower.limit</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum size of all memstores in a region server before flushes are forced. Defaults to 95% of hbase.regionserver.global.memstore.size (0.95). A 100% value for this value causes the minimum possible flushing to occur when updates are blocked due to memstore limiting. The default value in this configuration has been intentionally left emtpy in order to honor the old hbase.regionserver.global.memstore.lowerLimit property if present.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.regionserver.optionalcacheflushinterval" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.optionalcacheflushinterval</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum amount of time an edit lives in memory before being automatically flushed. Default 1 hour. Set it to 0 to disable automatic flushing.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3600000</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.dns.interface" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.dns.interface</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The name of the Network Interface from which a region server should report its IP address.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>default</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.dns.nameserver" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.dns.nameserver</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The host name or IP address of the name server (DNS) which a region server should use to determine the host name used by the master for communication and display purposes.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>default</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.region.split.policy" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.region.split.policy</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A split policy determines when a region should be split. The various other split policies that are available currently are BusyRegionSplitPolicy, ConstantSizeRegionSplitPolicy, DisabledRegionSplitPolicy, DelimitedKeyPrefixRegionSplitPolicy, KeyPrefixRegionSplitPolicy etc.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.regionserver.IncreasingToUpperBoundRegionSplitPolicy</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.regionSplitLimit" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.regionSplitLimit</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Limit for the number of regions after which no more region splitting should take place. This is not hard limit for the number of regions but acts as a guideline for the regionserver to stop splitting after a certain limit. Default is set to 1000.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1000</code></p> </div> </dd> </dl> </div> <div id="zookeeper.session.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>zookeeper.session.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>ZooKeeper session timeout in milliseconds. It is used in two different ways. First, this value is used in the ZK client that HBase uses to connect to the ensemble. It is also used by HBase when it starts a ZK server and it is passed as the 'maxSessionTimeout'. See <a href="http://hadoop.apache.org/zookeeper/docs/current/zookeeperProgrammers.html#ch_zkSessions" class="bare">http://hadoop.apache.org/zookeeper/docs/current/zookeeperProgrammers.html#ch_zkSessions</a>. For example, if a HBase region server connects to a ZK ensemble that’s also managed by HBase, then the session timeout will be the one specified by this configuration. But, a region server that connects to an ensemble managed with a different configuration will be subjected that ensemble’s maxSessionTimeout. So, even though HBase might propose using 90 seconds, the ensemble can have a max timeout lower than this and it will take precedence. The current default that ZK ships with is 40 seconds, which is lower than HBase’s.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>90000</code></p> </div> </dd> </dl> </div> <div id="zookeeper.znode.parent" class="dlist"> <dl> <dt class="hdlist1"><code>zookeeper.znode.parent</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Root ZNode for HBase in ZooKeeper. All of HBase’s ZooKeeper files that are configured with a relative path will go under this node. By default, all of HBase’s ZooKeeper file path are configured with a relative path, so they will all go under this directory unless changed.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>/hbase</code></p> </div> </dd> </dl> </div> <div id="zookeeper.znode.acl.parent" class="dlist"> <dl> <dt class="hdlist1"><code>zookeeper.znode.acl.parent</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Root ZNode for access control lists.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>acl</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.dns.interface" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.dns.interface</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The name of the Network Interface from which a ZooKeeper server should report its IP address.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>default</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.dns.nameserver" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.dns.nameserver</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The host name or IP address of the name server (DNS) which a ZooKeeper server should use to determine the host name used by the master for communication and display purposes.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>default</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.peerport" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.peerport</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Port used by ZooKeeper peers to talk to each other. See <a href="http://hadoop.apache.org/zookeeper/docs/r3.1.1/zookeeperStarted.html#sc_RunningReplicatedZooKeeper" class="bare">http://hadoop.apache.org/zookeeper/docs/r3.1.1/zookeeperStarted.html#sc_RunningReplicatedZooKeeper</a> for more information.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2888</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.leaderport" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.leaderport</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Port used by ZooKeeper for leader election. See <a href="http://hadoop.apache.org/zookeeper/docs/r3.1.1/zookeeperStarted.html#sc_RunningReplicatedZooKeeper" class="bare">http://hadoop.apache.org/zookeeper/docs/r3.1.1/zookeeperStarted.html#sc_RunningReplicatedZooKeeper</a> for more information.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3888</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.useMulti" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.useMulti</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Instructs HBase to make use of ZooKeeper’s multi-update functionality. This allows certain ZooKeeper operations to complete more quickly and prevents some issues with rare Replication failure scenarios (see the release note of HBASE-2611 for an example). IMPORTANT: only set this to true if all ZooKeeper servers in the cluster are on version 3.4+ and will not be downgraded. ZooKeeper versions before 3.4 do not support multi-update and will not fail gracefully if multi-update is invoked (see ZOOKEEPER-1495).</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.config.read.zookeeper.config" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.config.read.zookeeper.config</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to allow HBaseConfiguration to read the zoo.cfg file for ZooKeeper properties. Switching this to true is not recommended, since the functionality of reading ZK properties from a zoo.cfg file has been deprecated.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.property.initLimit" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.property.initLimit</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Property from ZooKeeper’s config zoo.cfg. The number of ticks that the initial synchronization phase can take.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.property.syncLimit" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.property.syncLimit</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Property from ZooKeeper’s config zoo.cfg. The number of ticks that can pass between sending a request and getting an acknowledgment.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>5</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.property.dataDir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.property.dataDir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Property from ZooKeeper’s config zoo.cfg. The directory where the snapshot is stored.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>${hbase.tmp.dir}/zookeeper</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.property.clientPort" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.property.clientPort</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Property from ZooKeeper’s config zoo.cfg. The port at which the clients will connect.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2181</code></p> </div> </dd> </dl> </div> <div id="hbase.zookeeper.property.maxClientCnxns" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.zookeeper.property.maxClientCnxns</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Property from ZooKeeper’s config zoo.cfg. Limit on number of concurrent connections (at the socket level) that a single client, identified by IP address, may make to a single member of the ZooKeeper ensemble. Set high to avoid zk connection issues running standalone and pseudo-distributed.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>300</code></p> </div> </dd> </dl> </div> <div id="hbase.client.write.buffer" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.write.buffer</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Default size of the HTable client write buffer in bytes. A bigger buffer takes more memory — on both the client and server side since server instantiates the passed write buffer to process it — but a larger buffer size reduces the number of RPCs made. For an estimate of server-side memory-used, evaluate hbase.client.write.buffer * hbase.regionserver.handler.count</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2097152</code></p> </div> </dd> </dl> </div> <div id="hbase.client.pause" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.pause</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>General client pause value. Used mostly as value to wait before running a retry of a failed get, region lookup, etc. See hbase.client.retries.number for description of how we backoff from this initial pause amount and how this pause works w/ retries.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>100</code></p> </div> </dd> </dl> </div> <div id="hbase.client.pause.cqtbe" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.pause.cqtbe</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Whether or not to use a special client pause for CallQueueTooBigException (cqtbe). Set this property to a higher value than hbase.client.pause if you observe frequent CQTBE from the same RegionServer and the call queue there keeps full</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.client.retries.number" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.retries.number</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum retries. Used as maximum for all retryable operations such as the getting of a cell’s value, starting a row update, etc. Retry interval is a rough function based on hbase.client.pause. At first we retry at this interval but then with backoff, we pretty quickly reach retrying every ten seconds. See HConstants#RETRY_BACKOFF for how the backup ramps up. Change this setting and hbase.client.pause to suit your workload.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>35</code></p> </div> </dd> </dl> </div> <div id="hbase.client.max.total.tasks" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.max.total.tasks</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of concurrent mutation tasks a single HTable instance will send to the cluster.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>100</code></p> </div> </dd> </dl> </div> <div id="hbase.client.max.perserver.tasks" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.max.perserver.tasks</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of concurrent mutation tasks a single HTable instance will send to a single region server.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>5</code></p> </div> </dd> </dl> </div> <div id="hbase.client.max.perregion.tasks" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.max.perregion.tasks</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of concurrent mutation tasks the client will maintain to a single Region. That is, if there is already hbase.client.max.perregion.tasks writes in progress for this region, new puts won’t be sent to this region until some writes finishes.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1</code></p> </div> </dd> </dl> </div> <div id="hbase.client.perserver.requests.threshold" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.perserver.requests.threshold</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The max number of concurrent pending requests for one server in all client threads (process level). Exceeding requests will be thrown ServerTooBusyException immediately to prevent user’s threads being occupied and blocked by only one slow region server. If you use a fix number of threads to access HBase in a synchronous way, set this to a suitable value which is related to the number of threads will help you. See <a href="https://issues.apache.org/jira/browse/HBASE-16388" class="bare">https://issues.apache.org/jira/browse/HBASE-16388</a> for details.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2147483647</code></p> </div> </dd> </dl> </div> <div id="hbase.client.scanner.caching" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.scanner.caching</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Number of rows that we try to fetch when calling next on a scanner if it is not served from (local, client) memory. This configuration works together with hbase.client.scanner.max.result.size to try and use the network efficiently. The default value is Integer.MAX_VALUE by default so that the network will fill the chunk size defined by hbase.client.scanner.max.result.size rather than be limited by a particular number of rows since the size of rows varies table to table. If you know ahead of time that you will not require more than a certain number of rows from a scan, this configuration should be set to that row limit via Scan#setCaching. Higher caching values will enable faster scanners but will eat up more memory and some calls of next may take longer and longer times when the cache is empty. Do not set this value such that the time between invocations is greater than the scanner timeout; i.e. hbase.client.scanner.timeout.period</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2147483647</code></p> </div> </dd> </dl> </div> <div id="hbase.client.keyvalue.maxsize" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.keyvalue.maxsize</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Specifies the combined maximum allowed size of a KeyValue instance. This is to set an upper boundary for a single entry saved in a storage file. Since they cannot be split it helps avoiding that a region cannot be split any further because the data is too large. It seems wise to set this to a fraction of the maximum region size. Setting it to zero or less disables the check.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10485760</code></p> </div> </dd> </dl> </div> <div id="hbase.server.keyvalue.maxsize" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.server.keyvalue.maxsize</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum allowed size of an individual cell, inclusive of value and all key components. A value of 0 or less disables the check. The default value is 10MB. This is a safety setting to protect the server from OOM situations.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10485760</code></p> </div> </dd> </dl> </div> <div id="hbase.client.scanner.timeout.period" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.scanner.timeout.period</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Client scanner lease period in milliseconds.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>60000</code></p> </div> </dd> </dl> </div> <div id="hbase.client.localityCheck.threadPoolSize" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.localityCheck.threadPoolSize</code></dt> <dd> <div class="paragraph"> <div class="title">Default</div> <p><code>2</code></p> </div> </dd> </dl> </div> <div id="hbase.bulkload.retries.number" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.bulkload.retries.number</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum retries. This is maximum number of iterations to atomic bulk loads are attempted in the face of splitting operations 0 means never give up.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.master.balancer.maxRitPercent" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.balancer.maxRitPercent</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The max percent of regions in transition when balancing. The default value is 1.0. So there are no balancer throttling. If set this config to 0.01, It means that there are at most 1% regions in transition when balancing. Then the cluster’s availability is at least 99% when balancing.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1.0</code></p> </div> </dd> </dl> </div> <div id="hbase.balancer.period" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.balancer.period</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Period at which the region balancer runs in the Master.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>300000</code></p> </div> </dd> </dl> </div> <div id="hbase.normalizer.period" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.normalizer.period</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Period at which the region normalizer runs in the Master.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1800000</code></p> </div> </dd> </dl> </div> <div id="hbase.regions.slop" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regions.slop</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Rebalance if any regionserver has average + (average * slop) regions. The default value of this parameter is 0.001 in StochasticLoadBalancer (the default load balancer), while the default is 0.2 in other load balancers (i.e., SimpleLoadBalancer).</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.001</code></p> </div> </dd> </dl> </div> <div id="hbase.server.thread.wakefrequency" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.server.thread.wakefrequency</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Time to sleep in between searches for work (in milliseconds). Used as sleep interval by service threads such as log roller.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10000</code></p> </div> </dd> </dl> </div> <div id="hbase.server.versionfile.writeattempts" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.server.versionfile.writeattempts</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>How many time to retry attempting to write a version file before just aborting. Each attempt is seperated by the hbase.server.thread.wakefrequency milliseconds.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.memstore.flush.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.memstore.flush.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Memstore will be flushed to disk if size of the memstore exceeds this number of bytes. Value is checked by a thread that runs every hbase.server.thread.wakefrequency.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>134217728</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.percolumnfamilyflush.size.lower.bound" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.percolumnfamilyflush.size.lower.bound</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If FlushLargeStoresPolicy is used, then every time that we hit the total memstore limit, we find out all the column families whose memstores exceed this value, and only flush them, while retaining the others whose memstores are lower than this limit. If none of the families have their memstore size more than this, all the memstores will be flushed (just as usual). This value should be less than half of the total memstore threshold (hbase.hregion.memstore.flush.size).</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16777216</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.preclose.flush.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.preclose.flush.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If the memstores in a region are this size or larger when we go to close, run a "pre-flush" to clear out memstores before we put up the region closed flag and take the region offline. On close, a flush is run under the close flag to empty memory. During this time the region is offline and we are not taking on any writes. If the memstore content is large, this flush could take a long time to complete. The preflush is meant to clean out the bulk of the memstore before putting up the close flag and taking the region offline so the flush that runs under the close flag has little to do.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>5242880</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.memstore.block.multiplier" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.memstore.block.multiplier</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Block updates if memstore has hbase.hregion.memstore.block.multiplier times hbase.hregion.memstore.flush.size bytes. Useful preventing runaway memstore during spikes in update traffic. Without an upper-bound, memstore fills such that when it flushes the resultant flush files take a long time to compact or split, or worse, we OOME.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>4</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.memstore.mslab.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.memstore.mslab.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enables the MemStore-Local Allocation Buffer, a feature which works to prevent heap fragmentation under heavy write loads. This can reduce the frequency of stop-the-world GC pauses on large heaps.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.max.filesize" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.max.filesize</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum HStoreFile size. If any one of a column families' HStoreFiles has grown to exceed this value, the hosting HRegion is split in two.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10737418240</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.majorcompaction" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.majorcompaction</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The time (in miliseconds) between 'major' compactions of all HStoreFiles in a region. Default: Set to 7 days. Major compactions tend to happen exactly when you need them least so enable them such that they run at off-peak for your deploy; or, since this setting is on a periodicity that is unlikely to match your loading, run the compactions via an external invocation out of a cron job or some such.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>604800000</code></p> </div> </dd> </dl> </div> <div id="hbase.hregion.majorcompaction.jitter" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hregion.majorcompaction.jitter</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Jitter outer bound for major compactions. On each regionserver, we multiply the hbase.region.majorcompaction interval by some random fraction that is inside the bounds of this maximum. We then add this + or - product to when the next major compaction is to run. The idea is that major compaction does happen on every regionserver at exactly the same time. The smaller this number, the closer the compactions come together.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.50</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.compactionThreshold" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.compactionThreshold</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If more than this number of HStoreFiles in any one HStore (one HStoreFile is written per flush of memstore) then a compaction is run to rewrite all HStoreFiles files as one. Larger numbers put off compaction but when it runs, it takes longer to complete.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.flusher.count" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.flusher.count</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The number of flush threads. With less threads, the memstore flushes will be queued. With more threads, the flush will be executed in parallel, increasing the hdfs load. This can lead as well to more compactions.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.blockingStoreFiles" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.blockingStoreFiles</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If more than this number of StoreFiles in any one Store (one StoreFile is written per flush of MemStore) then updates are blocked for this HRegion until a compaction is completed, or until hbase.hstore.blockingWaitTime has been exceeded.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.blockingWaitTime" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.blockingWaitTime</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The time an HRegion will block updates for after hitting the StoreFile limit defined by hbase.hstore.blockingStoreFiles. After this time has elapsed, the HRegion will stop blocking updates even if a compaction has not been completed.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>90000</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.compaction.max" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.compaction.max</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Max number of HStoreFiles to compact per 'minor' compaction.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.compaction.kv.max" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.compaction.kv.max</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>How many KeyValues to read and then write in a batch when flushing or compacting. Do less if big KeyValues and problems with OOME. Do more if wide, small rows.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.time.to.purge.deletes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.time.to.purge.deletes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The amount of time to delay purging of delete markers with future timestamps. If unset, or set to 0, all delete markers, including those with future timestamps, are purged during the next major compaction. Otherwise, a delete marker is kept until the major compaction which occurs after the marker’s timestamp plus the value of this setting, in milliseconds.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.majorcompaction.pagecache.drop" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.majorcompaction.pagecache.drop</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Specifies whether to drop pages read/written into the system page cache by major compactions. Setting it to true helps prevent major compactions from polluting the page cache, which is almost always required, especially for clusters with low/moderate memory to storage ratio.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.minorcompaction.pagecache.drop" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.minorcompaction.pagecache.drop</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Specifies whether to drop pages read/written into the system page cache by minor compactions. Setting it to true helps prevent minor compactions from polluting the page cache, which is most beneficial on clusters with low memory to storage ratio or very write heavy clusters. You may want to set it to false under moderate to low write workload when bulk of the reads are on the most recently written data.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.storescanner.parallel.seek.enable" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.storescanner.parallel.seek.enable</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enables StoreFileScanner parallel-seeking in StoreScanner, a feature which can reduce response latency under special conditions.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.storescanner.parallel.seek.threads" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.storescanner.parallel.seek.threads</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The default thread pool size if parallel-seeking feature enabled.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hfile.block.cache.size" class="dlist"> <dl> <dt class="hdlist1"><code>hfile.block.cache.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Percentage of maximum heap (-Xmx setting) to allocate to block cache used by HFile/StoreFile. Default of 0.4 means allocate 40%. Set to 0 to disable but it’s not recommended; you need at least enough cache to hold the storefile indices.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.4</code></p> </div> </dd> </dl> </div> <div id="hfile.block.index.cacheonwrite" class="dlist"> <dl> <dt class="hdlist1"><code>hfile.block.index.cacheonwrite</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This allows to put non-root multi-level index blocks into the block cache at the time the index is being written.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hfile.index.block.max.size" class="dlist"> <dl> <dt class="hdlist1"><code>hfile.index.block.max.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>When the size of a leaf-level, intermediate-level, or root-level index block in a multi-level block index grows to this size, the block is written out and a new block is started.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>131072</code></p> </div> </dd> </dl> </div> <div id="hbase.bucketcache.ioengine" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.bucketcache.ioengine</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Where to store the contents of the bucketcache. One of: heap, offheap, or file. If a file, set it to file:PATH_TO_FILE. See <a href="http://hbase.apache.org/book.html#offheap.blockcache" class="bare">http://hbase.apache.org/book.html#offheap.blockcache</a> for more information.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.bucketcache.combinedcache.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.bucketcache.combinedcache.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Whether or not the bucketcache is used in league with the LRU on-heap block cache. In this mode, indices and blooms are kept in the LRU blockcache and the data blocks are kept in the bucketcache.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.bucketcache.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.bucketcache.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A float that EITHER represents a percentage of total heap memory size to give to the cache (if < 1.0) OR, it is the total capacity in megabytes of BucketCache. Default: 0.0</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.bucketcache.bucket.sizes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.bucketcache.bucket.sizes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of sizes for buckets for the bucketcache. Can be multiple sizes. List block sizes in order from smallest to largest. The sizes you use will depend on your data access patterns. Must be a multiple of 256 else you will run into 'java.io.IOException: Invalid HFile block magic' when you go to read from cache. If you specify no values here, then you pick up the default bucketsizes set in code (See BucketAllocator#DEFAULT_BUCKET_SIZES).</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hfile.format.version" class="dlist"> <dl> <dt class="hdlist1"><code>hfile.format.version</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The HFile format version to use for new files. Version 3 adds support for tags in hfiles (See <a href="http://hbase.apache.org/book.html#hbase.tags" class="bare">http://hbase.apache.org/book.html#hbase.tags</a>). Distributed Log Replay requires that tags are enabled. Also see the configuration 'hbase.replication.rpc.codec'.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>3</code></p> </div> </dd> </dl> </div> <div id="hfile.block.bloom.cacheonwrite" class="dlist"> <dl> <dt class="hdlist1"><code>hfile.block.bloom.cacheonwrite</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enables cache-on-write for inline blocks of a compound Bloom filter.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="io.storefile.bloom.block.size" class="dlist"> <dl> <dt class="hdlist1"><code>io.storefile.bloom.block.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The size in bytes of a single block ("chunk") of a compound Bloom filter. This size is approximate, because Bloom blocks can only be inserted at data block boundaries, and the number of keys per data block varies.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>131072</code></p> </div> </dd> </dl> </div> <div id="hbase.rs.cacheblocksonwrite" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rs.cacheblocksonwrite</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Whether an HFile block should be added to the block cache when the block is finished.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.rpc.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rpc.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This is for the RPC layer to define how long (millisecond) HBase client applications take for a remote call to time out. It uses pings to check connections but will eventually throw a TimeoutException.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>60000</code></p> </div> </dd> </dl> </div> <div id="hbase.client.operation.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.operation.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Operation timeout is a top-level restriction (millisecond) that makes sure a blocking operation in Table will not be blocked more than this. In each operation, if rpc request fails because of timeout or other reason, it will retry until success or throw RetriesExhaustedException. But if the total time being blocking reach the operation timeout before retries exhausted, it will break early and throw SocketTimeoutException.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1200000</code></p> </div> </dd> </dl> </div> <div id="hbase.cells.scanned.per.heartbeat.check" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.cells.scanned.per.heartbeat.check</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The number of cells scanned in between heartbeat checks. Heartbeat checks occur during the processing of scans to determine whether or not the server should stop scanning in order to send back a heartbeat message to the client. Heartbeat messages are used to keep the client-server connection alive during long running scans. Small values mean that the heartbeat checks will occur more often and thus will provide a tighter bound on the execution time of the scan. Larger values mean that the heartbeat checks occur less frequently</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10000</code></p> </div> </dd> </dl> </div> <div id="hbase.rpc.shortoperation.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rpc.shortoperation.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This is another version of "hbase.rpc.timeout". For those RPC operation within cluster, we rely on this configuration to set a short timeout limitation for short operation. For example, short rpc timeout for region server’s trying to report to active master can benefit quicker master failover process.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10000</code></p> </div> </dd> </dl> </div> <div id="hbase.ipc.client.tcpnodelay" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.ipc.client.tcpnodelay</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set no delay on rpc socket connections. See <a href="http://docs.oracle.com/javase/1.5.0/docs/api/java/net/Socket.html#getTcpNoDelay(" class="bare">http://docs.oracle.com/javase/1.5.0/docs/api/java/net/Socket.html#getTcpNoDelay(</a>)</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.hostname" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.hostname</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This config is for experts: don’t set its value unless you really know what you are doing. When set to a non-empty value, this represents the (external facing) hostname for the underlying server. See <a href="https://issues.apache.org/jira/browse/HBASE-12954" class="bare">https://issues.apache.org/jira/browse/HBASE-12954</a> for details.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.regionserver.hostname.disable.master.reversedns" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.hostname.disable.master.reversedns</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This config is for experts: don’t set its value unless you really know what you are doing. When set to true, regionserver will use the current node hostname for the servername and HMaster will skip reverse DNS lookup and use the hostname sent by regionserver instead. Note that this config and hbase.regionserver.hostname are mutually exclusive. See <a href="https://issues.apache.org/jira/browse/HBASE-18226" class="bare">https://issues.apache.org/jira/browse/HBASE-18226</a> for more details.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.master.keytab.file" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.keytab.file</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Full path to the kerberos keytab file to use for logging in the configured HMaster server principal.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.master.kerberos.principal" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.kerberos.principal</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Ex. "hbase/_HOST@EXAMPLE.COM". The kerberos principal name that should be used to run the HMaster process. The principal name should be in the form: user/hostname@DOMAIN. If "_HOST" is used as the hostname portion, it will be replaced with the actual hostname of the running instance.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.regionserver.keytab.file" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.keytab.file</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Full path to the kerberos keytab file to use for logging in the configured HRegionServer server principal.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.regionserver.kerberos.principal" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.kerberos.principal</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Ex. "hbase/_HOST@EXAMPLE.COM". The kerberos principal name that should be used to run the HRegionServer process. The principal name should be in the form: user/hostname@DOMAIN. If "_HOST" is used as the hostname portion, it will be replaced with the actual hostname of the running instance. An entry for this principal must exist in the file specified in hbase.regionserver.keytab.file</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hadoop.policy.file" class="dlist"> <dl> <dt class="hdlist1"><code>hadoop.policy.file</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The policy configuration file used by RPC servers to make authorization decisions on client requests. Only used when HBase security is enabled.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>hbase-policy.xml</code></p> </div> </dd> </dl> </div> <div id="hbase.superuser" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.superuser</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>List of users or groups (comma-separated), who are allowed full privileges, regardless of stored ACLs, across the cluster. Only used when HBase security is enabled.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.auth.key.update.interval" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.auth.key.update.interval</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The update interval for master key for authentication tokens in servers in milliseconds. Only used when HBase security is enabled.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>86400000</code></p> </div> </dd> </dl> </div> <div id="hbase.auth.token.max.lifetime" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.auth.token.max.lifetime</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum lifetime in milliseconds after which an authentication token expires. Only used when HBase security is enabled.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>604800000</code></p> </div> </dd> </dl> </div> <div id="hbase.ipc.client.fallback-to-simple-auth-allowed" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.ipc.client.fallback-to-simple-auth-allowed</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>When a client is configured to attempt a secure connection, but attempts to connect to an insecure server, that server may instruct the client to switch to SASL SIMPLE (unsecure) authentication. This setting controls whether or not the client will accept this instruction from the server. When false (the default), the client will not allow the fallback to SIMPLE authentication, and will abort the connection.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.ipc.server.fallback-to-simple-auth-allowed" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.ipc.server.fallback-to-simple-auth-allowed</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>When a server is configured to require secure connections, it will reject connection attempts from clients using SASL SIMPLE (unsecure) authentication. This setting allows secure servers to accept SASL SIMPLE connections from clients when the client requests. When false (the default), the server will not allow the fallback to SIMPLE authentication, and will reject the connection. WARNING: This setting should ONLY be used as a temporary measure while converting clients over to secure authentication. It MUST BE DISABLED for secure operation.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.coprocessor.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.coprocessor.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enables or disables coprocessor loading. If 'false' (disabled), any other coprocessor related configuration will be ignored.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.coprocessor.user.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.coprocessor.user.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enables or disables user (aka. table) coprocessor loading. If 'false' (disabled), any table coprocessor attributes in table descriptors will be ignored. If "hbase.coprocessor.enabled" is 'false' this setting has no effect.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.coprocessor.region.classes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.coprocessor.region.classes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of Coprocessors that are loaded by default on all tables. For any override coprocessor method, these classes will be called in order. After implementing your own Coprocessor, just put it in HBase’s classpath and add the fully qualified class name here. A coprocessor can also be loaded on demand by setting HTableDescriptor.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.rest.port" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.port</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The port for the HBase REST server.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>8080</code></p> </div> </dd> </dl> </div> <div id="hbase.rest.readonly" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.readonly</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Defines the mode the REST server will be started in. Possible values are: false: All HTTP methods are permitted - GET/PUT/POST/DELETE. true: Only the GET method is permitted.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.rest.threads.max" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.threads.max</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of threads of the REST server thread pool. Threads in the pool are reused to process REST requests. This controls the maximum number of requests processed concurrently. It may help to control the memory used by the REST server to avoid OOM issues. If the thread pool is full, incoming requests will be queued up and wait for some free threads.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>100</code></p> </div> </dd> </dl> </div> <div id="hbase.rest.threads.min" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.threads.min</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The minimum number of threads of the REST server thread pool. The thread pool always has at least these number of threads so the REST server is ready to serve incoming requests.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2</code></p> </div> </dd> </dl> </div> <div id="hbase.rest.support.proxyuser" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.support.proxyuser</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enables running the REST server to support proxy-user mode.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.defaults.for.version.skip" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.defaults.for.version.skip</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to skip the 'hbase.defaults.for.version' check. Setting this to true can be useful in contexts other than the other side of a maven generation; i.e. running in an ide. You’ll want to set this boolean to true to avoid seeing the RuntimException complaint: "hbase-default.xml file seems to be for and old version of HBase (\${hbase.version}), this version is X.X.X-SNAPSHOT"</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.coprocessor.master.classes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.coprocessor.master.classes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of org.apache.hadoop.hbase.coprocessor.MasterObserver coprocessors that are loaded by default on the active HMaster process. For any implemented coprocessor methods, the listed classes will be called in order. After implementing your own MasterObserver, just put it in HBase’s classpath and add the fully qualified class name here.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.coprocessor.abortonerror" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.coprocessor.abortonerror</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to cause the hosting server (master or regionserver) to abort if a coprocessor fails to load, fails to initialize, or throws an unexpected Throwable object. Setting this to false will allow the server to continue execution but the system wide state of the coprocessor in question will become inconsistent as it will be properly executing in only a subset of servers, so this is most useful for debugging only.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.online.schema.update.enable" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.online.schema.update.enable</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set true to enable online schema changes.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.table.lock.enable" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.table.lock.enable</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to enable locking the table in zookeeper for schema change operations. Table locking from master prevents concurrent schema modifications to corrupt table state.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.table.max.rowsize" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.table.max.rowsize</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum size of single row in bytes (default is 1 Gb) for Get’ting or Scan’ning without in-row scan flag set. If row size exceeds this limit RowTooBigException is thrown to client.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1073741824</code></p> </div> </dd> </dl> </div> <div id="hbase.thrift.minWorkerThreads" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.thrift.minWorkerThreads</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The "core size" of the thread pool. New threads are created on every connection until this many threads are created.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16</code></p> </div> </dd> </dl> </div> <div id="hbase.thrift.maxWorkerThreads" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.thrift.maxWorkerThreads</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum size of the thread pool. When the pending request queue overflows, new threads are created until their number reaches this number. After that, the server starts dropping connections.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1000</code></p> </div> </dd> </dl> </div> <div id="hbase.thrift.maxQueuedRequests" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.thrift.maxQueuedRequests</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of pending Thrift connections waiting in the queue. If there are no idle threads in the pool, the server queues requests. Only when the queue overflows, new threads are added, up to hbase.thrift.maxQueuedRequests threads.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1000</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.thrift.framed" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.thrift.framed</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Use Thrift TFramedTransport on the server side. This is the recommended transport for thrift servers and requires a similar setting on the client side. Changing this to false will select the default transport, vulnerable to DoS when malformed requests are issued due to THRIFT-601.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.thrift.framed.max_frame_size_in_mb" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.thrift.framed.max_frame_size_in_mb</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Default frame size when using framed transport</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.thrift.compact" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.thrift.compact</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Use Thrift TCompactProtocol binary serialization protocol.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.rootdir.perms" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rootdir.perms</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>FS Permissions for the root directory in a secure(kerberos) setup. When master starts, it creates the rootdir with this permissions or sets the permissions if it does not match.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>700</code></p> </div> </dd> </dl> </div> <div id="hbase.wal.dir.perms" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.wal.dir.perms</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>FS Permissions for the root WAL directory in a secure(kerberos) setup. When master starts, it creates the WAL dir with this permissions or sets the permissions if it does not match.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>700</code></p> </div> </dd> </dl> </div> <div id="hbase.data.umask.enable" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.data.umask.enable</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Enable, if true, that file permissions should be assigned to the files written by the regionserver</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.data.umask" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.data.umask</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>File permissions that should be used to write data files when hbase.data.umask.enable is true</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>000</code></p> </div> </dd> </dl> </div> <div id="hbase.snapshot.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.snapshot.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to allow snapshots to be taken / restored / cloned.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.snapshot.restore.take.failsafe.snapshot" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.snapshot.restore.take.failsafe.snapshot</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to take a snapshot before the restore operation. The snapshot taken will be used in case of failure, to restore the previous state. At the end of the restore operation this snapshot will be deleted</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.snapshot.restore.failsafe.name" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.snapshot.restore.failsafe.name</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Name of the failsafe snapshot taken by the restore operation. You can use the {snapshot.name}, {table.name} and {restore.timestamp} variables to create a name based on what you are restoring.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>hbase-failsafe-{snapshot.name}-{restore.timestamp}</code></p> </div> </dd> </dl> </div> <div id="hbase.snapshot.working.dir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.snapshot.working.dir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Location where the snapshotting process will occur. The location of the completed snapshots will not change, but the temporary directory where the snapshot process occurs will be set to this location. This can be a separate filesystem than the root directory, for performance increase purposes. See HBASE-21098 for more information</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.server.compactchecker.interval.multiplier" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.server.compactchecker.interval.multiplier</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The number that determines how often we scan to see if compaction is necessary. Normally, compactions are done after some events (such as memstore flush), but if region didn’t receive a lot of writes for some time, or due to different compaction policies, it may be necessary to check it periodically. The interval between checks is hbase.server.compactchecker.interval.multiplier multiplied by hbase.server.thread.wakefrequency.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1000</code></p> </div> </dd> </dl> </div> <div id="hbase.lease.recovery.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.lease.recovery.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>How long we wait on dfs lease recovery in total before giving up.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>900000</code></p> </div> </dd> </dl> </div> <div id="hbase.lease.recovery.dfs.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.lease.recovery.dfs.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>How long between dfs recover lease invocations. Should be larger than the sum of the time it takes for the namenode to issue a block recovery command as part of datanode; dfs.heartbeat.interval and the time it takes for the primary datanode, performing block recovery to timeout on a dead datanode; usually dfs.client.socket-timeout. See the end of HBASE-8389 for more.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>64000</code></p> </div> </dd> </dl> </div> <div id="hbase.column.max.version" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.column.max.version</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>New column family descriptors will use this value as the default number of versions to keep.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>1</code></p> </div> </dd> </dl> </div> <div id="hbase.dfs.client.read.shortcircuit.buffer.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.dfs.client.read.shortcircuit.buffer.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If the DFSClient configuration dfs.client.read.shortcircuit.buffer.size is unset, we will use what is configured here as the short circuit read default direct byte buffer size. DFSClient native default is 1MB; HBase keeps its HDFS files open so number of file blocks * 1MB soon starts to add up and threaten OOME because of a shortage of direct memory. So, we set it down from the default. Make it > the default hbase block size set in the HColumnDescriptor which is usually 64k.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>131072</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.checksum.verify" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.checksum.verify</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If set to true (the default), HBase verifies the checksums for hfile blocks. HBase writes checksums inline with the data when it writes out hfiles. HDFS (as of this writing) writes checksums to a separate file than the data file necessitating extra seeks. Setting this flag saves some on i/o. Checksum verification by HDFS will be internally disabled on hfile streams when this flag is set. If the hbase-checksum verification fails, we will switch back to using HDFS checksums (so do not disable HDFS checksums! And besides this feature applies to hfiles only, not to WALs). If this parameter is set to false, then hbase will not verify any checksums, instead it will depend on checksum verification being done in the HDFS client.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>true</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.bytes.per.checksum" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.bytes.per.checksum</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Number of bytes in a newly created checksum chunk for HBase-level checksums in hfile blocks.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16384</code></p> </div> </dd> </dl> </div> <div id="hbase.hstore.checksum.algorithm" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.hstore.checksum.algorithm</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Name of an algorithm that is used to compute checksums. Possible values are NULL, CRC32, CRC32C.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>CRC32C</code></p> </div> </dd> </dl> </div> <div id="hbase.client.scanner.max.result.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.client.scanner.max.result.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum number of bytes returned when calling a scanner’s next method. Note that when a single row is larger than this limit the row is still returned completely. The default value is 2MB, which is good for 1ge networks. With faster and/or high latency networks this value should be increased.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>2097152</code></p> </div> </dd> </dl> </div> <div id="hbase.server.scanner.max.result.size" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.server.scanner.max.result.size</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Maximum number of bytes returned when calling a scanner’s next method. Note that when a single row is larger than this limit the row is still returned completely. The default value is 100MB. This is a safety setting to protect the server from OOM situations.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>104857600</code></p> </div> </dd> </dl> </div> <div id="hbase.status.published" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.status.published</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This setting activates the publication by the master of the status of the region server. When a region server dies and its recovery starts, the master will push this information to the client application, to let them cut the connection immediately instead of waiting for a timeout.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.status.publisher.class" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.status.publisher.class</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Implementation of the status publication with a multicast message.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.master.ClusterStatusPublisher$MulticastPublisher</code></p> </div> </dd> </dl> </div> <div id="hbase.status.listener.class" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.status.listener.class</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Implementation of the status listener with a multicast message.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.client.ClusterStatusListener$MulticastListener</code></p> </div> </dd> </dl> </div> <div id="hbase.status.multicast.address.ip" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.status.multicast.address.ip</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Multicast address to use for the status publication by multicast.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>226.1.1.3</code></p> </div> </dd> </dl> </div> <div id="hbase.status.multicast.address.port" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.status.multicast.address.port</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Multicast port to use for the status publication by multicast.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>16100</code></p> </div> </dd> </dl> </div> <div id="hbase.dynamic.jars.dir" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.dynamic.jars.dir</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The directory from which the custom filter/co-processor jars can be loaded dynamically by the region server without the need to restart. However, an already loaded filter/co-processor class would not be un-loaded. See HBASE-1936 for more details.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>${hbase.rootdir}/lib</code></p> </div> </dd> </dl> </div> <div id="hbase.security.authentication" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.security.authentication</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Controls whether or not secure authentication is enabled for HBase. Possible values are 'simple' (no authentication), and 'kerberos'.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>simple</code></p> </div> </dd> </dl> </div> <div id="hbase.rest.filter.classes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.filter.classes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Servlet filters for REST service.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.rest.filter.GzipFilter</code></p> </div> </dd> </dl> </div> <div id="hbase.master.loadbalancer.class" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.loadbalancer.class</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Class used to execute the regions balancing when the period occurs. See the class comment for more on how it works <a href="http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/master/balancer/StochasticLoadBalancer.html" class="bare">http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/master/balancer/StochasticLoadBalancer.html</a> It replaces the DefaultLoadBalancer as the default (since renamed as the SimpleLoadBalancer).</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.master.balancer.StochasticLoadBalancer</code></p> </div> </dd> </dl> </div> <div id="hbase.rest.csrf.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest.csrf.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Set to true to enable protection against cross-site request forgery (CSRF)</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.rest-csrf.browser-useragents-regex" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rest-csrf.browser-useragents-regex</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of regular expressions used to match against an HTTP request’s User-Agent header when protection against cross-site request forgery (CSRF) is enabled for REST server by setting hbase.rest.csrf.enabled to true. If the incoming User-Agent matches any of these regular expressions, then the request is considered to be sent by a browser, and therefore CSRF prevention is enforced. If the request’s User-Agent does not match any of these regular expressions, then the request is considered to be sent by something other than a browser, such as scripted automation. In this case, CSRF is not a potential attack vector, so the prevention is not enforced. This helps achieve backwards-compatibility with existing automation that has not been updated to send the CSRF prevention header.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code><sup>Mozilla.<strong>,</sup>Opera.</strong></code></p> </div> </dd> </dl> </div> <div id="hbase.security.exec.permission.checks" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.security.exec.permission.checks</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>If this setting is enabled and ACL based access control is active (the AccessController coprocessor is installed either as a system coprocessor or on a table as a table coprocessor) then you must grant all relevant users EXEC privilege if they require the ability to execute coprocessor endpoint calls. EXEC privilege, like any other permission, can be granted globally to a user, or to a user on a per table or per namespace basis. For more information on coprocessor endpoints, see the coprocessor section of the HBase online manual. For more information on granting or revoking permissions using the AccessController, see the security section of the HBase online manual.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.procedure.regionserver.classes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.procedure.regionserver.classes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of org.apache.hadoop.hbase.procedure.RegionServerProcedureManager procedure managers that are loaded by default on the active HRegionServer process. The lifecycle methods (init/start/stop) will be called by the active HRegionServer process to perform the specific globally barriered procedure. After implementing your own RegionServerProcedureManager, just put it in HBase’s classpath and add the fully qualified class name here.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.procedure.master.classes" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.procedure.master.classes</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma-separated list of org.apache.hadoop.hbase.procedure.MasterProcedureManager procedure managers that are loaded by default on the active HMaster process. A procedure is identified by its signature and users can use the signature and an instant name to trigger an execution of a globally barriered procedure. After implementing your own MasterProcedureManager, just put it in HBase’s classpath and add the fully qualified class name here.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p>none</p> </div> </dd> </dl> </div> <div id="hbase.coordinated.state.manager.class" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.coordinated.state.manager.class</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Fully qualified name of class implementing coordinated state manager.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.coordination.ZkCoordinatedStateManager</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.storefile.refresh.period" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.storefile.refresh.period</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The period (in milliseconds) for refreshing the store files for the secondary regions. 0 means this feature is disabled. Secondary regions sees new files (from flushes and compactions) from primary once the secondary region refreshes the list of files in the region (there is no notification mechanism). But too frequent refreshes might cause extra Namenode pressure. If the files cannot be refreshed for longer than HFile TTL (hbase.master.hfilecleaner.ttl) the requests are rejected. Configuring HFile TTL to a larger value is also recommended with this setting.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0</code></p> </div> </dd> </dl> </div> <div id="hbase.region.replica.replication.enabled" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.region.replica.replication.enabled</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Whether asynchronous WAL replication to the secondary region replicas is enabled or not. If this is enabled, a replication peer named "region_replica_replication" will be created which will tail the logs and replicate the mutatations to region replicas for tables that have region replication > 1. If this is enabled once, disabling this replication also requires disabling the replication peer using shell or ReplicationAdmin java class. Replication to secondary region replicas works over standard inter-cluster replication. So replication, if disabled explicitly, also has to be enabled by setting "hbase.replication" to true for this feature to work.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.http.filter.initializers" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.http.filter.initializers</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>A comma separated list of class names. Each class in the list must extend org.apache.hadoop.hbase.http.FilterInitializer. The corresponding Filter will be initialized. Then, the Filter will be applied to all user facing jsp and servlet web pages. The ordering of the list defines the ordering of the filters. The default StaticUserWebFilter add a user principal as defined by the hbase.http.staticuser.user property.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.http.lib.StaticUserWebFilter</code></p> </div> </dd> </dl> </div> <div id="hbase.security.visibility.mutations.checkauths" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.security.visibility.mutations.checkauths</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>This property if enabled, will check whether the labels in the visibility expression are associated with the user issuing the mutation</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>false</code></p> </div> </dd> </dl> </div> <div id="hbase.http.max.threads" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.http.max.threads</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of threads that the HTTP Server will create in its ThreadPool.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.replication.rpc.codec" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.replication.rpc.codec</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The codec that is to be used when replication is enabled so that the tags are also replicated. This is used along with HFileV3 which supports tags in them. If tags are not used or if the hfile version used is HFileV2 then KeyValueCodec can be used as the replication codec. Note that using KeyValueCodecWithTags for replication when there are no tags causes no harm.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.codec.KeyValueCodecWithTags</code></p> </div> </dd> </dl> </div> <div id="hbase.replication.source.maxthreads" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.replication.source.maxthreads</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The maximum number of threads any replication source will use for shipping edits to the sinks in parallel. This also limits the number of chunks each replication batch is broken into. Larger values can improve the replication throughput between the master and slave clusters. The default of 10 will rarely need to be changed.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>10</code></p> </div> </dd> </dl> </div> <div id="hbase.http.staticuser.user" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.http.staticuser.user</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The user name to filter as, on static web filters while rendering content. An example use is the HDFS web UI (user to be used for browsing files).</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>dr.stack</code></p> </div> </dd> </dl> </div> <div id="hbase.master.normalizer.class" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.master.normalizer.class</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Class used to execute the region normalization when the period occurs. See the class comment for more on how it works <a href="http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/master/normalizer/SimpleRegionNormalizer.html" class="bare">http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/master/normalizer/SimpleRegionNormalizer.html</a></p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>org.apache.hadoop.hbase.master.normalizer.SimpleRegionNormalizer</code></p> </div> </dd> </dl> </div> <div id="hbase.regionserver.handler.abort.on.error.percent" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.regionserver.handler.abort.on.error.percent</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>The percent of region server RPC threads failed to abort RS. -1 Disable aborting; 0 Abort if even a single handler has died; 0.x Abort only when this percent of handlers have died; 1 Abort only all of the handers have died.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>0.5</code></p> </div> </dd> </dl> </div> <div id="hbase.snapshot.master.timeout.millis" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.snapshot.master.timeout.millis</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Timeout for master for the snapshot procedure execution</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>300000</code></p> </div> </dd> </dl> </div> <div id="hbase.snapshot.region.timeout" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.snapshot.region.timeout</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Timeout for regionservers to keep threads in snapshot request pool waiting</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>300000</code></p> </div> </dd> </dl> </div> <div id="hbase.rpc.rows.warning.threshold" class="dlist"> <dl> <dt class="hdlist1"><code>hbase.rpc.rows.warning.threshold</code></dt> <dd> <div class="paragraph"> <div class="title">Description</div> <p>Number of rows in a batch operation above which a warning will be logged.</p> </div> <div class="paragraph"> <div class="title">Default</div> <p><code>5000</code></p> </div> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="hbase.env.sh"><a class="anchor" href="#hbase.env.sh"></a>7.3. <em>hbase-env.sh</em></h3> <div class="paragraph"> <p>Set HBase environment variables in this file. Examples include options to pass the JVM on start of an HBase daemon such as heap size and garbage collector configs. You can also set configurations for HBase configuration, log directories, niceness, ssh options, where to locate process pid files, etc. Open the file at <em>conf/hbase-env.sh</em> and peruse its content. Each option is fairly well documented. Add your own environment variables here if you want them read by HBase daemons on startup.</p> </div> <div class="paragraph"> <p>Changes here will require a cluster restart for HBase to notice the change.</p> </div> </div> <div class="sect2"> <h3 id="log4j"><a class="anchor" href="#log4j"></a>7.4. <em>log4j.properties</em></h3> <div class="paragraph"> <p>Edit this file to change rate at which HBase files are rolled and to change the level at which HBase logs messages.</p> </div> <div class="paragraph"> <p>Changes here will require a cluster restart for HBase to notice the change though log levels can be changed for particular daemons via the HBase UI.</p> </div> </div> <div class="sect2"> <h3 id="client_dependencies"><a class="anchor" href="#client_dependencies"></a>7.5. Client configuration and dependencies connecting to an HBase cluster</h3> <div class="paragraph"> <p>If you are running HBase in standalone mode, you don’t need to configure anything for your client to work provided that they are all on the same machine.</p> </div> <div class="paragraph"> <p>Since the HBase Master may move around, clients bootstrap by looking to ZooKeeper for current critical locations. ZooKeeper is where all these values are kept. Thus clients require the location of the ZooKeeper ensemble before they can do anything else. Usually this the ensemble location is kept out in the <em>hbase-site.xml</em> and is picked up by the client from the <code>CLASSPATH</code>.</p> </div> <div class="paragraph"> <p>If you are configuring an IDE to run a HBase client, you should include the <em>conf/</em> directory on your classpath so <em>hbase-site.xml</em> settings can be found (or add <em>src/test/resources</em> to pick up the hbase-site.xml used by tests).</p> </div> <div class="paragraph"> <p>Minimally, a client of HBase needs several libraries in its <code>CLASSPATH</code> when connecting to a cluster, including:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">commons-configuration (commons-configuration-<span class="float">1.6</span>.jar) commons-lang (commons-lang-<span class="float">2.5</span>.jar) commons-logging (commons-logging-<span class="float">1.1</span><span class="float">.1</span>.jar) hadoop-core (hadoop-core-<span class="float">1.0</span><span class="float">.0</span>.jar) hbase (hbase-<span class="float">0.92</span><span class="float">.0</span>.jar) log4j (log4j-<span class="float">1.2</span><span class="float">.16</span>.jar) slf4j-api (slf4j-api-<span class="float">1.5</span><span class="float">.8</span>.jar) slf4j-log4j (slf4j-log4j12-<span class="float">1.5</span><span class="float">.8</span>.jar) zookeeper (zookeeper-<span class="float">3.4</span><span class="float">.2</span>.jar)</code></pre> </div> </div> <div class="paragraph"> <p>An example basic <em>hbase-site.xml</em> for client only might look as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="preprocessor"><?xml version="1.0"?></span> <span class="preprocessor"><?xml-stylesheet type="text/xsl" href="configuration.xsl"?></span> <span class="tag"><configuration></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.quorum<span class="tag"></name></span> <span class="tag"><value></span>example1,example2,example3<span class="tag"></value></span> <span class="tag"><description></span>The directory shared by region servers. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"></configuration></span></code></pre> </div> </div> <div class="sect3"> <h4 id="java.client.config"><a class="anchor" href="#java.client.config"></a>7.5.1. Java client configuration</h4> <div class="paragraph"> <p>The configuration used by a Java client is kept in an <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HBaseConfiguration">HBaseConfiguration</a> instance.</p> </div> <div class="paragraph"> <p>The factory method on HBaseConfiguration, <code>HBaseConfiguration.create();</code>, on invocation, will read in the content of the first <em>hbase-site.xml</em> found on the client’s <code>CLASSPATH</code>, if one is present (Invocation will also factor in any <em>hbase-default.xml</em> found; an <em>hbase-default.xml</em> ships inside the <em>hbase.X.X.X.jar</em>). It is also possible to specify configuration directly without having to read from a <em>hbase-site.xml</em>. For example, to set the ZooKeeper ensemble for the cluster programmatically do as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); config.set(<span class="string"><span class="delimiter">"</span><span class="content">hbase.zookeeper.quorum</span><span class="delimiter">"</span></span>, <span class="string"><span class="delimiter">"</span><span class="content">localhost</span><span class="delimiter">"</span></span>); <span class="comment">// Here we are running zookeeper locally</span></code></pre> </div> </div> <div class="paragraph"> <p>If multiple ZooKeeper instances make up your ZooKeeper ensemble, they may be specified in a comma-separated list (just as in the <em>hbase-site.xml</em> file). This populated <code>Configuration</code> instance can then be passed to an <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html">Table</a>, and so on.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="example_config"><a class="anchor" href="#example_config"></a>8. Example Configurations</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_basic_distributed_hbase_install"><a class="anchor" href="#_basic_distributed_hbase_install"></a>8.1. Basic Distributed HBase Install</h3> <div class="paragraph"> <p>Here is an example basic configuration for a distributed ten node cluster: * The nodes are named <code>example0</code>, <code>example1</code>, etc., through node <code>example9</code> in this example. * The HBase Master and the HDFS NameNode are running on the node <code>example0</code>. * RegionServers run on nodes <code>example1</code>-<code>example9</code>. * A 3-node ZooKeeper ensemble runs on <code>example1</code>, <code>example2</code>, and <code>example3</code> on the default ports. * ZooKeeper data is persisted to the directory <em>/export/zookeeper</em>.</p> </div> <div class="paragraph"> <p>Below we show what the main configuration files — <em>hbase-site.xml</em>, <em>regionservers</em>, and <em>hbase-env.sh</em> — found in the HBase <em>conf</em> directory might look like.</p> </div> <div class="sect3"> <h4 id="hbase_site"><a class="anchor" href="#hbase_site"></a>8.1.1. <em>hbase-site.xml</em></h4> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="preprocessor"><?xml version="1.0"?></span> <span class="preprocessor"><?xml-stylesheet type="text/xsl" href="configuration.xsl"?></span> <span class="tag"><configuration></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.quorum<span class="tag"></name></span> <span class="tag"><value></span>example1,example2,example3<span class="tag"></value></span> <span class="tag"><description></span>The directory shared by RegionServers. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.property.dataDir<span class="tag"></name></span> <span class="tag"><value></span>/export/zookeeper<span class="tag"></value></span> <span class="tag"><description></span>Property from ZooKeeper config zoo.cfg. The directory where the snapshot is stored. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rootdir<span class="tag"></name></span> <span class="tag"><value></span>hdfs://example0:8020/hbase<span class="tag"></value></span> <span class="tag"><description></span>The directory shared by RegionServers. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.cluster.distributed<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span>The mode the cluster will be in. Possible values are false: standalone and pseudo-distributed setups with managed Zookeeper true: fully-distributed with unmanaged Zookeeper Quorum (see hbase-env.sh) <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"></configuration></span></code></pre> </div> </div> </div> <div class="sect3"> <h4 id="regionservers"><a class="anchor" href="#regionservers"></a>8.1.2. <em>regionservers</em></h4> <div class="paragraph"> <p>In this file you list the nodes that will run RegionServers. In our case, these nodes are <code>example1</code>-<code>example9</code>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">example1 example2 example3 example4 example5 example6 example7 example8 example9</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="hbase_env"><a class="anchor" href="#hbase_env"></a>8.1.3. <em>hbase-env.sh</em></h4> <div class="paragraph"> <p>The following lines in the <em>hbase-env.sh</em> file show how to set the <code>JAVA_HOME</code> environment variable (required for HBase 0.98.5 and newer) and set the heap to 4 GB (rather than the default value of 1 GB). If you copy and paste this example, be sure to adjust the <code>JAVA_HOME</code> to suit your environment.</p> </div> <div class="listingblock"> <div class="content"> <pre># The java implementation to use. export JAVA_HOME=/usr/java/jdk1.7.0/ # The maximum amount of heap to use. Default is left to JVM default. export HBASE_HEAPSIZE=4G</pre> </div> </div> <div class="paragraph"> <p>Use rsync to copy the content of the <em>conf</em> directory to all nodes of the cluster.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="important_configurations"><a class="anchor" href="#important_configurations"></a>9. The Important Configurations</h2> <div class="sectionbody"> <div class="paragraph"> <p>Below we list some <em>important</em> configurations. We’ve divided this section into required configuration and worth-a-look recommended configs.</p> </div> <div class="sect2"> <h3 id="required_configuration"><a class="anchor" href="#required_configuration"></a>9.1. Required Configurations</h3> <div class="paragraph"> <p>Review the <a href="#os">os</a> and <a href="#hadoop">hadoop</a> sections.</p> </div> <div class="sect3"> <h4 id="big.cluster.config"><a class="anchor" href="#big.cluster.config"></a>9.1.1. Big Cluster Configurations</h4> <div class="paragraph"> <p>If you have a cluster with a lot of regions, it is possible that a Regionserver checks in briefly after the Master starts while all the remaining RegionServers lag behind. This first server to check in will be assigned all regions which is not optimal. To prevent the above scenario from happening, up the <code>hbase.master.wait.on.regionservers.mintostart</code> property from its default value of 1. See <a href="https://issues.apache.org/jira/browse/HBASE-6389">HBASE-6389 Modify the conditions to ensure that Master waits for sufficient number of Region Servers before starting region assignments</a> for more detail.</p> </div> </div> <div class="sect3"> <h4 id="backup.master.fail.fast"><a class="anchor" href="#backup.master.fail.fast"></a>9.1.2. If a backup Master exists, make the primary Master fail fast</h4> <div class="paragraph"> <p>If the primary Master loses its connection with ZooKeeper, it will fall into a loop where it keeps trying to reconnect. Disable this functionality if you are running more than one Master: i.e. a backup Master. Failing to do so, the dying Master may continue to receive RPCs though another Master has assumed the role of primary. See the configuration <a href="#fail.fast.expired.active.master">fail.fast.expired.active.master</a>.</p> </div> </div> </div> <div class="sect2"> <h3 id="_recommended_configurations"><a class="anchor" href="#_recommended_configurations"></a>9.2. Recommended Configurations</h3> <div class="sect3"> <h4 id="recommended_configurations.zk"><a class="anchor" href="#recommended_configurations.zk"></a>9.2.1. ZooKeeper Configuration</h4> <div class="sect4"> <h5 id="sect.zookeeper.session.timeout"><a class="anchor" href="#sect.zookeeper.session.timeout"></a><code>zookeeper.session.timeout</code></h5> <div class="paragraph"> <p>The default timeout is three minutes (specified in milliseconds). This means that if a server crashes, it will be three minutes before the Master notices the crash and starts recovery. You might like to tune the timeout down to a minute or even less so the Master notices failures the sooner. Before changing this value, be sure you have your JVM garbage collection configuration under control otherwise, a long garbage collection that lasts beyond the ZooKeeper session timeout will take out your RegionServer (You might be fine with this — you probably want recovery to start on the server if a RegionServer has been in GC for a long period of time).</p> </div> <div class="paragraph"> <p>To change this configuration, edit <em>hbase-site.xml</em>, copy the changed file around the cluster and restart.</p> </div> <div class="paragraph"> <p>We set this value high to save our having to field questions up on the mailing lists asking why a RegionServer went down during a massive import. The usual cause is that their JVM is untuned and they are running into long GC pauses. Our thinking is that while users are getting familiar with HBase, we’d save them having to know all of its intricacies. Later when they’ve built some confidence, then they can play with configuration such as this.</p> </div> </div> <div class="sect4"> <h5 id="zookeeper.instances"><a class="anchor" href="#zookeeper.instances"></a>Number of ZooKeeper Instances</h5> <div class="paragraph"> <p>See <a href="#zookeeper">zookeeper</a>.</p> </div> </div> </div> <div class="sect3"> <h4 id="recommended.configurations.hdfs"><a class="anchor" href="#recommended.configurations.hdfs"></a>9.2.2. HDFS Configurations</h4> <div class="sect4"> <h5 id="dfs.datanode.failed.volumes.tolerated"><a class="anchor" href="#dfs.datanode.failed.volumes.tolerated"></a>dfs.datanode.failed.volumes.tolerated</h5> <div class="paragraph"> <p>This is the "…number of volumes that are allowed to fail before a DataNode stops offering service. By default any volume failure will cause a datanode to shutdown" from the <em>hdfs-default.xml</em> description. You might want to set this to about half the amount of your available disks.</p> </div> </div> </div> <div class="sect3"> <h4 id="hbase.regionserver.handler.count_description"><a class="anchor" href="#hbase.regionserver.handler.count_description"></a>9.2.3. <code>hbase.regionserver.handler.count</code></h4> <div class="paragraph"> <p>This setting defines the number of threads that are kept open to answer incoming requests to user tables. The rule of thumb is to keep this number low when the payload per request approaches the MB (big puts, scans using a large cache) and high when the payload is small (gets, small puts, ICVs, deletes). The total size of the queries in progress is limited by the setting <code>hbase.ipc.server.max.callqueue.size</code>.</p> </div> <div class="paragraph"> <p>It is safe to set that number to the maximum number of incoming clients if their payload is small, the typical example being a cluster that serves a website since puts aren’t typically buffered and most of the operations are gets.</p> </div> <div class="paragraph"> <p>The reason why it is dangerous to keep this setting high is that the aggregate size of all the puts that are currently happening in a region server may impose too much pressure on its memory, or even trigger an OutOfMemoryError. A RegionServer running on low memory will trigger its JVM’s garbage collector to run more frequently up to a point where GC pauses become noticeable (the reason being that all the memory used to keep all the requests' payloads cannot be trashed, no matter how hard the garbage collector tries). After some time, the overall cluster throughput is affected since every request that hits that RegionServer will take longer, which exacerbates the problem even more.</p> </div> <div class="paragraph"> <p>You can get a sense of whether you have too little or too many handlers by <a href="#rpc.logging">rpc.logging</a> on an individual RegionServer then tailing its logs (Queued requests consume memory).</p> </div> </div> <div class="sect3"> <h4 id="big_memory"><a class="anchor" href="#big_memory"></a>9.2.4. Configuration for large memory machines</h4> <div class="paragraph"> <p>HBase ships with a reasonable, conservative configuration that will work on nearly all machine types that people might want to test with. If you have larger machines — HBase has 8G and larger heap — you might the following configuration options helpful. TODO.</p> </div> </div> <div class="sect3"> <h4 id="config.compression"><a class="anchor" href="#config.compression"></a>9.2.5. Compression</h4> <div class="paragraph"> <p>You should consider enabling ColumnFamily compression. There are several options that are near-frictionless and in most all cases boost performance by reducing the size of StoreFiles and thus reducing I/O.</p> </div> <div class="paragraph"> <p>See <a href="#compression">compression</a> for more information.</p> </div> </div> <div class="sect3"> <h4 id="config.wals"><a class="anchor" href="#config.wals"></a>9.2.6. Configuring the size and number of WAL files</h4> <div class="paragraph"> <p>HBase uses <a href="#wal">wal</a> to recover the memstore data that has not been flushed to disk in case of an RS failure. These WAL files should be configured to be slightly smaller than HDFS block (by default a HDFS block is 64Mb and a WAL file is ~60Mb).</p> </div> <div class="paragraph"> <p>HBase also has a limit on the number of WAL files, designed to ensure there’s never too much data that needs to be replayed during recovery. This limit needs to be set according to memstore configuration, so that all the necessary data would fit. It is recommended to allocate enough WAL files to store at least that much data (when all memstores are close to full). For example, with 16Gb RS heap, default memstore settings (0.4), and default WAL file size (~60Mb), 16Gb*0.4/60, the starting point for WAL file count is ~109. However, as all memstores are not expected to be full all the time, less WAL files can be allocated.</p> </div> </div> <div class="sect3"> <h4 id="disable.splitting"><a class="anchor" href="#disable.splitting"></a>9.2.7. Managed Splitting</h4> <div class="paragraph"> <p>HBase generally handles splitting your regions, based upon the settings in your <em>hbase-default.xml</em> and <em>hbase-site.xml</em> configuration files. Important settings include <code>hbase.regionserver.region.split.policy</code>, <code>hbase.hregion.max.filesize</code>, <code>hbase.regionserver.regionSplitLimit</code>. A simplistic view of splitting is that when a region grows to <code>hbase.hregion.max.filesize</code>, it is split. For most use patterns, most of the time, you should use automatic splitting. See <a href="#manual_region_splitting_decisions">manual region splitting decisions</a> for more information about manual region splitting.</p> </div> <div class="paragraph"> <p>Instead of allowing HBase to split your regions automatically, you can choose to manage the splitting yourself. This feature was added in HBase 0.90.0. Manually managing splits works if you know your keyspace well, otherwise let HBase figure where to split for you. Manual splitting can mitigate region creation and movement under load. It also makes it so region boundaries are known and invariant (if you disable region splitting). If you use manual splits, it is easier doing staggered, time-based major compactions to spread out your network IO load.</p> </div> <div class="paragraph"> <div class="title">Disable Automatic Splitting</div> <p>To disable automatic splitting, set <code>hbase.hregion.max.filesize</code> to a very large value, such as <code>100 GB</code> It is not recommended to set it to its absolute maximum value of <code>Long.MAX_VALUE</code>.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Automatic Splitting Is Recommended</div> <div class="paragraph"> <p>If you disable automatic splits to diagnose a problem or during a period of fast data growth, it is recommended to re-enable them when your situation becomes more stable. The potential benefits of managing region splits yourself are not undisputed.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <div class="title">Determine the Optimal Number of Pre-Split Regions</div> <p>The optimal number of pre-split regions depends on your application and environment. A good rule of thumb is to start with 10 pre-split regions per server and watch as data grows over time. It is better to err on the side of too few regions and perform rolling splits later. The optimal number of regions depends upon the largest StoreFile in your region. The size of the largest StoreFile will increase with time if the amount of data grows. The goal is for the largest region to be just large enough that the compaction selection algorithm only compacts it during a timed major compaction. Otherwise, the cluster can be prone to compaction storms where a large number of regions under compaction at the same time. It is important to understand that the data growth causes compaction storms, and not the manual split decision.</p> </div> <div class="paragraph"> <p>If the regions are split into too many large regions, you can increase the major compaction interval by configuring <code>HConstants.MAJOR_COMPACTION_PERIOD</code>. HBase 0.90 introduced <code>org.apache.hadoop.hbase.util.RegionSplitter</code>, which provides a network-IO-safe rolling split of all regions.</p> </div> </div> <div class="sect3"> <h4 id="managed.compactions"><a class="anchor" href="#managed.compactions"></a>9.2.8. Managed Compactions</h4> <div class="paragraph"> <p>By default, major compactions are scheduled to run once in a 7-day period. Prior to HBase 0.96.x, major compactions were scheduled to happen once per day by default.</p> </div> <div class="paragraph"> <p>If you need to control exactly when and how often major compaction runs, you can disable managed major compactions. See the entry for <code>hbase.hregion.majorcompaction</code> in the <a href="#compaction.parameters">compaction.parameters</a> table for details.</p> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="title">Do Not Disable Major Compactions</div> <div class="paragraph"> <p>Major compactions are absolutely necessary for StoreFile clean-up. Do not disable them altogether. You can run major compactions manually via the HBase shell or via the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Admin.html#majorCompact(org.apache.hadoop.hbase.TableName)">Admin API</a>.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>For more information about compactions and the compaction file selection process, see <a href="#compaction">compaction</a></p> </div> </div> <div class="sect3"> <h4 id="spec.ex"><a class="anchor" href="#spec.ex"></a>9.2.9. Speculative Execution</h4> <div class="paragraph"> <p>Speculative Execution of MapReduce tasks is on by default, and for HBase clusters it is generally advised to turn off Speculative Execution at a system-level unless you need it for a specific case, where it can be configured per-job. Set the properties <code>mapreduce.map.speculative</code> and <code>mapreduce.reduce.speculative</code> to false.</p> </div> </div> </div> <div class="sect2"> <h3 id="other_configuration"><a class="anchor" href="#other_configuration"></a>9.3. Other Configurations</h3> <div class="sect3"> <h4 id="balancer_config"><a class="anchor" href="#balancer_config"></a>9.3.1. Balancer</h4> <div class="paragraph"> <p>The balancer is a periodic operation which is run on the master to redistribute regions on the cluster. It is configured via <code>hbase.balancer.period</code> and defaults to 300000 (5 minutes).</p> </div> <div class="paragraph"> <p>See <a href="#master.processes.loadbalancer">master.processes.loadbalancer</a> for more information on the LoadBalancer.</p> </div> </div> <div class="sect3"> <h4 id="disabling.blockcache"><a class="anchor" href="#disabling.blockcache"></a>9.3.2. Disabling Blockcache</h4> <div class="paragraph"> <p>Do not turn off block cache (You’d do it by setting <code>hbase.block.cache.size</code> to zero). Currently we do not do well if you do this because the RegionServer will spend all its time loading HFile indices over and over again. If your working set it such that block cache does you no good, at least size the block cache such that HFile indices will stay up in the cache (you can get a rough idea on the size you need by surveying RegionServer UIs; you’ll see index block size accounted near the top of the webpage).</p> </div> </div> <div class="sect3"> <h4 id="nagles"><a class="anchor" href="#nagles"></a>9.3.3. <a href="http://en.wikipedia.org/wiki/Nagle’s_algorithm">Nagle’s</a> or the small package problem</h4> <div class="paragraph"> <p>If a big 40ms or so occasional delay is seen in operations against HBase, try the Nagles' setting. For example, see the user mailing list thread, <a href="http://search-hadoop.com/m/pduLg2fydtE/Inconsistent+scan+performance+with+caching+set+&subj=Re+Inconsistent+scan+performance+with+caching+set+to+1">Inconsistent scan performance with caching set to 1</a> and the issue cited therein where setting <code>notcpdelay</code> improved scan speeds. You might also see the graphs on the tail of <a href="https://issues.apache.org/jira/browse/HBASE-7008">HBASE-7008 Set scanner caching to a better default</a> where our Lars Hofhansl tries various data sizes w/ Nagle’s on and off measuring the effect.</p> </div> </div> <div class="sect3"> <h4 id="mttr"><a class="anchor" href="#mttr"></a>9.3.4. Better Mean Time to Recover (MTTR)</h4> <div class="paragraph"> <p>This section is about configurations that will make servers come back faster after a fail. See the Deveraj Das an Nicolas Liochon blog post <a href="http://hortonworks.com/blog/introduction-to-hbase-mean-time-to-recover-mttr/">Introduction to HBase Mean Time to Recover (MTTR)</a> for a brief introduction.</p> </div> <div class="paragraph"> <p>The issue <a href="https://issues.apache.org/jira/browse/HBASE-8389">HBASE-8354 forces Namenode into loop with lease recovery requests</a> is messy but has a bunch of good discussion toward the end on low timeouts and how to effect faster recovery including citation of fixes added to HDFS. Read the Varun Sharma comments. The below suggested configurations are Varun’s suggestions distilled and tested. Make sure you are running on a late-version HDFS so you have the fixes he refers too and himself adds to HDFS that help HBase MTTR (e.g. HDFS-3703, HDFS-3712, and HDFS-4791 — Hadoop 2 for sure has them and late Hadoop 1 has some). Set the following in the RegionServer.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.lease.recovery.dfs.timeout<span class="tag"></name></span> <span class="tag"><value></span>23000<span class="tag"></value></span> <span class="tag"><description></span>How much time we allow elapse between calls to recover lease. Should be larger than the dfs timeout.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.client.socket-timeout<span class="tag"></name></span> <span class="tag"><value></span>10000<span class="tag"></value></span> <span class="tag"><description></span>Down the DFS timeout from 60 to 10 seconds.<span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>And on the NameNode/DataNode side, set the following to enable 'staleness' introduced in HDFS-3703, HDFS-3912.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>dfs.client.socket-timeout<span class="tag"></name></span> <span class="tag"><value></span>10000<span class="tag"></value></span> <span class="tag"><description></span>Down the DFS timeout from 60 to 10 seconds.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.datanode.socket.write.timeout<span class="tag"></name></span> <span class="tag"><value></span>10000<span class="tag"></value></span> <span class="tag"><description></span>Down the DFS timeout from 8 * 60 to 10 seconds.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>ipc.client.connect.timeout<span class="tag"></name></span> <span class="tag"><value></span>3000<span class="tag"></value></span> <span class="tag"><description></span>Down from 60 seconds to 3.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>ipc.client.connect.max.retries.on.timeouts<span class="tag"></name></span> <span class="tag"><value></span>2<span class="tag"></value></span> <span class="tag"><description></span>Down from 45 seconds to 3 (2 == 3 retries).<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.namenode.avoid.read.stale.datanode<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span>Enable stale state in hdfs<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.namenode.stale.datanode.interval<span class="tag"></name></span> <span class="tag"><value></span>20000<span class="tag"></value></span> <span class="tag"><description></span>Down from default 30 seconds<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.namenode.avoid.write.stale.datanode<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span>Enable stale state in hdfs<span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> </div> <div class="sect3"> <h4 id="jmx_config"><a class="anchor" href="#jmx_config"></a>9.3.5. JMX</h4> <div class="paragraph"> <p>JMX (Java Management Extensions) provides built-in instrumentation that enables you to monitor and manage the Java VM. To enable monitoring and management from remote systems, you need to set system property <code>com.sun.management.jmxremote.port</code> (the port number through which you want to enable JMX RMI connections) when you start the Java VM. See the <a href="http://docs.oracle.com/javase/6/docs/technotes/guides/management/agent.html">official documentation</a> for more information. Historically, besides above port mentioned, JMX opens two additional random TCP listening ports, which could lead to port conflict problem. (See <a href="https://issues.apache.org/jira/browse/HBASE-10289">HBASE-10289</a> for details)</p> </div> <div class="paragraph"> <p>As an alternative, You can use the coprocessor-based JMX implementation provided by HBase. To enable it in 0.99 or above, add below property in <em>hbase-site.xml</em>:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.regionserver.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.JMXListener<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> DO NOT set <code>com.sun.management.jmxremote.port</code> for Java VM at the same time. </td> </tr> </table> </div> <div class="paragraph"> <p>Currently it supports Master and RegionServer Java VM. By default, the JMX listens on TCP port 10102, you can further configure the port using below properties:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>regionserver.rmi.registry.port<span class="tag"></name></span> <span class="tag"><value></span>61130<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>regionserver.rmi.connector.port<span class="tag"></name></span> <span class="tag"><value></span>61140<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>The registry port can be shared with connector port in most cases, so you only need to configure regionserver.rmi.registry.port. However if you want to use SSL communication, the 2 ports must be configured to different values.</p> </div> <div class="paragraph"> <p>By default the password authentication and SSL communication is disabled. To enable password authentication, you need to update <em>hbase-env.sh</em> like below:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">export HBASE_JMX_BASE="-Dcom.sun.management.jmxremote.authenticate=true \ -Dcom.sun.management.jmxremote.password.file=your_password_file \ -Dcom.sun.management.jmxremote.access.file=your_access_file" export HBASE_MASTER_OPTS="$HBASE_MASTER_OPTS $HBASE_JMX_BASE " export HBASE_REGIONSERVER_OPTS="$HBASE_REGIONSERVER_OPTS $HBASE_JMX_BASE "</code></pre> </div> </div> <div class="paragraph"> <p>See example password/access file under <em>$JRE_HOME/lib/management</em>.</p> </div> <div class="paragraph"> <p>To enable SSL communication with password authentication, follow below steps:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">#1. generate a key pair, stored in myKeyStore keytool -genkey -alias jconsole -keystore myKeyStore #2. export it to file jconsole.cert keytool -export -alias jconsole -keystore myKeyStore -file jconsole.cert #3. copy jconsole.cert to jconsole client machine, import it to jconsoleKeyStore keytool -import -alias jconsole -keystore jconsoleKeyStore -file jconsole.cert</code></pre> </div> </div> <div class="paragraph"> <p>And then update <em>hbase-env.sh</em> like below:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">export HBASE_JMX_BASE="-Dcom.sun.management.jmxremote.ssl=true \ -Djavax.net.ssl.keyStore=/home/tianq/myKeyStore \ -Djavax.net.ssl.keyStorePassword=your_password_in_step_1 \ -Dcom.sun.management.jmxremote.authenticate=true \ -Dcom.sun.management.jmxremote.password.file=your_password file \ -Dcom.sun.management.jmxremote.access.file=your_access_file" export HBASE_MASTER_OPTS="$HBASE_MASTER_OPTS $HBASE_JMX_BASE " export HBASE_REGIONSERVER_OPTS="$HBASE_REGIONSERVER_OPTS $HBASE_JMX_BASE "</code></pre> </div> </div> <div class="paragraph"> <p>Finally start <code>jconsole</code> on the client using the key store:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">jconsole -J-Djavax.net.ssl.trustStore=/home/tianq/jconsoleKeyStore</code></pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> To enable the HBase JMX implementation on Master, you also need to add below property in <em>hbase-site.xml</em>: </td> </tr> </table> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><ame></span>hbase.coprocessor.master.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.JMXListener<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>The corresponding properties for port configuration are <code>master.rmi.registry.port</code> (by default 10101) and <code>master.rmi.connector.port</code> (by default the same as registry.port)</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="dyn_config"><a class="anchor" href="#dyn_config"></a>10. Dynamic Configuration</h2> <div class="sectionbody"> <div class="paragraph"> <p>Since HBase 1.0.0, it is possible to change a subset of the configuration without requiring a server restart. In the HBase shell, there are new operators, <code>update_config</code> and <code>update_all_config</code> that will prompt a server or all servers to reload configuration.</p> </div> <div class="paragraph"> <p>Only a subset of all configurations can currently be changed in the running server. Here is an incomplete list: <code>hbase.regionserver.thread.compaction.large</code>, <code>hbase.regionserver.thread.compaction.small</code>, <code>hbase.regionserver.thread.split</code>, <code>hbase.regionserver.thread.merge</code>, as well as compaction policy and configurations and adjustment to offpeak hours. For the full list consult the patch attached to <a href="https://issues.apache.org/jira/browse/HBASE-12147">HBASE-12147 Porting Online Config Change from 89-fb</a>.</p> </div> </div> </div> <h1 id="upgrading" class="sect0"><a class="anchor" href="#upgrading"></a>Upgrading</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>You cannot skip major versions when upgrading. If you are upgrading from version 0.90.x to 0.94.x, you must first go from 0.90.x to 0.92.x and then go from 0.92.x to 0.94.x.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> It may be possible to skip across versions — for example go from 0.92.2 straight to 0.98.0 just following the 0.96.x upgrade instructions — but these scenarios are untested. </td> </tr> </table> </div> <div class="paragraph"> <p>Review <a href="#configuration">Apache HBase Configuration</a>, in particular <a href="#hadoop"><a href="http://hadoop.apache.org">Hadoop</a></a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="hbase.versioning"><a class="anchor" href="#hbase.versioning"></a>11. HBase version number and compatibility</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase has two versioning schemes, pre-1.0 and post-1.0. Both are detailed below.</p> </div> <div class="sect2"> <h3 id="hbase.versioning.post10"><a class="anchor" href="#hbase.versioning.post10"></a>11.1. Post 1.0 versions</h3> <div class="paragraph"> <p>Starting with the 1.0.0 release, HBase is working towards <a href="http://semver.org/">Semantic Versioning</a> for its release versioning. In summary:</p> </div> <div class="ulist"> <div class="title">Given a version number MAJOR.MINOR.PATCH, increment the:</div> <ul> <li> <p>MAJOR version when you make incompatible API changes,</p> </li> <li> <p>MINOR version when you add functionality in a backwards-compatible manner, and</p> </li> <li> <p>PATCH version when you make backwards-compatible bug fixes.</p> </li> <li> <p>Additional labels for pre-release and build metadata are available as extensions to the MAJOR.MINOR.PATCH format.</p> </li> </ul> </div> <div id="hbase.versioning.compat" class="paragraph"> <div class="title">Compatibility Dimensions</div> <p>In addition to the usual API versioning considerations HBase has other compatibility dimensions that we need to consider.</p> </div> <div class="ulist"> <div class="title">Client-Server wire protocol compatibility</div> <ul> <li> <p>Allows updating client and server out of sync.</p> </li> <li> <p>We could only allow upgrading the server first. I.e. the server would be backward compatible to an old client, that way new APIs are OK.</p> </li> <li> <p>Example: A user should be able to use an old client to connect to an upgraded cluster.</p> </li> </ul> </div> <div class="ulist"> <div class="title">Server-Server protocol compatibility</div> <ul> <li> <p>Servers of different versions can co-exist in the same cluster.</p> </li> <li> <p>The wire protocol between servers is compatible.</p> </li> <li> <p>Workers for distributed tasks, such as replication and log splitting, can co-exist in the same cluster.</p> </li> <li> <p>Dependent protocols (such as using ZK for coordination) will also not be changed.</p> </li> <li> <p>Example: A user can perform a rolling upgrade.</p> </li> </ul> </div> <div class="ulist"> <div class="title">File format compatibility</div> <ul> <li> <p>Support file formats backward and forward compatible</p> </li> <li> <p>Example: File, ZK encoding, directory layout is upgraded automatically as part of an HBase upgrade. User can rollback to the older version and everything will continue to work.</p> </li> </ul> </div> <div class="ulist"> <div class="title">Client API compatibility</div> <ul> <li> <p>Allow changing or removing existing client APIs.</p> </li> <li> <p>An API needs to be deprecated for a whole major version before we will change/remove it.</p> <div class="ulist"> <ul> <li> <p>An example: An API was deprecated in 2.0.1 and will be marked for deletion in 4.0.0. On the other hand, an API deprecated in 2.0.0 can be removed in 3.0.0.</p> </li> </ul> </div> </li> <li> <p>APIs available in a patch version will be available in all later patch versions. However, new APIs may be added which will not be available in earlier patch versions.</p> </li> <li> <p>Example: A user using a newly deprecated api does not need to modify application code with hbase api calls until the next major version.</p> </li> </ul> </div> <div class="ulist"> <div class="title">Client Binary compatibility</div> <ul> <li> <p>Client code written to APIs available in a given patch release can run unchanged (no recompilation needed) against the new jars of later patch versions.</p> </li> <li> <p>Client code written to APIs available in a given patch release might not run against the old jars from an earlier patch version.</p> </li> <li> <p>Example: Old compiled client code will work unchanged with the new jars.</p> </li> </ul> </div> <div class="ulist"> <div class="title">Server-Side Limited API compatibility (taken from Hadoop)</div> <ul> <li> <p>Internal APIs are marked as Stable, Evolving, or Unstable</p> </li> <li> <p>This implies binary compatibility for coprocessors and plugins (pluggable classes, including replication) as long as these are only using marked interfaces/classes.</p> </li> <li> <p>Example: Old compiled Coprocessor, Filter, or Plugin code will work unchanged with the new jars.</p> </li> </ul> </div> <div class="ulist"> <div class="title">Dependency Compatibility</div> <ul> <li> <p>An upgrade of HBase will not require an incompatible upgrade of a dependent project, including the Java runtime.</p> </li> <li> <p>Example: An upgrade of Hadoop will not invalidate any of the compatibilities guarantees we made.</p> </li> </ul> </div> <div class="ulist"> <div class="title">Operational Compatibility</div> <ul> <li> <p>Metric changes</p> </li> <li> <p>Behavioral changes of services</p> </li> <li> <p>Web page APIs</p> </li> </ul> </div> <div class="ulist"> <div class="title">Summary</div> <ul> <li> <p>A patch upgrade is a drop-in replacement. Any change that is not Java binary compatible would not be allowed.<sup class="footnote">[<a id="_footnoteref_1" class="footnote" href="#_footnote_1" title="View footnote.">1</a>]</sup>. Downgrading versions within patch releases may not be compatible.</p> </li> <li> <p>A minor upgrade requires no application/client code modification. Ideally it would be a drop-in replacement but client code, coprocessors, filters, etc might have to be recompiled if new jars are used.</p> </li> <li> <p>A major upgrade allows the HBase community to make breaking changes.</p> </li> </ul> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 3. Compatibility Matrix <sup class="footnote">[<a id="_footnoteref_2" class="footnote" href="#_footnote_2" title="View footnote.">2</a>]</sup></caption> <colgroup> <col style="width: 25%;"> <col style="width: 25%;"> <col style="width: 25%;"> <col style="width: 25%;"> </colgroup> <tbody> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Major</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Minor</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Patch</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Client-Server wire Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Server-Server Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">File Format Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N <sup class="footnote">[<a id="_footnoteref_3" class="footnote" href="#_footnote_3" title="View footnote.">3</a>]</sup></p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Client API Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Client Binary Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-left valign-top" colspan="4"><p class="tableblock">Server-Side Limited API Compatibility</p></td> </tr> <tr> <td class="tableblock halign-right valign-top"><p class="tableblock">Stable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-right valign-top"><p class="tableblock">Evolving</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-right valign-top"><p class="tableblock">Unstable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Dependency Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Operational Compatibility</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">N</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Y</p></td> </tr> </tbody> </table> <div class="sect3"> <h4 id="hbase.client.api.surface"><a class="anchor" href="#hbase.client.api.surface"></a>11.1.1. HBase API Surface</h4> <div class="paragraph"> <p>HBase has a lot of API points, but for the compatibility matrix above, we differentiate between Client API, Limited Private API, and Private API. HBase uses a version of <a href="https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-common/Compatibility.html">Hadoop’s Interface classification</a>. HBase’s Interface classification classes can be found <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/classification/package-summary.html">here</a>.</p> </div> <div class="ulist"> <ul> <li> <p>InterfaceAudience: captures the intended audience, possible values are Public (for end users and external projects), LimitedPrivate (for other Projects, Coprocessors or other plugin points), and Private (for internal use).</p> </li> <li> <p>InterfaceStability: describes what types of interface changes are permitted. Possible values are Stable, Evolving, Unstable, and Deprecated.</p> </li> </ul> </div> <div id="hbase.client.api" class="dlist"> <dl> <dt class="hdlist1">HBase Client API</dt> <dd> <p>HBase Client API consists of all the classes or methods that are marked with InterfaceAudience.Public interface. All main classes in hbase-client and dependent modules have either InterfaceAudience.Public, InterfaceAudience.LimitedPrivate, or InterfaceAudience.Private marker. Not all classes in other modules (hbase-server, etc) have the marker. If a class is not annotated with one of these, it is assumed to be a InterfaceAudience.Private class.</p> </dd> </dl> </div> <div id="hbase.limitetprivate.api" class="dlist"> <dl> <dt class="hdlist1">HBase LimitedPrivate API</dt> <dd> <p>LimitedPrivate annotation comes with a set of target consumers for the interfaces. Those consumers are coprocessors, phoenix, replication endpoint implemnetations or similar. At this point, HBase only guarantees source and binary compatibility for these interfaces between patch versions.</p> </dd> </dl> </div> <div id="hbase.private.api" class="dlist"> <dl> <dt class="hdlist1">HBase Private API</dt> <dd> <p>All classes annotated with InterfaceAudience.Private or all classes that do not have the annotation are for HBase internal use only. The interfaces and method signatures can change at any point in time. If you are relying on a particular interface that is marked Private, you should open a jira to propose changing the interface to be Public or LimitedPrivate, or an interface exposed for this purpose.</p> </dd> </dl> </div> </div> </div> <div class="sect2"> <h3 id="hbase.versioning.pre10"><a class="anchor" href="#hbase.versioning.pre10"></a>11.2. Pre 1.0 versions</h3> <div class="paragraph"> <p>Before the semantic versioning scheme pre-1.0, HBase tracked either Hadoop’s versions (0.2x) or 0.9x versions. If you are into the arcane, checkout our old wiki page on <a href="http://wiki.apache.org/hadoop/Hbase/HBaseVersions">HBase Versioning</a> which tries to connect the HBase version dots. Below sections cover ONLY the releases before 1.0.</p> </div> <div id="hbase.development.series" class="paragraph"> <div class="title">Odd/Even Versioning or "Development" Series Releases</div> <p>Ahead of big releases, we have been putting up preview versions to start the feedback cycle turning-over earlier. These "Development" Series releases, always odd-numbered, come with no guarantees, not even regards being able to upgrade between two sequential releases (we reserve the right to break compatibility across "Development" Series releases). Needless to say, these releases are not for production deploys. They are a preview of what is coming in the hope that interested parties will take the release for a test drive and flag us early if we there are issues we’ve missed ahead of our rolling a production-worthy release.</p> </div> <div class="paragraph"> <p>Our first "Development" Series was the 0.89 set that came out ahead of HBase 0.90.0. HBase 0.95 is another "Development" Series that portends HBase 0.96.0. 0.99.x is the last series in "developer preview" mode before 1.0. Afterwards, we will be using semantic versioning naming scheme (see above).</p> </div> <div id="hbase.binary.compatibility" class="paragraph"> <div class="title">Binary Compatibility</div> <p>When we say two HBase versions are compatible, we mean that the versions are wire and binary compatible. Compatible HBase versions means that clients can talk to compatible but differently versioned servers. It means too that you can just swap out the jars of one version and replace them with the jars of another, compatible version and all will just work. Unless otherwise specified, HBase point versions are (mostly) binary compatible. You can safely do rolling upgrades between binary compatible versions; i.e. across point versions: e.g. from 0.94.5 to 0.94.6. See link:[Does compatibility between versions also mean binary compatibility?] discussion on the HBase dev mailing list.</p> </div> </div> <div class="sect2"> <h3 id="hbase.rolling.upgrade"><a class="anchor" href="#hbase.rolling.upgrade"></a>11.3. Rolling Upgrades</h3> <div class="paragraph"> <p>A rolling upgrade is the process by which you update the servers in your cluster a server at a time. You can rolling upgrade across HBase versions if they are binary or wire compatible. See <a href="#hbase.rolling.restart">Rolling Upgrade Between Versions that are Binary/Wire Compatible</a> for more on what this means. Coarsely, a rolling upgrade is a graceful stop each server, update the software, and then restart. You do this for each server in the cluster. Usually you upgrade the Master first and then the RegionServers. See <a href="#rolling">Rolling Restart</a> for tools that can help use the rolling upgrade process.</p> </div> <div class="paragraph"> <p>For example, in the below, HBase was symlinked to the actual HBase install. On upgrade, before running a rolling restart over the cluser, we changed the symlink to point at the new HBase software version and then ran</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_HOME=~/hadoop-2.6.0-CRC-SNAPSHOT ~/hbase/bin/rolling-restart.sh --config ~/conf_hbase</code></pre> </div> </div> <div class="paragraph"> <p>The rolling-restart script will first gracefully stop and restart the master, and then each of the RegionServers in turn. Because the symlink was changed, on restart the server will come up using the new HBase version. Check logs for errors as the rolling upgrade proceeds.</p> </div> <div id="hbase.rolling.restart" class="paragraph"> <div class="title">Rolling Upgrade Between Versions that are Binary/Wire Compatible</div> <p>Unless otherwise specified, HBase point versions are binary compatible. You can do a <a href="#hbase.rolling.upgrade">Rolling Upgrades</a> between HBase point versions. For example, you can go to 0.94.6 from 0.94.5 by doing a rolling upgrade across the cluster replacing the 0.94.5 binary with a 0.94.6 binary.</p> </div> <div class="paragraph"> <p>In the minor version-particular sections below, we call out where the versions are wire/protocol compatible and in this case, it is also possible to do a <a href="#hbase.rolling.upgrade">Rolling Upgrades</a>. For example, in <a href="#upgrade1.0.rolling.upgrade">Rolling upgrade from 0.98.x to HBase 1.0.0</a>, we state that it is possible to do a rolling upgrade between hbase-0.98.x and hbase-1.0.0.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_upgrade_paths"><a class="anchor" href="#_upgrade_paths"></a>12. Upgrade Paths</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="upgrade1.4"><a class="anchor" href="#upgrade1.4"></a>12.1. Upgrading to 1.4+</h3> <div class="sect3"> <h4 id="_replication_peer_s_tablecfs_config"><a class="anchor" href="#_replication_peer_s_tablecfs_config"></a>12.1.1. Replication peer’s TableCFs config</h4> <div class="paragraph"> <p>Before 1.4, the table name can’t include namespace for replication peer’s TableCFs config. It was fixed by add TableCFs to ReplicationPeerConfig which was stored on Zookeeper. So when upgrade to 1.4, you have to update the original ReplicationPeerConfig data on Zookeeper firstly. There are four steps to upgrade when your cluster have a replication peer with TableCFs config.</p> </div> <div class="ulist"> <ul> <li> <p>Disable the replication peer.</p> </li> <li> <p>If master has permission to write replication peer znode, then rolling update master directly. If not, use TableCFsUpdater tool to update the replication peer’s config.</p> </li> </ul> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.replication.master.TableCFsUpdater update</pre> </div> </div> <div class="ulist"> <ul> <li> <p>Rolling update regionservers.</p> </li> <li> <p>Enable the replication peer.</p> </li> </ul> </div> <div class="paragraph"> <p>Notes:</p> </div> <div class="ulist"> <ul> <li> <p>Can’t use the old client(before 1.4) to change the replication peer’s config. Because the client will write config to Zookeeper directly, the old client will miss TableCFs config. And the old client write TableCFs config to the old tablecfs znode, it will not work for new version regionserver.</p> </li> </ul> </div> </div> </div> <div class="sect2"> <h3 id="upgrade1.0"><a class="anchor" href="#upgrade1.0"></a>12.2. Upgrading from 0.98.x to 1.x</h3> <div class="paragraph"> <p>In this section we first note the significant changes that come in with 1.0.0+ HBase and then we go over the upgrade process. Be sure to read the significant changes section with care so you avoid surprises.</p> </div> <div class="sect3"> <h4 id="_changes_of_note"><a class="anchor" href="#_changes_of_note"></a>12.2.1. Changes of Note!</h4> <div class="paragraph"> <p>In here we list important changes that are in 1.0.0+ since 0.98.x., changes you should be aware that will go into effect once you upgrade.</p> </div> <div id="zookeeper.3.4" class="paragraph"> <div class="title">ZooKeeper 3.4 is required in HBase 1.0.0+</div> <p>See <a href="#zookeeper.requirements">ZooKeeper Requirements</a>.</p> </div> <div id="default.ports.changed" class="paragraph"> <div class="title">HBase Default Ports Changed</div> <p>The ports used by HBase changed. They used to be in the 600XX range. In HBase 1.0.0 they have been moved up out of the ephemeral port range and are 160XX instead (Master web UI was 60010 and is now 16010; the RegionServer web UI was 60030 and is now 16030, etc.). If you want to keep the old port locations, copy the port setting configs from <em>hbase-default.xml</em> into <em>hbase-site.xml</em>, change them back to the old values from the HBase 0.98.x era, and ensure you’ve distributed your configurations before you restart.</p> </div> <div id="upgrade1.0.hbase.bucketcache.percentage.in.combinedcache" class="paragraph"> <div class="title">hbase.bucketcache.percentage.in.combinedcache configuration has been REMOVED</div> <p>You may have made use of this configuration if you are using BucketCache. If NOT using BucketCache, this change does not effect you. Its removal means that your L1 LruBlockCache is now sized using <code>hfile.block.cache.size</code> — i.e. the way you would size the on-heap L1 LruBlockCache if you were NOT doing BucketCache — and the BucketCache size is not whatever the setting for <code>hbase.bucketcache.size</code> is. You may need to adjust configs to get the LruBlockCache and BucketCache sizes set to what they were in 0.98.x and previous. If you did not set this config., its default value was 0.9. If you do nothing, your BucketCache will increase in size by 10%. Your L1 LruBlockCache will become <code>hfile.block.cache.size</code> times your java heap size (<code>hfile.block.cache.size</code> is a float between 0.0 and 1.0). To read more, see <a href="https://issues.apache.org/jira/browse/HBASE-11520">HBASE-11520 Simplify offheap cache config by removing the confusing "hbase.bucketcache.percentage.in.combinedcache"</a>.</p> </div> <div id="hbase-12068" class="paragraph"> <div class="title">If you have your own customer filters.</div> <p>See the release notes on the issue <a href="https://issues.apache.org/jira/browse/HBASE-12068">HBASE-12068 [Branch-1] Avoid need to always do KeyValueUtil#ensureKeyValue for Filter transformCell</a>; be sure to follow the recommendations therein.</p> </div> <div id="dlr" class="paragraph"> <div class="title">Distributed Log Replay</div> <p><a href="#distributed.log.replay">Distributed Log Replay</a> is off by default in HBase 1.0.0. Enabling it can make a big difference improving HBase MTTR. Enable this feature if you are doing a clean stop/start when you are upgrading. You cannot rolling upgrade to this feature (caveat if you are running on a version of HBase in excess of HBase 0.98.4 — see <a href="https://issues.apache.org/jira/browse/HBASE-12577">HBASE-12577 Disable distributed log replay by default</a> for more).</p> </div> <div class="paragraph"> <div class="title">Availability of Date Tiered Compaction.</div> <p>The Date Tiered Compaction feature available as of 0.98.19 is available in the 1.y release line starting in release 1.3.0. If you have enabled this feature for any tables you must upgrade to version 1.3.0 or later. If you attempt to use an earlier 1.y release, any tables configured to use date tiered compaction will fail to have their regions open.</p> </div> </div> <div class="sect3"> <h4 id="upgrade1.0.rolling.upgrade"><a class="anchor" href="#upgrade1.0.rolling.upgrade"></a>12.2.2. Rolling upgrade from 0.98.x to HBase 1.0.0</h4> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">From 0.96.x to 1.0.0</div> You cannot do a <a href="#hbase.rolling.upgrade">rolling upgrade</a> from 0.96.x to 1.0.0 without first doing a rolling upgrade to 0.98.x. See comment in <a href="https://issues.apache.org/jira/browse/HBASE-11164?focusedCommentId=14182330&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14182330">HBASE-11164 Document and test rolling updates from 0.98 → 1.0</a> for the why. Also because HBase 1.0.0 enables HFile v3 by default, <a href="https://issues.apache.org/jira/browse/HBASE-9801">HBASE-9801 Change the default HFile version to V3</a>, and support for HFile v3 only arrives in 0.98, this is another reason you cannot rolling upgrade from HBase 0.96.x; if the rolling upgrade stalls, the 0.96.x servers cannot open files written by the servers running the newer HBase 1.0.0 with HFile’s of version 3. </td> </tr> </table> </div> <div class="paragraph"> <p>There are no known issues running a <a href="#hbase.rolling.upgrade">rolling upgrade</a> from HBase 0.98.x to HBase 1.0.0.</p> </div> </div> <div class="sect3"> <h4 id="upgrade1.0.from.0.94"><a class="anchor" href="#upgrade1.0.from.0.94"></a>12.2.3. Upgrading to 1.0 from 0.94</h4> <div class="paragraph"> <p>You cannot rolling upgrade from 0.94.x to 1.x.x. You must stop your cluster, install the 1.x.x software, run the migration described at <a href="#executing.the.0.96.upgrade">Executing the 0.96 Upgrade</a> (substituting 1.x.x. wherever we make mention of 0.96.x in the section below), and then restart. Be sure to upgrade your ZooKeeper if it is a version less than the required 3.4.x.</p> </div> </div> </div> <div class="sect2"> <h3 id="upgrade0.98"><a class="anchor" href="#upgrade0.98"></a>12.3. Upgrading from 0.96.x to 0.98.x</h3> <div class="paragraph"> <p>A rolling upgrade from 0.96.x to 0.98.x works. The two versions are not binary compatible.</p> </div> <div class="paragraph"> <p>Additional steps are required to take advantage of some of the new features of 0.98.x, including cell visibility labels, cell ACLs, and transparent server side encryption. See <a href="#security">Securing Apache HBase</a> for more information. Significant performance improvements include a change to the write ahead log threading model that provides higher transaction throughput under high load, reverse scanners, MapReduce over snapshot files, and striped compaction.</p> </div> <div class="paragraph"> <p>Clients and servers can run with 0.98.x and 0.96.x versions. However, applications may need to be recompiled due to changes in the Java API.</p> </div> </div> <div class="sect2"> <h3 id="_upgrading_from_0_94_x_to_0_98_x"><a class="anchor" href="#_upgrading_from_0_94_x_to_0_98_x"></a>12.4. Upgrading from 0.94.x to 0.98.x</h3> <div class="paragraph"> <p>A rolling upgrade from 0.94.x directly to 0.98.x does not work. The upgrade path follows the same procedures as <a href="#upgrade0.96">Upgrading from 0.94.x to 0.96.x</a>. Additional steps are required to use some of the new features of 0.98.x. See <a href="#upgrade0.98">Upgrading from 0.96.x to 0.98.x</a> for an abbreviated list of these features.</p> </div> </div> <div class="sect2"> <h3 id="upgrade0.96"><a class="anchor" href="#upgrade0.96"></a>12.5. Upgrading from 0.94.x to 0.96.x</h3> <div class="sect3"> <h4 id="_the_singularity"><a class="anchor" href="#_the_singularity"></a>12.5.1. The "Singularity"</h4> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">HBase 0.96.x was EOL’d, September 1st, 2014</div> Do not deploy 0.96.x Deploy at least 0.98.x. See <a href="https://issues.apache.org/jira/browse/HBASE-11642">EOL 0.96</a>. </td> </tr> </table> </div> <div class="paragraph"> <p>You will have to stop your old 0.94.x cluster completely to upgrade. If you are replicating between clusters, both clusters will have to go down to upgrade. Make sure it is a clean shutdown. The less WAL files around, the faster the upgrade will run (the upgrade will split any log files it finds in the filesystem as part of the upgrade process). All clients must be upgraded to 0.96 too.</p> </div> <div class="paragraph"> <p>The API has changed. You will need to recompile your code against 0.96 and you may need to adjust applications to go against new APIs (TODO: List of changes).</p> </div> </div> <div class="sect3"> <h4 id="executing.the.0.96.upgrade"><a class="anchor" href="#executing.the.0.96.upgrade"></a>12.5.2. Executing the 0.96 Upgrade</h4> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">HDFS and ZooKeeper must be up!</div> HDFS and ZooKeeper should be up and running during the upgrade process. </td> </tr> </table> </div> <div class="paragraph"> <p>HBase 0.96.0 comes with an upgrade script. Run</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ bin/hbase upgrade</code></pre> </div> </div> <div class="paragraph"> <p>to see its usage. The script has two main modes: <code>-check</code>, and <code>-execute</code>.</p> </div> <div class="paragraph"> <div class="title">check</div> <p>The check step is run against a running 0.94 cluster. Run it from a downloaded 0.96.x binary. The check step is looking for the presence of HFile v1 files. These are unsupported in HBase 0.96.0. To have them rewritten as HFile v2 you must run a compaction.</p> </div> <div class="paragraph"> <p>The check step prints stats at the end of its run (grep for <code>“Result:”</code> in the log) printing absolute path of the tables it scanned, any HFile v1 files found, the regions containing said files (these regions will need a major compaction), and any corrupted files if found. A corrupt file is unreadable, and so is undefined (neither HFile v1 nor HFile v2).</p> </div> <div class="paragraph"> <p>To run the check step, run</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ bin/hbase upgrade -check</code></pre> </div> </div> <div class="paragraph"> <p>Here is sample output:</p> </div> <div class="listingblock"> <div class="content"> <pre>Tables Processed: hdfs://localhost:41020/myHBase/.META. hdfs://localhost:41020/myHBase/usertable hdfs://localhost:41020/myHBase/TestTable hdfs://localhost:41020/myHBase/t Count of HFileV1: 2 HFileV1: hdfs://localhost:41020/myHBase/usertable /fa02dac1f38d03577bd0f7e666f12812/family/249450144068442524 hdfs://localhost:41020/myHBase/usertable /ecdd3eaee2d2fcf8184ac025555bb2af/family/249450144068442512 Count of corrupted files: 1 Corrupted Files: hdfs://localhost:41020/myHBase/usertable/fa02dac1f38d03577bd0f7e666f12812/family/1 Count of Regions with HFileV1: 2 Regions to Major Compact: hdfs://localhost:41020/myHBase/usertable/fa02dac1f38d03577bd0f7e666f12812 hdfs://localhost:41020/myHBase/usertable/ecdd3eaee2d2fcf8184ac025555bb2af There are some HFileV1, or corrupt files (files with incorrect major version)</pre> </div> </div> <div class="paragraph"> <p>In the above sample output, there are two HFile v1 files in two regions, and one corrupt file. Corrupt files should probably be removed. The regions that have HFile v1s need to be major compacted. To major compact, start up the hbase shell and review how to compact an individual region. After the major compaction is done, rerun the check step and the HFile v1 files should be gone, replaced by HFile v2 instances.</p> </div> <div class="paragraph"> <p>By default, the check step scans the HBase root directory (defined as <code>hbase.rootdir</code> in the configuration). To scan a specific directory only, pass the <code>-dir</code> option.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ bin/hbase upgrade -check -dir /myHBase/testTable</code></pre> </div> </div> <div class="paragraph"> <p>The above command would detect HFile v1 files in the <em>/myHBase/testTable</em> directory.</p> </div> <div class="paragraph"> <p>Once the check step reports all the HFile v1 files have been rewritten, it is safe to proceed with the upgrade.</p> </div> <div class="paragraph"> <div class="title">execute</div> <p>After the <em>check</em> step shows the cluster is free of HFile v1, it is safe to proceed with the upgrade. Next is the <em>execute</em> step. You must <strong>SHUTDOWN YOUR 0.94.x CLUSTER</strong> before you can run the execute step. The execute step will not run if it detects running HBase masters or RegionServers.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="paragraph"> <p>HDFS and ZooKeeper should be up and running during the upgrade process. If zookeeper is managed by HBase, then you can start zookeeper so it is available to the upgrade by running</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./hbase/bin/hbase-daemon.sh start zookeeper</code></pre> </div> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>The execute upgrade step is made of three substeps.</p> </div> <div class="ulist"> <ul> <li> <p>Namespaces: HBase 0.96.0 has support for namespaces. The upgrade needs to reorder directories in the filesystem for namespaces to work.</p> </li> <li> <p>ZNodes: All znodes are purged so that new ones can be written in their place using a new protobuf’ed format and a few are migrated in place: e.g. replication and table state znodes</p> </li> <li> <p>WAL Log Splitting: If the 0.94.x cluster shutdown was not clean, we’ll split WAL logs as part of migration before we startup on 0.96.0. This WAL splitting runs slower than the native distributed WAL splitting because it is all inside the single upgrade process (so try and get a clean shutdown of the 0.94.0 cluster if you can).</p> </li> </ul> </div> <div class="paragraph"> <p>To run the <em>execute</em> step, make sure that first you have copied HBase 0.96.0 binaries everywhere under servers and under clients. Make sure the 0.94.0 cluster is down. Then do as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ bin/hbase upgrade -execute</code></pre> </div> </div> <div class="paragraph"> <p>Here is some sample output.</p> </div> <div class="listingblock"> <div class="content"> <pre>Starting Namespace upgrade Created version file at hdfs://localhost:41020/myHBase with version=7 Migrating table testTable to hdfs://localhost:41020/myHBase/.data/default/testTable ..... Created version file at hdfs://localhost:41020/myHBase with version=8 Successfully completed NameSpace upgrade. Starting Znode upgrade ..... Successfully completed Znode upgrade Starting Log splitting ... Successfully completed Log splitting</pre> </div> </div> <div class="paragraph"> <p>If the output from the execute step looks good, stop the zookeeper instance you started to do the upgrade:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./hbase/bin/hbase-daemon.sh stop zookeeper</code></pre> </div> </div> <div class="paragraph"> <p>Now start up hbase-0.96.0.</p> </div> </div> </div> <div class="sect2"> <h3 id="s096.migration.troubleshooting"><a class="anchor" href="#s096.migration.troubleshooting"></a>12.6. Troubleshooting</h3> <div id="s096.migration.troubleshooting.old.client" class="paragraph"> <div class="title">Old Client connecting to 0.96 cluster</div> <p>It will fail with an exception like the below. Upgrade.</p> </div> <div class="listingblock"> <div class="content"> <pre>17:22:15 Exception in thread "main" java.lang.IllegalArgumentException: Not a host:port pair: PBUF 17:22:15 * 17:22:15 api-compat-8.ent.cloudera.com �� ( 17:22:15 at org.apache.hadoop.hbase.util.Addressing.parseHostname(Addressing.java:60) 17:22:15 at org.apache.hadoop.hbase.ServerName.&init>(ServerName.java:101) 17:22:15 at org.apache.hadoop.hbase.ServerName.parseVersionedServerName(ServerName.java:283) 17:22:15 at org.apache.hadoop.hbase.MasterAddressTracker.bytesToServerName(MasterAddressTracker.java:77) 17:22:15 at org.apache.hadoop.hbase.MasterAddressTracker.getMasterAddress(MasterAddressTracker.java:61) 17:22:15 at org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.getMaster(HConnectionManager.java:703) 17:22:15 at org.apache.hadoop.hbase.client.HBaseAdmin.&init>(HBaseAdmin.java:126) 17:22:15 at Client_4_3_0.setup(Client_4_3_0.java:716) 17:22:15 at Client_4_3_0.main(Client_4_3_0.java:63)</pre> </div> </div> <div class="sect3"> <h4 id="_upgrading_code_meta_code_to_use_protocol_buffers_protobuf"><a class="anchor" href="#_upgrading_code_meta_code_to_use_protocol_buffers_protobuf"></a>12.6.1. Upgrading <code>META</code> to use Protocol Buffers (Protobuf)</h4> <div class="paragraph"> <p>When you upgrade from versions prior to 0.96, <code>META</code> needs to be converted to use protocol buffers. This is controlled by the configuration option <code>hbase.MetaMigrationConvertingToPB</code>, which is set to <code>true</code> by default. Therefore, by default, no action is required on your part.</p> </div> <div class="paragraph"> <p>The migration is a one-time event. However, every time your cluster starts, <code>META</code> is scanned to ensure that it does not need to be converted. If you have a very large number of regions, this scan can take a long time. Starting in 0.98.5, you can set <code>hbase.MetaMigrationConvertingToPB</code> to <code>false</code> in <em>hbase-site.xml</em>, to disable this start-up scan. This should be considered an expert-level setting.</p> </div> </div> </div> <div class="sect2"> <h3 id="upgrade0.94"><a class="anchor" href="#upgrade0.94"></a>12.7. Upgrading from 0.92.x to 0.94.x</h3> <div class="paragraph"> <p>We used to think that 0.92 and 0.94 were interface compatible and that you can do a rolling upgrade between these versions but then we figured that <a href="https://issues.apache.org/jira/browse/HBASE-5357">HBASE-5357 Use builder pattern in HColumnDescriptor</a> changed method signatures so rather than return <code>void</code> they instead return <code>HColumnDescriptor</code>. This will throw`java.lang.NoSuchMethodError: org.apache.hadoop.hbase.HColumnDescriptor.setMaxVersions(I)V` so 0.92 and 0.94 are NOT compatible. You cannot do a rolling upgrade between them.</p> </div> </div> <div class="sect2"> <h3 id="upgrade0.92"><a class="anchor" href="#upgrade0.92"></a>12.8. Upgrading from 0.90.x to 0.92.x</h3> <div class="sect3"> <h4 id="_upgrade_guide"><a class="anchor" href="#_upgrade_guide"></a>12.8.1. Upgrade Guide</h4> <div class="paragraph"> <p>You will find that 0.92.0 runs a little differently to 0.90.x releases. Here are a few things to watch out for upgrading from 0.90.x to 0.92.0.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">tl:dr</div> <div class="paragraph"> <p>These are the important things to know before upgrading. . Once you upgrade, you can’t go back.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>MSLAB is on by default. Watch that heap usage if you have a lot of regions.</p> </li> <li> <p>Distributed Log Splitting is on by default. It should make RegionServer failover faster.</p> </li> <li> <p>There’s a separate tarball for security.</p> </li> <li> <p>If <code>-XX:MaxDirectMemorySize</code> is set in your <em>hbase-env.sh</em>, it’s going to enable the experimental off-heap cache (You may not want this).</p> </li> </ol> </div> </td> </tr> </table> </div> <div class="paragraph"> <div class="title">You can’t go back!</div> <p>To move to 0.92.0, all you need to do is shutdown your cluster, replace your HBase 0.90.x with HBase 0.92.0 binaries (be sure you clear out all 0.90.x instances) and restart (You cannot do a rolling restart from 0.90.x to 0.92.x — you must restart). On startup, the <code>.META.</code> table content is rewritten removing the table schema from the <code>info:regioninfo</code> column. Also, any flushes done post first startup will write out data in the new 0.92.0 file format, <a href="#hfilev2">HBase file format with inline blocks (version 2)</a>. This means you cannot go back to 0.90.x once you’ve started HBase 0.92.0 over your HBase data directory.</p> </div> <div class="paragraph"> <div class="title">MSLAB is ON by default</div> <p>In 0.92.0, the <code><a href="#hbase.hregion.memstore.mslab.enabled">hbase.hregion.memstore.mslab.enabled</a></code> flag is set to <code>true</code> (See <a href="#gcpause">Long GC pauses</a>). In 0.90.x it was false. When it is enabled, memstores will step allocate memory in MSLAB 2MB chunks even if the memstore has zero or just a few small elements. This is fine usually but if you had lots of regions per RegionServer in a 0.90.x cluster (and MSLAB was off), you may find yourself OOME’ing on upgrade because the <code>thousands of regions * number of column families * 2MB MSLAB</code> (at a minimum) puts your heap over the top. Set <code>hbase.hregion.memstore.mslab.enabled</code> to <code>false</code> or set the MSLAB size down from 2MB by setting <code>hbase.hregion.memstore.mslab.chunksize</code> to something less.</p> </div> <div id="dls" class="paragraph"> <div class="title">Distributed Log Splitting is on by default</div> <p>Previous, WAL logs on crash were split by the Master alone. In 0.92.0, log splitting is done by the cluster (See <a href="https://issues.apache.org/jira/browse/hbase-1364">HBASE-1364 [performance] Distributed splitting of regionserver commit logs</a> or see the blog post <a href="http://blog.cloudera.com/blog/2012/07/hbase-log-splitting/">Apache HBase Log Splitting</a>). This should cut down significantly on the amount of time it takes splitting logs and getting regions back online again.</p> </div> <div class="paragraph"> <div class="title">Memory accounting is different now</div> <p>In 0.92.0, <a href="#hfilev2">HBase file format with inline blocks (version 2)</a> indices and bloom filters take up residence in the same LRU used caching blocks that come from the filesystem. In 0.90.x, the HFile v1 indices lived outside of the LRU so they took up space even if the index was on a ‘cold’ file, one that wasn’t being actively used. With the indices now in the LRU, you may find you have less space for block caching. Adjust your block cache accordingly. See the <a href="#block.cache">Block Cache</a> for more detail. The block size default size has been changed in 0.92.0 from 0.2 (20 percent of heap) to 0.25.</p> </div> <div class="paragraph"> <div class="title">On the Hadoop version to use</div> <p>Run 0.92.0 on Hadoop 1.0.x (or CDH3u3). The performance benefits are worth making the move. Otherwise, our Hadoop prescription is as it has been; you need an Hadoop that supports a working sync. See <a href="#hadoop"><a href="http://hadoop.apache.org">Hadoop</a></a>.</p> </div> <div class="paragraph"> <p>If running on Hadoop 1.0.x (or CDH3u3), enable local read. See <a href="http://files.meetup.com/1350427/hug_ebay_jdcryans.pdf">Practical Caching</a> presentation for ruminations on the performance benefits ‘going local’ (and for how to enable local reads).</p> </div> <div class="paragraph"> <div class="title">HBase 0.92.0 ships with ZooKeeper 3.4.2</div> <p>If you can, upgrade your ZooKeeper. If you can’t, 3.4.2 clients should work against 3.3.X ensembles (HBase makes use of 3.4.2 API).</p> </div> <div class="paragraph"> <div class="title">Online alter is off by default</div> <p>In 0.92.0, we’ve added an experimental online schema alter facility (See <a href="#hbase.online.schema.update.enable">hbase.online.schema.update.enable</a>). It’s off by default. Enable it at your own risk. Online alter and splitting tables do not play well together so be sure your cluster quiescent using this feature (for now).</p> </div> <div class="paragraph"> <div class="title">WebUI</div> <p>The web UI has had a few additions made in 0.92.0. It now shows a list of the regions currently transitioning, recent compactions/flushes, and a process list of running processes (usually empty if all is well and requests are being handled promptly). Other additions including requests by region, a debugging servlet dump, etc.</p> </div> <div class="paragraph"> <div class="title">Security tarball</div> <p>We now ship with two tarballs; secure and insecure HBase. Documentation on how to setup a secure HBase is on the way.</p> </div> <div class="paragraph"> <div class="title">Changes in HBase replication</div> <p>0.92.0 adds two new features: multi-slave and multi-master replication. The way to enable this is the same as adding a new peer, so in order to have multi-master you would just run add_peer for each cluster that acts as a master to the other slave clusters. Collisions are handled at the timestamp level which may or may not be what you want, this needs to be evaluated on a per use case basis. Replication is still experimental in 0.92 and is disabled by default, run it at your own risk.</p> </div> <div class="paragraph"> <div class="title">RegionServer now aborts if OOME</div> <p>If an OOME, we now have the JVM kill -9 the RegionServer process so it goes down fast. Previous, a RegionServer might stick around after incurring an OOME limping along in some wounded state. To disable this facility, and recommend you leave it in place, you’d need to edit the bin/hbase file. Look for the addition of the -XX:OnOutOfMemoryError="kill -9 %p" arguments (See <a href="https://issues.apache.org/jira/browse/HBASE-4769">HBASE-4769 - ‘Abort RegionServer Immediately on OOME’</a>).</p> </div> <div class="paragraph"> <div class="title">HFile v2 and the “Bigger, Fewer” Tendency</div> <p>0.92.0 stores data in a new format, <a href="#hfilev2">HBase file format with inline blocks (version 2)</a>. As HBase runs, it will move all your data from HFile v1 to HFile v2 format. This auto-migration will run in the background as flushes and compactions run. HFile v2 allows HBase run with larger regions/files. In fact, we encourage that all HBasers going forward tend toward Facebook axiom #1, run with larger, fewer regions. If you have lots of regions now — more than 100s per host — you should look into setting your region size up after you move to 0.92.0 (In 0.92.0, default size is now 1G, up from 256M), and then running online merge tool (See <a href="https://issues.apache.org/jira/browse/HBASE-1621">HBASE-1621 merge tool should work on online cluster, but disabled table</a>).</p> </div> </div> </div> <div class="sect2"> <h3 id="upgrade0.90"><a class="anchor" href="#upgrade0.90"></a>12.9. Upgrading to HBase 0.90.x from 0.20.x or 0.89.x</h3> <div class="paragraph"> <p>This version of 0.90.x HBase can be started on data written by HBase 0.20.x or HBase 0.89.x. There is no need of a migration step. HBase 0.89.x and 0.90.x does write out the name of region directories differently — it names them with a md5 hash of the region name rather than a jenkins hash — so this means that once started, there is no going back to HBase 0.20.x.</p> </div> <div class="paragraph"> <p>Be sure to remove the <em>hbase-default.xml</em> from your <em>conf</em> directory on upgrade. A 0.20.x version of this file will have sub-optimal configurations for 0.90.x HBase. The <em>hbase-default.xml</em> file is now bundled into the HBase jar and read from there. If you would like to review the content of this file, see it in the src tree at <em>src/main/resources/hbase-default.xml</em> or see <a href="#hbase_default_configurations">HBase Default Configuration</a>.</p> </div> <div class="paragraph"> <p>Finally, if upgrading from 0.20.x, check your .META. schema in the shell. In the past we would recommend that users run with a 16kb MEMSTORE_FLUSHSIZE. Run</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> scan '-ROOT-'</pre> </div> </div> <div class="paragraph"> <p>in the shell. This will output the current <code>.META.</code> schema. Check <code>MEMSTORE_FLUSHSIZE</code> size. Is it 16kb (16384)? If so, you will need to change this (The 'normal'/default value is 64MB (67108864)). Run the script <code>bin/set_meta_memstore_size.rb</code>. This will make the necessary edit to your <code>.META.</code> schema. Failure to run this change will make for a slow cluster. See <a href="https://issues.apache.org/jira/browse/HBASE-3499">HBASE-3499 Users upgrading to 0.90.0 need to have their .META. table updated with the right MEMSTORE_SIZE</a>.</p> </div> </div> </div> </div> <h1 id="shell" class="sect0"><a class="anchor" href="#shell"></a>The Apache HBase Shell</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>The Apache HBase Shell is <a href="http://jruby.org">(J)Ruby</a>'s IRB with some HBase particular commands added. Anything you can do in IRB, you should be able to do in the HBase Shell.</p> </div> <div class="paragraph"> <p>To run the HBase shell, do as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./bin/hbase shell</code></pre> </div> </div> <div class="paragraph"> <p>Type <code>help</code> and then <code><RETURN></code> to see a listing of shell commands and options. Browse at least the paragraphs at the end of the help output for the gist of how variables and command arguments are entered into the HBase shell; in particular note how table names, rows, and columns, etc., must be quoted.</p> </div> <div class="paragraph"> <p>See <a href="#shell_exercises">shell exercises</a> for example basic shell operation.</p> </div> <div class="paragraph"> <p>Here is a nicely formatted listing of <a href="http://learnhbase.wordpress.com/2013/03/02/hbase-shell-commands/">all shell commands</a> by Rajeshbabu Chintaguntla.</p> </div> </div> </div> <div class="sect1"> <h2 id="scripting"><a class="anchor" href="#scripting"></a>13. Scripting with Ruby</h2> <div class="sectionbody"> <div class="paragraph"> <p>For examples scripting Apache HBase, look in the HBase <em>bin</em> directory. Look at the files that end in <em>*.rb</em>. To run one of these files, do as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./bin/hbase org.jruby.Main PATH_TO_SCRIPT</code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="_running_the_shell_in_non_interactive_mode"><a class="anchor" href="#_running_the_shell_in_non_interactive_mode"></a>14. Running the Shell in Non-Interactive Mode</h2> <div class="sectionbody"> <div class="paragraph"> <p>A new non-interactive mode has been added to the HBase Shell (<a href="https://issues.apache.org/jira/browse/HBASE-11658">HBASE-11658)</a>. Non-interactive mode captures the exit status (success or failure) of HBase Shell commands and passes that status back to the command interpreter. If you use the normal interactive mode, the HBase Shell will only ever return its own exit status, which will nearly always be <code>0</code> for success.</p> </div> <div class="paragraph"> <p>To invoke non-interactive mode, pass the <code>-n</code> or <code>--non-interactive</code> option to HBase Shell.</p> </div> </div> </div> <div class="sect1"> <h2 id="hbase.shell.noninteractive"><a class="anchor" href="#hbase.shell.noninteractive"></a>15. HBase Shell in OS Scripts</h2> <div class="sectionbody"> <div class="paragraph"> <p>You can use the HBase shell from within operating system script interpreters like the Bash shell which is the default command interpreter for most Linux and UNIX distributions. The following guidelines use Bash syntax, but could be adjusted to work with C-style shells such as csh or tcsh, and could probably be modified to work with the Microsoft Windows script interpreter as well. Submissions are welcome.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Spawning HBase Shell commands in this way is slow, so keep that in mind when you are deciding when combining HBase operations with the operating system command line is appropriate. </td> </tr> </table> </div> <div class="exampleblock"> <div class="title">Example 8. Passing Commands to the HBase Shell</div> <div class="content"> <div class="paragraph"> <p>You can pass commands to the HBase Shell in non-interactive mode (see <a href="#hbasee.shell.noninteractive">hbasee.shell.noninteractive</a>) using the <code>echo</code> command and the <code>|</code> (pipe) operator. Be sure to escape characters in the HBase commands which would otherwise be interpreted by the shell. Some debug-level output has been truncated from the example below.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ echo "describe 'test1'" | ./hbase shell -n Version 0.98.3-hadoop2, rd5e65a9144e315bb0a964e7730871af32f5018d5, Sat May 31 19:56:09 PDT 2014 describe 'test1' DESCRIPTION ENABLED 'test1', {NAME => 'cf', DATA_BLOCK_ENCODING => 'NON true E', BLOOMFILTER => 'ROW', REPLICATION_SCOPE => '0', VERSIONS => '1', COMPRESSION => 'NONE', MIN_VERSIO NS => '0', TTL => 'FOREVER', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false' , BLOCKCACHE => 'true'} 1 row(s) in 3.2410 seconds</code></pre> </div> </div> <div class="paragraph"> <p>To suppress all output, echo it to <em>/dev/null:</em></p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ echo "describe 'test'" | ./hbase shell -n > /dev/null 2>&1</code></pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 9. Checking the Result of a Scripted Command</div> <div class="content"> <div class="paragraph"> <p>Since scripts are not designed to be run interactively, you need a way to check whether your command failed or succeeded. The HBase shell uses the standard convention of returning a value of <code>0</code> for successful commands, and some non-zero value for failed commands. Bash stores a command’s return value in a special environment variable called <code>$?</code>. Because that variable is overwritten each time the shell runs any command, you should store the result in a different, script-defined variable.</p> </div> <div class="paragraph"> <p>This is a naive script that shows one way to store the return value and make a decision based upon it.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">#!/bin/bash echo "describe 'test'" | ./hbase shell -n > /dev/null 2>&1 status=$? echo "The status was " $status if ($status == 0); then echo "The command succeeded" else echo "The command may have failed." fi return $status</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="_checking_for_success_or_failure_in_scripts"><a class="anchor" href="#_checking_for_success_or_failure_in_scripts"></a>15.1. Checking for Success or Failure In Scripts</h3> <div class="paragraph"> <p>Getting an exit code of <code>0</code> means that the command you scripted definitely succeeded. However, getting a non-zero exit code does not necessarily mean the command failed. The command could have succeeded, but the client lost connectivity, or some other event obscured its success. This is because RPC commands are stateless. The only way to be sure of the status of an operation is to check. For instance, if your script creates a table, but returns a non-zero exit value, you should check whether the table was actually created before trying again to create it.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_read_hbase_shell_commands_from_a_command_file"><a class="anchor" href="#_read_hbase_shell_commands_from_a_command_file"></a>16. Read HBase Shell Commands from a Command File</h2> <div class="sectionbody"> <div class="paragraph"> <p>You can enter HBase Shell commands into a text file, one command per line, and pass that file to the HBase Shell.</p> </div> <div class="exampleblock"> <div class="title">Example 10. Example Command File</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>create 'test', 'cf' list 'test' put 'test', 'row1', 'cf:a', 'value1' put 'test', 'row2', 'cf:b', 'value2' put 'test', 'row3', 'cf:c', 'value3' put 'test', 'row4', 'cf:d', 'value4' scan 'test' get 'test', 'row1' disable 'test' enable 'test'</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 11. Directing HBase Shell to Execute the Commands</div> <div class="content"> <div class="paragraph"> <p>Pass the path to the command file as the only argument to the <code>hbase shell</code> command. Each command is executed and its output is shown. If you do not include the <code>exit</code> command in your script, you are returned to the HBase shell prompt. There is no way to programmatically check each individual command for success or failure. Also, though you see the output for each command, the commands themselves are not echoed to the screen so it can be difficult to line up the command with its output.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./hbase shell ./sample_commands.txt 0 row(s) in 3.4170 seconds TABLE test 1 row(s) in 0.0590 seconds 0 row(s) in 0.1540 seconds 0 row(s) in 0.0080 seconds 0 row(s) in 0.0060 seconds 0 row(s) in 0.0060 seconds ROW COLUMN+CELL row1 column=cf:a, timestamp=1407130286968, value=value1 row2 column=cf:b, timestamp=1407130286997, value=value2 row3 column=cf:c, timestamp=1407130287007, value=value3 row4 column=cf:d, timestamp=1407130287015, value=value4 4 row(s) in 0.0420 seconds COLUMN CELL cf:a timestamp=1407130286968, value=value1 1 row(s) in 0.0110 seconds 0 row(s) in 1.5630 seconds 0 row(s) in 0.4360 seconds</code></pre> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="_passing_vm_options_to_the_shell"><a class="anchor" href="#_passing_vm_options_to_the_shell"></a>17. Passing VM Options to the Shell</h2> <div class="sectionbody"> <div class="paragraph"> <p>You can pass VM options to the HBase Shell using the <code>HBASE_SHELL_OPTS</code> environment variable. You can set this in your environment, for instance by editing <em>~/.bashrc</em>, or set it as part of the command to launch HBase Shell. The following example sets several garbage-collection-related variables, just for the lifetime of the VM running the HBase Shell. The command should be run all on a single line, but is broken by the <code>\</code> character, for readability.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HBASE_SHELL_OPTS="-verbose:gc -XX:+PrintGCApplicationStoppedTime -XX:+PrintGCDateStamps \ -XX:+PrintGCDetails -Xloggc:$HBASE_HOME/logs/gc-hbase.log" ./bin/hbase shell</code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="_overriding_configuration_starting_the_hbase_shell"><a class="anchor" href="#_overriding_configuration_starting_the_hbase_shell"></a>18. Overriding configuration starting the HBase Shell</h2> <div class="sectionbody"> <div class="paragraph"> <p>As of hbase-2.0.5/hbase-2.1.3/hbase-2.2.0/hbase-1.4.10/hbase-1.5.0, you can pass or override hbase configuration as specified in <code>hbase-*.xml</code> by passing your key/values prefixed with <code>-D</code> on the command-line as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./bin/hbase shell -Dhbase.zookeeper.quorum=ZK0.remote.cluster.example.org,ZK1.remote.cluster.example.org,ZK2.remote.cluster.example.org -Draining=false ... hbase(main):001:0> @shell.hbase.configuration.get("hbase.zookeeper.quorum") => "ZK0.remote.cluster.example.org,ZK1.remote.cluster.example.org,ZK2.remote.cluster.example.org" hbase(main):002:0> @shell.hbase.configuration.get("raining") => "false"</code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="_shell_tricks"><a class="anchor" href="#_shell_tricks"></a>19. Shell Tricks</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_table_variables"><a class="anchor" href="#_table_variables"></a>19.1. Table variables</h3> <div class="paragraph"> <p>HBase 0.95 adds shell commands that provides jruby-style object-oriented references for tables. Previously all of the shell commands that act upon a table have a procedural style that always took the name of the table as an argument. HBase 0.95 introduces the ability to assign a table to a jruby variable. The table reference can be used to perform data read write operations such as puts, scans, and gets well as admin functionality such as disabling, dropping, describing tables.</p> </div> <div class="paragraph"> <p>For example, previously you would always specify a table name:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):000:0> create ‘t’, ‘f’ 0 row(s) in 1.0970 seconds hbase(main):001:0> put 't', 'rold', 'f', 'v' 0 row(s) in 0.0080 seconds hbase(main):002:0> scan 't' ROW COLUMN+CELL rold column=f:, timestamp=1378473207660, value=v 1 row(s) in 0.0130 seconds hbase(main):003:0> describe 't' DESCRIPTION ENABLED 't', {NAME => 'f', DATA_BLOCK_ENCODING => 'NONE', BLOOMFILTER => 'ROW', REPLICATION_ true SCOPE => '0', VERSIONS => '1', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => '2 147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false ', BLOCKCACHE => 'true'} 1 row(s) in 1.4430 seconds hbase(main):004:0> disable 't' 0 row(s) in 14.8700 seconds hbase(main):005:0> drop 't' 0 row(s) in 23.1670 seconds hbase(main):006:0></pre> </div> </div> <div class="paragraph"> <p>Now you can assign the table to a variable and use the results in jruby shell code.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):007 > t = create 't', 'f' 0 row(s) in 1.0970 seconds => Hbase::Table - t hbase(main):008 > t.put 'r', 'f', 'v' 0 row(s) in 0.0640 seconds hbase(main):009 > t.scan ROW COLUMN+CELL r column=f:, timestamp=1331865816290, value=v 1 row(s) in 0.0110 seconds hbase(main):010:0> t.describe DESCRIPTION ENABLED 't', {NAME => 'f', DATA_BLOCK_ENCODING => 'NONE', BLOOMFILTER => 'ROW', REPLICATION_ true SCOPE => '0', VERSIONS => '1', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => '2 147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false ', BLOCKCACHE => 'true'} 1 row(s) in 0.0210 seconds hbase(main):038:0> t.disable 0 row(s) in 6.2350 seconds hbase(main):039:0> t.drop 0 row(s) in 0.2340 seconds</pre> </div> </div> <div class="paragraph"> <p>If the table has already been created, you can assign a Table to a variable by using the get_table method:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):011 > create 't','f' 0 row(s) in 1.2500 seconds => Hbase::Table - t hbase(main):012:0> tab = get_table 't' 0 row(s) in 0.0010 seconds => Hbase::Table - t hbase(main):013:0> tab.put ‘r1’ ,’f’, ‘v’ 0 row(s) in 0.0100 seconds hbase(main):014:0> tab.scan ROW COLUMN+CELL r1 column=f:, timestamp=1378473876949, value=v 1 row(s) in 0.0240 seconds hbase(main):015:0></pre> </div> </div> <div class="paragraph"> <p>The list functionality has also been extended so that it returns a list of table names as strings. You can then use jruby to script table operations based on these names. The list_snapshots command also acts similarly.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):016 > tables = list(‘t.*’) TABLE t 1 row(s) in 0.1040 seconds => #<#<Class:0x7677ce29>:0x21d377a4> hbase(main):017:0> tables.map { |t| disable t ; drop t} 0 row(s) in 2.2510 seconds => [nil] hbase(main):018:0></pre> </div> </div> </div> <div class="sect2"> <h3 id="irbrc"><a class="anchor" href="#irbrc"></a>19.2. <em>irbrc</em></h3> <div class="paragraph"> <p>Create an <em>.irbrc</em> file for yourself in your home directory. Add customizations. A useful one is command history so commands are save across Shell invocations:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ more .irbrc require 'irb/ext/save-history' IRB.conf[:SAVE_HISTORY] = 100 IRB.conf[:HISTORY_FILE] = "#{ENV['HOME']}/.irb-save-history"</code></pre> </div> </div> <div class="paragraph"> <p>If you’d like to avoid printing the result of evaluting each expression to stderr, for example the array of tables returned from the "list" command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ echo "IRB.conf[:ECHO] = false" >>~/.irbrc</code></pre> </div> </div> <div class="paragraph"> <p>See the <code>ruby</code> documentation of <em>.irbrc</em> to learn about other possible configurations.</p> </div> </div> <div class="sect2"> <h3 id="_log_data_to_timestamp"><a class="anchor" href="#_log_data_to_timestamp"></a>19.3. LOG data to timestamp</h3> <div class="paragraph"> <p>To convert the date '08/08/16 20:56:29' from an hbase log into a timestamp, do:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):021:0> import java.text.SimpleDateFormat hbase(main):022:0> import java.text.ParsePosition hbase(main):023:0> SimpleDateFormat.new("yy/MM/dd HH:mm:ss").parse("08/08/16 20:56:29", ParsePosition.new(0)).getTime() => 1218920189000</pre> </div> </div> <div class="paragraph"> <p>To go the other direction:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):021:0> import java.util.Date hbase(main):022:0> Date.new(1218920189000).toString() => "Sat Aug 16 20:56:29 UTC 2008"</pre> </div> </div> <div class="paragraph"> <p>To output in a format that is exactly like that of the HBase log format will take a little messing with <a href="http://download.oracle.com/javase/6/docs/api/java/text/SimpleDateFormat.html">SimpleDateFormat</a>.</p> </div> </div> <div class="sect2"> <h3 id="_debug"><a class="anchor" href="#_debug"></a>19.4. Debug</h3> <div class="sect3"> <h4 id="_shell_debug_switch"><a class="anchor" href="#_shell_debug_switch"></a>19.4.1. Shell debug switch</h4> <div class="paragraph"> <p>You can set a debug switch in the shell to see more output — e.g. more of the stack trace on exception — when you run a command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase> debug <RETURN></code></pre> </div> </div> </div> <div class="sect3"> <h4 id="_debug_log_level"><a class="anchor" href="#_debug_log_level"></a>19.4.2. DEBUG log level</h4> <div class="paragraph"> <p>To enable DEBUG level logging in the shell, launch it with the <code>-d</code> option.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./bin/hbase shell -d</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="_commands"><a class="anchor" href="#_commands"></a>19.5. Commands</h3> <div class="sect3"> <h4 id="_count"><a class="anchor" href="#_count"></a>19.5.1. count</h4> <div class="paragraph"> <p>Count command returns the number of rows in a table. It’s quite fast when configured with the right CACHE</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase> count <span class="string"><span class="delimiter">'</span><span class="content"><tablename></span><span class="delimiter">'</span></span>, CACHE => <span class="integer">1000</span></code></pre> </div> </div> <div class="paragraph"> <p>The above count fetches 1000 rows at a time. Set CACHE lower if your rows are big. Default is to fetch one row at a time.</p> </div> </div> </div> </div> </div> <h1 id="datamodel" class="sect0"><a class="anchor" href="#datamodel"></a>Data Model</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>In HBase, data is stored in tables, which have rows and columns. This is a terminology overlap with relational databases (RDBMSs), but this is not a helpful analogy. Instead, it can be helpful to think of an HBase table as a multi-dimensional map.</p> </div> <div class="dlist"> <div class="title">HBase Data Model Terminology</div> <dl> <dt class="hdlist1">Table</dt> <dd> <p>An HBase table consists of multiple rows.</p> </dd> <dt class="hdlist1">Row</dt> <dd> <p>A row in HBase consists of a row key and one or more columns with values associated with them. Rows are sorted alphabetically by the row key as they are stored. For this reason, the design of the row key is very important. The goal is to store data in such a way that related rows are near each other. A common row key pattern is a website domain. If your row keys are domains, you should probably store them in reverse (org.apache.www, org.apache.mail, org.apache.jira). This way, all of the Apache domains are near each other in the table, rather than being spread out based on the first letter of the subdomain.</p> </dd> <dt class="hdlist1">Column</dt> <dd> <p>A column in HBase consists of a column family and a column qualifier, which are delimited by a <code>:</code> (colon) character.</p> </dd> <dt class="hdlist1">Column Family</dt> <dd> <p>Column families physically colocate a set of columns and their values, often for performance reasons. Each column family has a set of storage properties, such as whether its values should be cached in memory, how its data is compressed or its row keys are encoded, and others. Each row in a table has the same column families, though a given row might not store anything in a given column family.</p> </dd> <dt class="hdlist1">Column Qualifier</dt> <dd> <p>A column qualifier is added to a column family to provide the index for a given piece of data. Given a column family <code>content</code>, a column qualifier might be <code>content:html</code>, and another might be <code>content:pdf</code>. Though column families are fixed at table creation, column qualifiers are mutable and may differ greatly between rows.</p> </dd> <dt class="hdlist1">Cell</dt> <dd> <p>A cell is a combination of row, column family, and column qualifier, and contains a value and a timestamp, which represents the value’s version.</p> </dd> <dt class="hdlist1">Timestamp</dt> <dd> <p>A timestamp is written alongside each value, and is the identifier for a given version of a value. By default, the timestamp represents the time on the RegionServer when the data was written, but you can specify a different timestamp value when you put data into the cell.</p> </dd> </dl> </div> </div> </div> <div class="sect1"> <h2 id="conceptual.view"><a class="anchor" href="#conceptual.view"></a>20. Conceptual View</h2> <div class="sectionbody"> <div class="paragraph"> <p>You can read a very understandable explanation of the HBase data model in the blog post <a href="https://dzone.com/articles/understanding-hbase-and-bigtab">Understanding HBase and BigTable</a> by Jim R. Wilson. Another good explanation is available in the PDF <a href="http://0b4af6cdc2f0c5998459-c0245c5c937c5dedcca3f1764ecc9b2f.r43.cf2.rackcdn.com/9353-login1210_khurana.pdf">Introduction to Basic Schema Design</a> by Amandeep Khurana.</p> </div> <div class="paragraph"> <p>It may help to read different perspectives to get a solid understanding of HBase schema design. The linked articles cover the same ground as the information in this section.</p> </div> <div class="paragraph"> <p>The following example is a slightly modified form of the one on page 2 of the <a href="http://research.google.com/archive/bigtable.html">BigTable</a> paper. There is a table called <code>webtable</code> that contains two rows (<code>com.cnn.www</code> and <code>com.example.www</code>) and three column families named <code>contents</code>, <code>anchor</code>, and <code>people</code>. In this example, for the first row (<code>com.cnn.www</code>), <code>anchor</code> contains two columns (<code>anchor:cssnsi.com</code>, <code>anchor:my.look.ca</code>) and <code>contents</code> contains one column (<code>contents:html</code>). This example contains 5 versions of the row with the row key <code>com.cnn.www</code>, and one version of the row with the row key <code>com.example.www</code>. The <code>contents:html</code> column qualifier contains the entire HTML of a given website. Qualifiers of the <code>anchor</code> column family each contain the external site which links to the site represented by the row, along with the text it used in the anchor of its link. The <code>people</code> column family represents people associated with the site.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Column Names</div> <div class="paragraph"> <p>By convention, a column name is made of its column family prefix and a <em>qualifier</em>. For example, the column <em>contents:html</em> is made up of the column family <code>contents</code> and the <code>html</code> qualifier. The colon character (<code>:</code>) delimits the column family from the column family <em>qualifier</em>.</p> </div> </td> </tr> </table> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 4. Table <code>webtable</code></caption> <colgroup> <col style="width: 20%;"> <col style="width: 20%;"> <col style="width: 20%;"> <col style="width: 20%;"> <col style="width: 20%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Row Key</th> <th class="tableblock halign-left valign-top">Time Stamp</th> <th class="tableblock halign-left valign-top">ColumnFamily <code>contents</code></th> <th class="tableblock halign-left valign-top">ColumnFamily <code>anchor</code></th> <th class="tableblock halign-left valign-top">ColumnFamily <code>people</code></th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t9</p></td> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">anchor:cnnsi.com = "CNN"</p></td> <td class="tableblock halign-left valign-top"></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t8</p></td> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">anchor:my.look.ca = "CNN.com"</p></td> <td class="tableblock halign-left valign-top"></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t6</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">contents:html = "<html>…"</p></td> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t5</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">contents:html = "<html>…"</p></td> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t3</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">contents:html = "<html>…"</p></td> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"></td> </tr> </tbody> </table> <div class="paragraph"> <p>Cells in this table that appear to be empty do not take space, or in fact exist, in HBase. This is what makes HBase "sparse." A tabular view is not the only possible way to look at data in HBase, or even the most accurate. The following represents the same information as a multi-dimensional map. This is only a mock-up for illustrative purposes and may not be strictly accurate.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="json">{ <span class="key"><span class="delimiter">"</span><span class="content">com.cnn.www</span><span class="delimiter">"</span></span>: { <span class="error">c</span><span class="error">o</span><span class="error">n</span><span class="error">t</span><span class="error">e</span><span class="error">n</span><span class="error">t</span><span class="error">s</span>: { <span class="error">t</span><span class="integer">6</span>: <span class="error">c</span><span class="error">o</span><span class="error">n</span><span class="error">t</span><span class="error">e</span><span class="error">n</span><span class="error">t</span><span class="error">s</span>:<span class="error">h</span><span class="error">t</span><span class="error">m</span><span class="error">l</span>: <span class="string"><span class="delimiter">"</span><span class="content"><html>...</span><span class="delimiter">"</span></span> <span class="error">t</span><span class="integer">5</span>: <span class="error">c</span><span class="error">o</span><span class="error">n</span><span class="error">t</span><span class="error">e</span><span class="error">n</span><span class="error">t</span><span class="error">s</span>:<span class="error">h</span><span class="error">t</span><span class="error">m</span><span class="error">l</span>: <span class="string"><span class="delimiter">"</span><span class="content"><html>...</span><span class="delimiter">"</span></span> <span class="error">t</span><span class="integer">3</span>: <span class="error">c</span><span class="error">o</span><span class="error">n</span><span class="error">t</span><span class="error">e</span><span class="error">n</span><span class="error">t</span><span class="error">s</span>:<span class="error">h</span><span class="error">t</span><span class="error">m</span><span class="error">l</span>: <span class="string"><span class="delimiter">"</span><span class="content"><html>...</span><span class="delimiter">"</span></span> } <span class="error">a</span><span class="error">n</span><span class="error">c</span><span class="error">h</span><span class="error">o</span><span class="error">r</span>: { <span class="error">t</span><span class="integer">9</span>: <span class="error">a</span><span class="error">n</span><span class="error">c</span><span class="error">h</span><span class="error">o</span><span class="error">r</span>:<span class="error">c</span><span class="error">n</span><span class="error">n</span><span class="error">s</span><span class="error">i</span><span class="error">.</span><span class="error">c</span><span class="error">o</span><span class="error">m</span> <span class="error">=</span> <span class="string"><span class="delimiter">"</span><span class="content">CNN</span><span class="delimiter">"</span></span> <span class="error">t</span><span class="integer">8</span>: <span class="error">a</span><span class="error">n</span><span class="error">c</span><span class="error">h</span><span class="error">o</span><span class="error">r</span>:<span class="error">m</span><span class="error">y</span><span class="error">.</span><span class="error">l</span><span class="error">o</span><span class="error">o</span><span class="error">k</span><span class="error">.</span><span class="error">c</span><span class="error">a</span> <span class="error">=</span> <span class="string"><span class="delimiter">"</span><span class="content">CNN.com</span><span class="delimiter">"</span></span> } <span class="error">p</span><span class="error">e</span><span class="error">o</span><span class="error">p</span><span class="error">l</span><span class="error">e</span>: {} } <span class="key"><span class="delimiter">"</span><span class="content">com.example.www</span><span class="delimiter">"</span></span>: { <span class="error">c</span><span class="error">o</span><span class="error">n</span><span class="error">t</span><span class="error">e</span><span class="error">n</span><span class="error">t</span><span class="error">s</span>: { <span class="error">t</span><span class="integer">5</span>: <span class="error">c</span><span class="error">o</span><span class="error">n</span><span class="error">t</span><span class="error">e</span><span class="error">n</span><span class="error">t</span><span class="error">s</span>:<span class="error">h</span><span class="error">t</span><span class="error">m</span><span class="error">l</span>: <span class="string"><span class="delimiter">"</span><span class="content"><html>...</span><span class="delimiter">"</span></span> } <span class="error">a</span><span class="error">n</span><span class="error">c</span><span class="error">h</span><span class="error">o</span><span class="error">r</span>: {} <span class="error">p</span><span class="error">e</span><span class="error">o</span><span class="error">p</span><span class="error">l</span><span class="error">e</span>: { <span class="error">t</span><span class="integer">5</span>: <span class="error">p</span><span class="error">e</span><span class="error">o</span><span class="error">p</span><span class="error">l</span><span class="error">e</span>:<span class="error">a</span><span class="error">u</span><span class="error">t</span><span class="error">h</span><span class="error">o</span><span class="error">r</span>: <span class="string"><span class="delimiter">"</span><span class="content">John Doe</span><span class="delimiter">"</span></span> } } }</code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="physical.view"><a class="anchor" href="#physical.view"></a>21. Physical View</h2> <div class="sectionbody"> <div class="paragraph"> <p>Although at a conceptual level tables may be viewed as a sparse set of rows, they are physically stored by column family. A new column qualifier (column_family:column_qualifier) can be added to an existing column family at any time.</p> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 5. ColumnFamily <code>anchor</code></caption> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Row Key</th> <th class="tableblock halign-left valign-top">Time Stamp</th> <th class="tableblock halign-left valign-top">Column Family <code>anchor</code></th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t9</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>anchor:cnnsi.com = "CNN"</code></p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t8</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>anchor:my.look.ca = "CNN.com"</code></p></td> </tr> </tbody> </table> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 6. ColumnFamily <code>contents</code></caption> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Row Key</th> <th class="tableblock halign-left valign-top">Time Stamp</th> <th class="tableblock halign-left valign-top">ColumnFamily <code>contents:</code></th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t6</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">contents:html = "<html>…"</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t5</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">contents:html = "<html>…"</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">"com.cnn.www"</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">t3</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">contents:html = "<html>…"</p></td> </tr> </tbody> </table> <div class="paragraph"> <p>The empty cells shown in the conceptual view are not stored at all. Thus a request for the value of the <code>contents:html</code> column at time stamp <code>t8</code> would return no value. Similarly, a request for an <code>anchor:my.look.ca</code> value at time stamp <code>t9</code> would return no value. However, if no timestamp is supplied, the most recent value for a particular column would be returned. Given multiple versions, the most recent is also the first one found, since timestamps are stored in descending order. Thus a request for the values of all columns in the row <code>com.cnn.www</code> if no timestamp is specified would be: the value of <code>contents:html</code> from timestamp <code>t6</code>, the value of <code>anchor:cnnsi.com</code> from timestamp <code>t9</code>, the value of <code>anchor:my.look.ca</code> from timestamp <code>t8</code>.</p> </div> <div class="paragraph"> <p>For more information about the internals of how Apache HBase stores data, see <a href="#regions.arch">regions.arch</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_namespace"><a class="anchor" href="#_namespace"></a>22. Namespace</h2> <div class="sectionbody"> <div class="paragraph"> <p>A namespace is a logical grouping of tables analogous to a database in relation database systems. This abstraction lays the groundwork for upcoming multi-tenancy related features:</p> </div> <div class="ulist"> <ul> <li> <p>Quota Management (<a href="https://issues.apache.org/jira/browse/HBASE-8410">HBASE-8410</a>) - Restrict the amount of resources (ie regions, tables) a namespace can consume.</p> </li> <li> <p>Namespace Security Administration (<a href="https://issues.apache.org/jira/browse/HBASE-9206">HBASE-9206</a>) - Provide another level of security administration for tenants.</p> </li> <li> <p>Region server groups (<a href="https://issues.apache.org/jira/browse/HBASE-6721">HBASE-6721</a>) - A namespace/table can be pinned onto a subset of RegionServers thus guaranteeing a course level of isolation.</p> </li> </ul> </div> <div class="sect2"> <h3 id="namespace_creation"><a class="anchor" href="#namespace_creation"></a>22.1. Namespace management</h3> <div class="paragraph"> <p>A namespace can be created, removed or altered. Namespace membership is determined during table creation by specifying a fully-qualified table name of the form:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><table</span> <span class="attribute-name">namespace</span><span class="tag">></span>:<span class="tag"><table</span> <span class="attribute-name">qualifier</span><span class="tag">></span></code></pre> </div> </div> <div class="exampleblock"> <div class="title">Example 12. Examples</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">#Create a namespace create_namespace 'my_ns'</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">#create my_table in my_ns namespace create 'my_ns:my_table', 'fam'</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">#drop namespace drop_namespace 'my_ns'</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">#alter namespace alter_namespace 'my_ns', {METHOD => 'set', 'PROPERTY_NAME' => 'PROPERTY_VALUE'}</code></pre> </div> </div> </div> </div> </div> <div class="sect2"> <h3 id="namespace_special"><a class="anchor" href="#namespace_special"></a>22.2. Predefined namespaces</h3> <div class="paragraph"> <p>There are two predefined special namespaces:</p> </div> <div class="ulist"> <ul> <li> <p>hbase - system namespace, used to contain HBase internal tables</p> </li> <li> <p>default - tables with no explicit specified namespace will automatically fall into this namespace</p> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 13. Examples</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">#namespace=foo and table qualifier=bar create 'foo:bar', 'fam' #namespace=default and table qualifier=bar create 'bar', 'fam'</code></pre> </div> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="_table"><a class="anchor" href="#_table"></a>23. Table</h2> <div class="sectionbody"> <div class="paragraph"> <p>Tables are declared up front at schema definition time.</p> </div> </div> </div> <div class="sect1"> <h2 id="_row"><a class="anchor" href="#_row"></a>24. Row</h2> <div class="sectionbody"> <div class="paragraph"> <p>Row keys are uninterpreted bytes. Rows are lexicographically sorted with the lowest order appearing first in a table. The empty byte array is used to denote both the start and end of a tables' namespace.</p> </div> </div> </div> <div class="sect1"> <h2 id="columnfamily"><a class="anchor" href="#columnfamily"></a>25. Column Family</h2> <div class="sectionbody"> <div class="paragraph"> <p>Columns in Apache HBase are grouped into <em>column families</em>. All column members of a column family have the same prefix. For example, the columns <em>courses:history</em> and <em>courses:math</em> are both members of the <em>courses</em> column family. The colon character (<code>:</code>) delimits the column family from the column family qualifier. The column family prefix must be composed of <em>printable</em> characters. The qualifying tail, the column family <em>qualifier</em>, can be made of any arbitrary bytes. Column families must be declared up front at schema definition time whereas columns do not need to be defined at schema time but can be conjured on the fly while the table is up an running.</p> </div> <div class="paragraph"> <p>Physically, all column family members are stored together on the filesystem. Because tunings and storage specifications are done at the column family level, it is advised that all column family members have the same general access pattern and size characteristics.</p> </div> </div> </div> <div class="sect1"> <h2 id="_cells"><a class="anchor" href="#_cells"></a>26. Cells</h2> <div class="sectionbody"> <div class="paragraph"> <p>A <em>{row, column, version}</em> tuple exactly specifies a <code>cell</code> in HBase. Cell content is uninterpreted bytes</p> </div> </div> </div> <div class="sect1"> <h2 id="_data_model_operations"><a class="anchor" href="#_data_model_operations"></a>27. Data Model Operations</h2> <div class="sectionbody"> <div class="paragraph"> <p>The four primary data model operations are Get, Put, Scan, and Delete. Operations are applied via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html">Table</a> instances.</p> </div> <div class="sect2"> <h3 id="_get"><a class="anchor" href="#_get"></a>27.1. Get</h3> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</a> returns attributes for a specified row. Gets are executed via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#get(org.apache.hadoop.hbase.client.Get)">Table.get</a>.</p> </div> </div> <div class="sect2"> <h3 id="_put"><a class="anchor" href="#_put"></a>27.2. Put</h3> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Put.html">Put</a> either adds new rows to a table (if the key is new) or can update existing rows (if the key already exists). Puts are executed via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#put(org.apache.hadoop.hbase.client.Put)">Table.put</a> (writeBuffer) or link:http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#batch(java.util.List, java.lang.Object[])[Table.batch] (non-writeBuffer).</p> </div> </div> <div class="sect2"> <h3 id="scan"><a class="anchor" href="#scan"></a>27.3. Scans</h3> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</a> allow iteration over multiple rows for specified attributes.</p> </div> <div class="paragraph"> <p>The following is an example of a Scan on a Table instance. Assume that a table is populated with rows with keys "row1", "row2", "row3", and then another set of rows with the keys "abc1", "abc2", and "abc3". The following example shows how to set a Scan instance to return the rows beginning with "row".</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR = <span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>.getBytes(); ... Table table = ... <span class="comment">// instantiate a Table instance</span> Scan scan = <span class="keyword">new</span> Scan(); scan.addColumn(CF, ATTR); scan.setRowPrefixFilter(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">row</span><span class="delimiter">"</span></span>)); ResultScanner rs = table.getScanner(scan); <span class="keyword">try</span> { <span class="keyword">for</span> (<span class="predefined-type">Result</span> r = rs.next(); r != <span class="predefined-constant">null</span>; r = rs.next()) { <span class="comment">// process result...</span> } } <span class="keyword">finally</span> { rs.close(); <span class="comment">// always close the ResultScanner!</span> }</code></pre> </div> </div> <div class="paragraph"> <p>Note that generally the easiest way to specify a specific stop point for a scan is by using the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/InclusiveStopFilter.html">InclusiveStopFilter</a> class.</p> </div> </div> <div class="sect2"> <h3 id="_delete"><a class="anchor" href="#_delete"></a>27.4. Delete</h3> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Delete.html">Delete</a> removes a row from a table. Deletes are executed via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#delete(org.apache.hadoop.hbase.client.Delete)">Table.delete</a>.</p> </div> <div class="paragraph"> <p>HBase does not modify data in place, and so deletes are handled by creating new markers called <em>tombstones</em>. These tombstones, along with the dead values, are cleaned up on major compactions.</p> </div> <div class="paragraph"> <p>See <a href="#version.delete">version.delete</a> for more information on deleting versions of columns, and see <a href="#compaction">compaction</a> for more information on compactions.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="versions"><a class="anchor" href="#versions"></a>28. Versions</h2> <div class="sectionbody"> <div class="paragraph"> <p>A <em>{row, column, version}</em> tuple exactly specifies a <code>cell</code> in HBase. It’s possible to have an unbounded number of cells where the row and column are the same but the cell address differs only in its version dimension.</p> </div> <div class="paragraph"> <p>While rows and column keys are expressed as bytes, the version is specified using a long integer. Typically this long contains time instances such as those returned by <code>java.util.Date.getTime()</code> or <code>System.currentTimeMillis()</code>, that is: <em class="quote">the difference, measured in milliseconds, between the current time and midnight, January 1, 1970 UTC</em>.</p> </div> <div class="paragraph"> <p>The HBase version dimension is stored in decreasing order, so that when reading from a store file, the most recent values are found first.</p> </div> <div class="paragraph"> <p>There is a lot of confusion over the semantics of <code>cell</code> versions, in HBase. In particular:</p> </div> <div class="ulist"> <ul> <li> <p>If multiple writes to a cell have the same version, only the last written is fetchable.</p> </li> <li> <p>It is OK to write cells in a non-increasing version order.</p> </li> </ul> </div> <div class="paragraph"> <p>Below we describe how the version dimension in HBase currently works. See <a href="https://issues.apache.org/jira/browse/HBASE-2406">HBASE-2406</a> for discussion of HBase versions. <a href="http://outerthought.org/blog/417-ot.html">Bending time in HBase</a> makes for a good read on the version, or time, dimension in HBase. It has more detail on versioning than is provided here. As of this writing, the limitation <em>Overwriting values at existing timestamps</em> mentioned in the article no longer holds in HBase. This section is basically a synopsis of this article by Bruno Dumon.</p> </div> <div class="sect2"> <h3 id="specify.number.of.versions"><a class="anchor" href="#specify.number.of.versions"></a>28.1. Specifying the Number of Versions to Store</h3> <div class="paragraph"> <p>The maximum number of versions to store for a given column is part of the column schema and is specified at table creation, or via an <code>alter</code> command, via <code>HColumnDescriptor.DEFAULT_VERSIONS</code>. Prior to HBase 0.96, the default number of versions kept was <code>3</code>, but in 0.96 and newer has been changed to <code>1</code>.</p> </div> <div class="exampleblock"> <div class="title">Example 14. Modify the Maximum Number of Versions for a Column Family</div> <div class="content"> <div class="paragraph"> <p>This example uses HBase Shell to keep a maximum of 5 versions of all columns in column family <code>f1</code>. You could also use <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a>.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> alter ‘t1′, NAME => ‘f1′, VERSIONS => 5</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 15. Modify the Minimum Number of Versions for a Column Family</div> <div class="content"> <div class="paragraph"> <p>You can also specify the minimum number of versions to store per column family. By default, this is set to 0, which means the feature is disabled. The following example sets the minimum number of versions on all columns in column family <code>f1</code> to <code>2</code>, via HBase Shell. You could also use <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a>.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> alter ‘t1′, NAME => ‘f1′, MIN_VERSIONS => 2</pre> </div> </div> </div> </div> <div class="paragraph"> <p>Starting with HBase 0.98.2, you can specify a global default for the maximum number of versions kept for all newly-created columns, by setting <code>hbase.column.max.version</code> in <em>hbase-site.xml</em>. See <a href="#hbase.column.max.version">hbase.column.max.version</a>.</p> </div> </div> <div class="sect2"> <h3 id="versions.ops"><a class="anchor" href="#versions.ops"></a>28.2. Versions and HBase Operations</h3> <div class="paragraph"> <p>In this section we look at the behavior of the version dimension for each of the core HBase operations.</p> </div> <div class="sect3"> <h4 id="_get_scan"><a class="anchor" href="#_get_scan"></a>28.2.1. Get/Scan</h4> <div class="paragraph"> <p>Gets are implemented on top of Scans. The below discussion of <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</a> applies equally to <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scans</a>.</p> </div> <div class="paragraph"> <p>By default, i.e. if you specify no explicit version, when doing a <code>get</code>, the cell whose version has the largest value is returned (which may or may not be the latest one written, see later). The default behavior can be modified in the following ways:</p> </div> <div class="ulist"> <ul> <li> <p>to return more than one version, see <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html#setMaxVersions()">Get.setMaxVersions()</a></p> </li> <li> <p>to return versions other than the latest, see <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html#setTimeRange(long,%20long)">Get.setTimeRange()</a></p> <div class="paragraph"> <p>To retrieve the latest version that is less than or equal to a given value, thus giving the 'latest' state of the record at a certain point in time, just use a range from 0 to the desired version and set the max versions to 1.</p> </div> </li> </ul> </div> </div> <div class="sect3"> <h4 id="_default_get_example"><a class="anchor" href="#_default_get_example"></a>28.2.2. Default Get Example</h4> <div class="paragraph"> <p>The following Get will only retrieve the current version of the row</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR = <span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>.getBytes(); ... Get get = <span class="keyword">new</span> Get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">row1</span><span class="delimiter">"</span></span>)); <span class="predefined-type">Result</span> r = table.get(get); <span class="type">byte</span><span class="type">[]</span> b = r.getValue(CF, ATTR); <span class="comment">// returns current version of value</span></code></pre> </div> </div> </div> <div class="sect3"> <h4 id="_versioned_get_example"><a class="anchor" href="#_versioned_get_example"></a>28.2.3. Versioned Get Example</h4> <div class="paragraph"> <p>The following Get will return the last 3 versions of the row.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR = <span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>.getBytes(); ... Get get = <span class="keyword">new</span> Get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">row1</span><span class="delimiter">"</span></span>)); get.setMaxVersions(<span class="integer">3</span>); <span class="comment">// will return last 3 versions of row</span> <span class="predefined-type">Result</span> r = table.get(get); <span class="type">byte</span><span class="type">[]</span> b = r.getValue(CF, ATTR); <span class="comment">// returns current version of value</span> <span class="predefined-type">List</span><KeyValue> kv = r.getColumn(CF, ATTR); <span class="comment">// returns all versions of this column</span></code></pre> </div> </div> </div> <div class="sect3"> <h4 id="_put_2"><a class="anchor" href="#_put_2"></a>28.2.4. Put</h4> <div class="paragraph"> <p>Doing a put always creates a new version of a <code>cell</code>, at a certain timestamp. By default the system uses the server’s <code>currentTimeMillis</code>, but you can specify the version (= the long integer) yourself, on a per-column level. This means you could assign a time in the past or the future, or use the long value for non-time purposes.</p> </div> <div class="paragraph"> <p>To overwrite an existing value, do a put at exactly the same row, column, and version as that of the cell you want to overwrite.</p> </div> <div class="sect4"> <h5 id="_implicit_version_example"><a class="anchor" href="#_implicit_version_example"></a>Implicit Version Example</h5> <div class="paragraph"> <p>The following Put will be implicitly versioned by HBase with the current time.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR = <span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>.getBytes(); ... Put put = <span class="keyword">new</span> Put(Bytes.toBytes(row)); put.add(CF, ATTR, Bytes.toBytes( data)); table.put(put);</code></pre> </div> </div> </div> <div class="sect4"> <h5 id="_explicit_version_example"><a class="anchor" href="#_explicit_version_example"></a>Explicit Version Example</h5> <div class="paragraph"> <p>The following Put has the version timestamp explicitly set.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR = <span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>.getBytes(); ... Put put = <span class="keyword">new</span> Put( Bytes.toBytes(row)); <span class="type">long</span> explicitTimeInMs = <span class="integer">555</span>; <span class="comment">// just an example</span> put.add(CF, ATTR, explicitTimeInMs, Bytes.toBytes(data)); table.put(put);</code></pre> </div> </div> <div class="paragraph"> <p>Caution: the version timestamp is used internally by HBase for things like time-to-live calculations. It’s usually best to avoid setting this timestamp yourself. Prefer using a separate timestamp attribute of the row, or have the timestamp as a part of the row key, or both.</p> </div> </div> </div> <div class="sect3"> <h4 id="version.delete"><a class="anchor" href="#version.delete"></a>28.2.5. Delete</h4> <div class="paragraph"> <p>There are three different types of internal delete markers. See Lars Hofhansl’s blog for discussion of his attempt adding another, <a href="http://hadoop-hbase.blogspot.com/2012/01/scanning-in-hbase.html">Scanning in HBase: Prefix Delete Marker</a>.</p> </div> <div class="ulist"> <ul> <li> <p>Delete: for a specific version of a column.</p> </li> <li> <p>Delete column: for all versions of a column.</p> </li> <li> <p>Delete family: for all columns of a particular ColumnFamily</p> </li> </ul> </div> <div class="paragraph"> <p>When deleting an entire row, HBase will internally create a tombstone for each ColumnFamily (i.e., not each individual column).</p> </div> <div class="paragraph"> <p>Deletes work by creating <em>tombstone</em> markers. For example, let’s suppose we want to delete a row. For this you can specify a version, or else by default the <code>currentTimeMillis</code> is used. What this means is <em>delete all cells where the version is less than or equal to this version</em>. HBase never modifies data in place, so for example a delete will not immediately delete (or mark as deleted) the entries in the storage file that correspond to the delete condition. Rather, a so-called <em>tombstone</em> is written, which will mask the deleted values. When HBase does a major compaction, the tombstones are processed to actually remove the dead values, together with the tombstones themselves. If the version you specified when deleting a row is larger than the version of any value in the row, then you can consider the complete row to be deleted.</p> </div> <div class="paragraph"> <p>For an informative discussion on how deletes and versioning interact, see the thread <a href="http://comments.gmane.org/gmane.comp.java.hadoop.hbase.user/28421">Put w/timestamp → Deleteall → Put w/ timestamp fails</a> up on the user mailing list.</p> </div> <div class="paragraph"> <p>Also see <a href="#keyvalue">keyvalue</a> for more information on the internal KeyValue format.</p> </div> <div class="paragraph"> <p>Delete markers are purged during the next major compaction of the store, unless the <code>KEEP_DELETED_CELLS</code> option is set in the column family (See <a href="#cf.keep.deleted">Keeping Deleted Cells</a>). To keep the deletes for a configurable amount of time, you can set the delete TTL via the hbase.hstore.time.to.purge.deletes property in <em>hbase-site.xml</em>. If <code>hbase.hstore.time.to.purge.deletes</code> is not set, or set to 0, all delete markers, including those with timestamps in the future, are purged during the next major compaction. Otherwise, a delete marker with a timestamp in the future is kept until the major compaction which occurs after the time represented by the marker’s timestamp plus the value of <code>hbase.hstore.time.to.purge.deletes</code>, in milliseconds.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> This behavior represents a fix for an unexpected change that was introduced in HBase 0.94, and was fixed in <a href="https://issues.apache.org/jira/browse/HBASE-10118">HBASE-10118</a>. The change has been backported to HBase 0.94 and newer branches. </td> </tr> </table> </div> </div> </div> <div class="sect2"> <h3 id="_current_limitations"><a class="anchor" href="#_current_limitations"></a>28.3. Current Limitations</h3> <div class="sect3"> <h4 id="_deletes_mask_puts"><a class="anchor" href="#_deletes_mask_puts"></a>28.3.1. Deletes mask Puts</h4> <div class="paragraph"> <p>Deletes mask puts, even puts that happened after the delete was entered. See <a href="https://issues.apache.org/jira/browse/HBASE-2256">HBASE-2256</a>. Remember that a delete writes a tombstone, which only disappears after then next major compaction has run. Suppose you do a delete of everything ⇐ T. After this you do a new put with a timestamp ⇐ T. This put, even if it happened after the delete, will be masked by the delete tombstone. Performing the put will not fail, but when you do a get you will notice the put did have no effect. It will start working again after the major compaction has run. These issues should not be a problem if you use always-increasing versions for new puts to a row. But they can occur even if you do not care about time: just do delete and put immediately after each other, and there is some chance they happen within the same millisecond.</p> </div> </div> <div class="sect3"> <h4 id="major.compactions.change.query.results"><a class="anchor" href="#major.compactions.change.query.results"></a>28.3.2. Major compactions change query results</h4> <div class="paragraph"> <p><em>…create three cell versions at t1, t2 and t3, with a maximum-versions setting of 2. So when getting all versions, only the values at t2 and t3 will be returned. But if you delete the version at t2 or t3, the one at t1 will appear again. Obviously, once a major compaction has run, such behavior will not be the case anymore…</em> (See <em>Garbage Collection</em> in <a href="http://outerthought.org/blog/417-ot.html">Bending time in HBase</a>.)</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="dm.sort"><a class="anchor" href="#dm.sort"></a>29. Sort Order</h2> <div class="sectionbody"> <div class="paragraph"> <p>All data model operations HBase return data in sorted order. First by row, then by ColumnFamily, followed by column qualifier, and finally timestamp (sorted in reverse, so newest records are returned first).</p> </div> </div> </div> <div class="sect1"> <h2 id="dm.column.metadata"><a class="anchor" href="#dm.column.metadata"></a>30. Column Metadata</h2> <div class="sectionbody"> <div class="paragraph"> <p>There is no store of column metadata outside of the internal KeyValue instances for a ColumnFamily. Thus, while HBase can support not only a wide number of columns per row, but a heterogeneous set of columns between rows as well, it is your responsibility to keep track of the column names.</p> </div> <div class="paragraph"> <p>The only way to get a complete set of columns that exist for a ColumnFamily is to process all the rows. For more information about how HBase stores data internally, see <a href="#keyvalue">keyvalue</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_joins"><a class="anchor" href="#_joins"></a>31. Joins</h2> <div class="sectionbody"> <div class="paragraph"> <p>Whether HBase supports joins is a common question on the dist-list, and there is a simple answer: it doesn’t, at not least in the way that RDBMS' support them (e.g., with equi-joins or outer-joins in SQL). As has been illustrated in this chapter, the read data model operations in HBase are Get and Scan.</p> </div> <div class="paragraph"> <p>However, that doesn’t mean that equivalent join functionality can’t be supported in your application, but you have to do it yourself. The two primary strategies are either denormalizing the data upon writing to HBase, or to have lookup tables and do the join between HBase tables in your application or MapReduce code (and as RDBMS' demonstrate, there are several strategies for this depending on the size of the tables, e.g., nested loops vs. hash-joins). So which is the best approach? It depends on what you are trying to do, and as such there isn’t a single answer that works for every use case.</p> </div> </div> </div> <div class="sect1"> <h2 id="_acid"><a class="anchor" href="#_acid"></a>32. ACID</h2> <div class="sectionbody"> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/acid-semantics.html">ACID Semantics</a>. Lars Hofhansl has also written a note on <a href="http://hadoop-hbase.blogspot.com/2012/03/acid-in-hbase.html">ACID in HBase</a>.</p> </div> </div> </div> <h1 id="schema" class="sect0"><a class="anchor" href="#schema"></a>HBase and Schema Design</h1> <div class="openblock partintro"> <div class="content"> A good general introduction on the strength and weaknesses modelling on the various non-rdbms datastores is Ian Varley’s Master thesis, <a href="http://ianvarley.com/UT/MR/Varley_MastersReport_Full_2009-08-07.pdf">No Relation: The Mixed Blessings of Non-Relational Databases</a>. Also, read <a href="#keyvalue">keyvalue</a> for how HBase stores data internally, and the section on <a href="#schema.casestudies">schema.casestudies</a>. </div> </div> <div class="sect1"> <h2 id="schema.creation"><a class="anchor" href="#schema.creation"></a>33. Schema Creation</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase schemas can be created or updated using the <a href="#shell">The Apache HBase Shell</a> or by using <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Admin.html">Admin</a> in the Java API.</p> </div> <div class="paragraph"> <p>Tables must be disabled when making ColumnFamily modifications, for example:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); Admin admin = <span class="keyword">new</span> Admin(conf); <span class="predefined-type">String</span> table = <span class="string"><span class="delimiter">"</span><span class="content">myTable</span><span class="delimiter">"</span></span>; admin.disableTable(table); HColumnDescriptor cf1 = ...; admin.addColumn(table, cf1); <span class="comment">// adding new ColumnFamily</span> HColumnDescriptor cf2 = ...; admin.modifyColumn(table, cf2); <span class="comment">// modifying existing ColumnFamily</span> admin.enableTable(table);</code></pre> </div> </div> <div class="paragraph"> <p>See <a href="#client_dependencies">client dependencies</a> for more information about configuring client connections.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> online schema changes are supported in the 0.92.x codebase, but the 0.90.x codebase requires the table to be disabled. </td> </tr> </table> </div> <div class="sect2"> <h3 id="schema.updates"><a class="anchor" href="#schema.updates"></a>33.1. Schema Updates</h3> <div class="paragraph"> <p>When changes are made to either Tables or ColumnFamilies (e.g. region size, block size), these changes take effect the next time there is a major compaction and the StoreFiles get re-written.</p> </div> <div class="paragraph"> <p>See <a href="#store">store</a> for more information on StoreFiles.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="number.of.cfs"><a class="anchor" href="#number.of.cfs"></a>34. On the number of column families</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase currently does not do well with anything above two or three column families so keep the number of column families in your schema low. Currently, flushing and compactions are done on a per Region basis so if one column family is carrying the bulk of the data bringing on flushes, the adjacent families will also be flushed even though the amount of data they carry is small. When many column families exist the flushing and compaction interaction can make for a bunch of needless i/o (To be addressed by changing flushing and compaction to work on a per column family basis). For more information on compactions, see <a href="#compaction">[compaction]</a>.</p> </div> <div class="paragraph"> <p>Try to make do with one column family if you can in your schemas. Only introduce a second and third column family in the case where data access is usually column scoped; i.e. you query one column family or the other but usually not both at the one time.</p> </div> <div class="sect2"> <h3 id="number.of.cfs.card"><a class="anchor" href="#number.of.cfs.card"></a>34.1. Cardinality of ColumnFamilies</h3> <div class="paragraph"> <p>Where multiple ColumnFamilies exist in a single table, be aware of the cardinality (i.e., number of rows). If ColumnFamilyA has 1 million rows and ColumnFamilyB has 1 billion rows, ColumnFamilyA’s data will likely be spread across many, many regions (and RegionServers). This makes mass scans for ColumnFamilyA less efficient.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="rowkey.design"><a class="anchor" href="#rowkey.design"></a>35. Rowkey Design</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_hotspotting"><a class="anchor" href="#_hotspotting"></a>35.1. Hotspotting</h3> <div class="paragraph"> <p>Rows in HBase are sorted lexicographically by row key. This design optimizes for scans, allowing you to store related rows, or rows that will be read together, near each other. However, poorly designed row keys are a common source of <em class="firstterm">hotspotting</em>. Hotspotting occurs when a large amount of client traffic is directed at one node, or only a few nodes, of a cluster. This traffic may represent reads, writes, or other operations. The traffic overwhelms the single machine responsible for hosting that region, causing performance degradation and potentially leading to region unavailability. This can also have adverse effects on other regions hosted by the same region server as that host is unable to service the requested load. It is important to design data access patterns such that the cluster is fully and evenly utilized.</p> </div> <div class="paragraph"> <p>To prevent hotspotting on writes, design your row keys such that rows that truly do need to be in the same region are, but in the bigger picture, data is being written to multiple regions across the cluster, rather than one at a time. Some common techniques for avoiding hotspotting are described below, along with some of their advantages and drawbacks.</p> </div> <div class="paragraph"> <div class="title">Salting</div> <p>Salting in this sense has nothing to do with cryptography, but refers to adding random data to the start of a row key. In this case, salting refers to adding a randomly-assigned prefix to the row key to cause it to sort differently than it otherwise would. The number of possible prefixes correspond to the number of regions you want to spread the data across. Salting can be helpful if you have a few "hot" row key patterns which come up over and over amongst other more evenly-distributed rows. Consider the following example, which shows that salting can spread write load across multiple RegionServers, and illustrates some of the negative implications for reads.</p> </div> <div class="exampleblock"> <div class="title">Example 16. Salting Example</div> <div class="content"> <div class="paragraph"> <p>Suppose you have the following list of row keys, and your table is split such that there is one region for each letter of the alphabet. Prefix 'a' is one region, prefix 'b' is another. In this table, all rows starting with 'f' are in the same region. This example focuses on rows with keys like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>foo0001 foo0002 foo0003 foo0004</pre> </div> </div> <div class="paragraph"> <p>Now, imagine that you would like to spread these across four different regions. You decide to use four different salts: <code>a</code>, <code>b</code>, <code>c</code>, and <code>d</code>. In this scenario, each of these letter prefixes will be on a different region. After applying the salts, you have the following rowkeys instead. Since you can now write to four separate regions, you theoretically have four times the throughput when writing that you would have if all the writes were going to the same region.</p> </div> <div class="listingblock"> <div class="content"> <pre>a-foo0003 b-foo0001 c-foo0004 d-foo0002</pre> </div> </div> <div class="paragraph"> <p>Then, if you add another row, it will randomly be assigned one of the four possible salt values and end up near one of the existing rows.</p> </div> <div class="listingblock"> <div class="content"> <pre>a-foo0003 b-foo0001 c-foo0003 c-foo0004 d-foo0002</pre> </div> </div> <div class="paragraph"> <p>Since this assignment will be random, you will need to do more work if you want to retrieve the rows in lexicographic order. In this way, salting attempts to increase throughput on writes, but has a cost during reads.</p> </div> </div> </div> <div class="paragraph"> <div class="title">Hashing</div> <p>Instead of a random assignment, you could use a one-way <em class="firstterm">hash</em> that would cause a given row to always be "salted" with the same prefix, in a way that would spread the load across the RegionServers, but allow for predictability during reads. Using a deterministic hash allows the client to reconstruct the complete rowkey and use a Get operation to retrieve that row as normal.</p> </div> <div class="exampleblock"> <div class="title">Example 17. Hashing Example</div> <div class="content"> Given the same situation in the salting example above, you could instead apply a one-way hash that would cause the row with key <code>foo0003</code> to always, and predictably, receive the <code>a</code> prefix. Then, to retrieve that row, you would already know the key. You could also optimize things so that certain pairs of keys were always in the same region, for instance. </div> </div> <div class="paragraph"> <div class="title">Reversing the Key</div> <p>A third common trick for preventing hotspotting is to reverse a fixed-width or numeric row key so that the part that changes the most often (the least significant digit) is first. This effectively randomizes row keys, but sacrifices row ordering properties.</p> </div> <div class="paragraph"> <p>See <a href="https://communities.intel.com/community/itpeernetwork/datastack/blog/2013/11/10/discussion-on-designing-hbase-tables" class="bare">https://communities.intel.com/community/itpeernetwork/datastack/blog/2013/11/10/discussion-on-designing-hbase-tables</a>, and <a href="http://phoenix.apache.org/salted.html">article on Salted Tables</a> from the Phoenix project, and the discussion in the comments of <a href="https://issues.apache.org/jira/browse/HBASE-11682">HBASE-11682</a> for more information about avoiding hotspotting.</p> </div> </div> <div class="sect2"> <h3 id="timeseries"><a class="anchor" href="#timeseries"></a>35.2. Monotonically Increasing Row Keys/Timeseries Data</h3> <div class="paragraph"> <p>In the HBase chapter of Tom White’s book <a href="http://oreilly.com/catalog/9780596521981">Hadoop: The Definitive Guide</a> (O’Reilly) there is a an optimization note on watching out for a phenomenon where an import process walks in lock-step with all clients in concert pounding one of the table’s regions (and thus, a single node), then moving onto the next region, etc. With monotonically increasing row-keys (i.e., using a timestamp), this will happen. See this comic by IKai Lan on why monotonically increasing row keys are problematic in BigTable-like datastores: <a href="http://ikaisays.com/2011/01/25/app-engine-datastore-tip-monotonically-increasing-values-are-bad/">monotonically increasing values are bad</a>. The pile-up on a single region brought on by monotonically increasing keys can be mitigated by randomizing the input records to not be in sorted order, but in general it’s best to avoid using a timestamp or a sequence (e.g. 1, 2, 3) as the row-key.</p> </div> <div class="paragraph"> <p>If you do need to upload time series data into HBase, you should study <a href="http://opentsdb.net/">OpenTSDB</a> as a successful example. It has a page describing the link: <a href="http://opentsdb.net/schema.html">schema</a> it uses in HBase. The key format in OpenTSDB is effectively [metric_type][event_timestamp], which would appear at first glance to contradict the previous advice about not using a timestamp as the key. However, the difference is that the timestamp is not in the <em>lead</em> position of the key, and the design assumption is that there are dozens or hundreds (or more) of different metric types. Thus, even with a continual stream of input data with a mix of metric types, the Puts are distributed across various points of regions in the table.</p> </div> <div class="paragraph"> <p>See <a href="#schema.casestudies">schema.casestudies</a> for some rowkey design examples.</p> </div> </div> <div class="sect2"> <h3 id="keysize"><a class="anchor" href="#keysize"></a>35.3. Try to minimize row and column sizes</h3> <div class="paragraph"> <p>In HBase, values are always freighted with their coordinates; as a cell value passes through the system, it’ll be accompanied by its row, column name, and timestamp - always. If your rows and column names are large, especially compared to the size of the cell value, then you may run up against some interesting scenarios. One such is the case described by Marc Limotte at the tail of <a href="https://issues.apache.org/jira/browse/HBASE-3551?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13005272#comment-13005272">HBASE-3551</a> (recommended!). Therein, the indices that are kept on HBase storefiles (<a href="#hfile">StoreFile (HFile)</a>) to facilitate random access may end up occupying large chunks of the HBase allotted RAM because the cell value coordinates are large. Mark in the above cited comment suggests upping the block size so entries in the store file index happen at a larger interval or modify the table schema so it makes for smaller rows and column names. Compression will also make for larger indices. See the thread <a href="http://search-hadoop.com/m/hemBv1LiN4Q1/a+question+storefileIndexSize&subj=a+question+storefileIndexSize">a question storefileIndexSize</a> up on the user mailing list.</p> </div> <div class="paragraph"> <p>Most of the time small inefficiencies don’t matter all that much. Unfortunately, this is a case where they do. Whatever patterns are selected for ColumnFamilies, attributes, and rowkeys they could be repeated several billion times in your data.</p> </div> <div class="paragraph"> <p>See <a href="#keyvalue">keyvalue</a> for more information on HBase stores data internally to see why this is important.</p> </div> <div class="sect3"> <h4 id="keysize.cf"><a class="anchor" href="#keysize.cf"></a>35.3.1. Column Families</h4> <div class="paragraph"> <p>Try to keep the ColumnFamily names as small as possible, preferably one character (e.g. "d" for data/default).</p> </div> <div class="paragraph"> <p>See <a href="#keyvalue">[keyvalue]</a> for more information on HBase stores data internally to see why this is important.</p> </div> </div> <div class="sect3"> <h4 id="keysize.attributes"><a class="anchor" href="#keysize.attributes"></a>35.3.2. Attributes</h4> <div class="paragraph"> <p>Although verbose attribute names (e.g., "myVeryImportantAttribute") are easier to read, prefer shorter attribute names (e.g., "via") to store in HBase.</p> </div> <div class="paragraph"> <p>See <a href="#keyvalue">keyvalue</a> for more information on HBase stores data internally to see why this is important.</p> </div> </div> <div class="sect3"> <h4 id="keysize.row"><a class="anchor" href="#keysize.row"></a>35.3.3. Rowkey Length</h4> <div class="paragraph"> <p>Keep them as short as is reasonable such that they can still be useful for required data access (e.g. Get vs. Scan). A short key that is useless for data access is not better than a longer key with better get/scan properties. Expect tradeoffs when designing rowkeys.</p> </div> </div> <div class="sect3"> <h4 id="keysize.patterns"><a class="anchor" href="#keysize.patterns"></a>35.3.4. Byte Patterns</h4> <div class="paragraph"> <p>A long is 8 bytes. You can store an unsigned number up to 18,446,744,073,709,551,615 in those eight bytes. If you stored this number as a String — presuming a byte per character — you need nearly 3x the bytes.</p> </div> <div class="paragraph"> <p>Not convinced? Below is some sample code that you can run on your own.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="comment">// long</span> <span class="comment">//</span> <span class="type">long</span> l = <span class="integer">1234567890L</span>; <span class="type">byte</span><span class="type">[]</span> lb = Bytes.toBytes(l); <span class="predefined-type">System</span>.out.println(<span class="string"><span class="delimiter">"</span><span class="content">long bytes length: </span><span class="delimiter">"</span></span> + lb.length); <span class="comment">// returns 8</span> <span class="predefined-type">String</span> s = <span class="predefined-type">String</span>.valueOf(l); <span class="type">byte</span><span class="type">[]</span> sb = Bytes.toBytes(s); <span class="predefined-type">System</span>.out.println(<span class="string"><span class="delimiter">"</span><span class="content">long as string length: </span><span class="delimiter">"</span></span> + sb.length); <span class="comment">// returns 10</span> <span class="comment">// hash</span> <span class="comment">//</span> <span class="predefined-type">MessageDigest</span> md = <span class="predefined-type">MessageDigest</span>.getInstance(<span class="string"><span class="delimiter">"</span><span class="content">MD5</span><span class="delimiter">"</span></span>); <span class="type">byte</span><span class="type">[]</span> digest = md.digest(Bytes.toBytes(s)); <span class="predefined-type">System</span>.out.println(<span class="string"><span class="delimiter">"</span><span class="content">md5 digest bytes length: </span><span class="delimiter">"</span></span> + digest.length); <span class="comment">// returns 16</span> <span class="predefined-type">String</span> sDigest = <span class="keyword">new</span> <span class="predefined-type">String</span>(digest); <span class="type">byte</span><span class="type">[]</span> sbDigest = Bytes.toBytes(sDigest); <span class="predefined-type">System</span>.out.println(<span class="string"><span class="delimiter">"</span><span class="content">md5 digest as string length: </span><span class="delimiter">"</span></span> + sbDigest.length); <span class="comment">// returns 26</span></code></pre> </div> </div> <div class="paragraph"> <p>Unfortunately, using a binary representation of a type will make your data harder to read outside of your code. For example, this is what you will see in the shell when you increment a value:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">001</span>:<span class="integer">0</span>> incr <span class="string"><span class="delimiter">'</span><span class="content">t</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">f:q</span><span class="delimiter">'</span></span>, <span class="integer">1</span> COUNTER VALUE = <span class="integer">1</span> hbase(main):<span class="octal">002</span>:<span class="integer">0</span>> get <span class="string"><span class="delimiter">'</span><span class="content">t</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r</span><span class="delimiter">'</span></span> COLUMN CELL f:q timestamp=<span class="integer">1369163040570</span>, value=<span class="error">\</span>x00<span class="error">\</span>x00<span class="error">\</span>x00<span class="error">\</span>x00<span class="error">\</span>x00<span class="error">\</span>x00<span class="error">\</span>x00<span class="error">\</span>x01 <span class="integer">1</span> row(s) in <span class="float">0.0310</span> seconds</code></pre> </div> </div> <div class="paragraph"> <p>The shell makes a best effort to print a string, and it this case it decided to just print the hex. The same will happen to your row keys inside the region names. It can be okay if you know what’s being stored, but it might also be unreadable if arbitrary data can be put in the same cells. This is the main trade-off.</p> </div> </div> </div> <div class="sect2"> <h3 id="reverse.timestamp"><a class="anchor" href="#reverse.timestamp"></a>35.4. Reverse Timestamps</h3> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Reverse Scan API</div> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-4811">HBASE-4811</a> implements an API to scan a table or a range within a table in reverse, reducing the need to optimize your schema for forward or reverse scanning. This feature is available in HBase 0.98 and later. See <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html#setReversed%28boolean" class="bare">https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html#setReversed%28boolean</a> for more information.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>A common problem in database processing is quickly finding the most recent version of a value. A technique using reverse timestamps as a part of the key can help greatly with a special case of this problem. Also found in the HBase chapter of Tom White’s book Hadoop: The Definitive Guide (O’Reilly), the technique involves appending (<code>Long.MAX_VALUE - timestamp</code>) to the end of any key, e.g. [key][reverse_timestamp].</p> </div> <div class="paragraph"> <p>The most recent value for [key] in a table can be found by performing a Scan for [key] and obtaining the first record. Since HBase keys are in sorted order, this key sorts before any older row-keys for [key] and thus is first.</p> </div> <div class="paragraph"> <p>This technique would be used instead of using <a href="#schema.versions">Number of Versions</a> where the intent is to hold onto all versions "forever" (or a very long time) and at the same time quickly obtain access to any other version by using the same Scan technique.</p> </div> </div> <div class="sect2"> <h3 id="rowkey.scope"><a class="anchor" href="#rowkey.scope"></a>35.5. Rowkeys and ColumnFamilies</h3> <div class="paragraph"> <p>Rowkeys are scoped to ColumnFamilies. Thus, the same rowkey could exist in each ColumnFamily that exists in a table without collision.</p> </div> </div> <div class="sect2"> <h3 id="changing.rowkeys"><a class="anchor" href="#changing.rowkeys"></a>35.6. Immutability of Rowkeys</h3> <div class="paragraph"> <p>Rowkeys cannot be changed. The only way they can be "changed" in a table is if the row is deleted and then re-inserted. This is a fairly common question on the HBase dist-list so it pays to get the rowkeys right the first time (and/or before you’ve inserted a lot of data).</p> </div> </div> <div class="sect2"> <h3 id="rowkey.regionsplits"><a class="anchor" href="#rowkey.regionsplits"></a>35.7. Relationship Between RowKeys and Region Splits</h3> <div class="paragraph"> <p>If you pre-split your table, it is <em>critical</em> to understand how your rowkey will be distributed across the region boundaries. As an example of why this is important, consider the example of using displayable hex characters as the lead position of the key (e.g., "0000000000000000" to "ffffffffffffffff"). Running those key ranges through <code>Bytes.split</code> (which is the split strategy used when creating regions in <code>Admin.createTable(byte[] startKey, byte[] endKey, numRegions)</code> for 10 regions will generate the following splits…</p> </div> <div class="listingblock"> <div class="content"> <pre>48 48 48 48 48 48 48 48 48 48 48 48 48 48 48 48 // 0 54 -10 -10 -10 -10 -10 -10 -10 -10 -10 -10 -10 -10 -10 -10 -10 // 6 61 -67 -67 -67 -67 -67 -67 -67 -67 -67 -67 -67 -67 -67 -67 -68 // = 68 -124 -124 -124 -124 -124 -124 -124 -124 -124 -124 -124 -124 -124 -124 -126 // D 75 75 75 75 75 75 75 75 75 75 75 75 75 75 75 72 // K 82 18 18 18 18 18 18 18 18 18 18 18 18 18 18 14 // R 88 -40 -40 -40 -40 -40 -40 -40 -40 -40 -40 -40 -40 -40 -40 -44 // X 95 -97 -97 -97 -97 -97 -97 -97 -97 -97 -97 -97 -97 -97 -97 -102 // _ 102 102 102 102 102 102 102 102 102 102 102 102 102 102 102 102 // f</pre> </div> </div> <div class="paragraph"> <p>(note: the lead byte is listed to the right as a comment.) Given that the first split is a '0' and the last split is an 'f', everything is great, right? Not so fast.</p> </div> <div class="paragraph"> <p>The problem is that all the data is going to pile up in the first 2 regions and the last region thus creating a "lumpy" (and possibly "hot") region problem. To understand why, refer to an <a href="http://www.asciitable.com">ASCII Table</a>. '0' is byte 48, and 'f' is byte 102, but there is a huge gap in byte values (bytes 58 to 96) that will <em>never appear in this keyspace</em> because the only values are [0-9] and [a-f]. Thus, the middle regions regions will never be used. To make pre-spliting work with this example keyspace, a custom definition of splits (i.e., and not relying on the built-in split method) is required.</p> </div> <div class="paragraph"> <p>Lesson #1: Pre-splitting tables is generally a best practice, but you need to pre-split them in such a way that all the regions are accessible in the keyspace. While this example demonstrated the problem with a hex-key keyspace, the same problem can happen with <em>any</em> keyspace. Know your data.</p> </div> <div class="paragraph"> <p>Lesson #2: While generally not advisable, using hex-keys (and more generally, displayable data) can still work with pre-split tables as long as all the created regions are accessible in the keyspace.</p> </div> <div class="paragraph"> <p>To conclude this example, the following is an example of how appropriate splits can be pre-created for hex-keys:.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">boolean</span> createTable(Admin admin, HTableDescriptor table, <span class="type">byte</span><span class="type">[]</span><span class="type">[]</span> splits) <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="keyword">try</span> { admin.createTable( table, splits ); <span class="keyword">return</span> <span class="predefined-constant">true</span>; } <span class="keyword">catch</span> (TableExistsException e) { logger.info(<span class="string"><span class="delimiter">"</span><span class="content">table </span><span class="delimiter">"</span></span> + table.getNameAsString() + <span class="string"><span class="delimiter">"</span><span class="content"> already exists</span><span class="delimiter">"</span></span>); <span class="comment">// the table already exists...</span> <span class="keyword">return</span> <span class="predefined-constant">false</span>; } } <span class="directive">public</span> <span class="directive">static</span> <span class="type">byte</span><span class="type">[]</span><span class="type">[]</span> getHexSplits(<span class="predefined-type">String</span> startKey, <span class="predefined-type">String</span> endKey, <span class="type">int</span> numRegions) { <span class="type">byte</span><span class="type">[]</span><span class="type">[]</span> splits = <span class="keyword">new</span> <span class="type">byte</span>[numRegions-<span class="integer">1</span>]<span class="type">[]</span>; <span class="predefined-type">BigInteger</span> lowestKey = <span class="keyword">new</span> <span class="predefined-type">BigInteger</span>(startKey, <span class="integer">16</span>); <span class="predefined-type">BigInteger</span> highestKey = <span class="keyword">new</span> <span class="predefined-type">BigInteger</span>(endKey, <span class="integer">16</span>); <span class="predefined-type">BigInteger</span> range = highestKey.subtract(lowestKey); <span class="predefined-type">BigInteger</span> regionIncrement = range.divide(<span class="predefined-type">BigInteger</span>.valueOf(numRegions)); lowestKey = lowestKey.add(regionIncrement); <span class="keyword">for</span>(<span class="type">int</span> i=<span class="integer">0</span>; i < numRegions-<span class="integer">1</span>;i++) { <span class="predefined-type">BigInteger</span> key = lowestKey.add(regionIncrement.multiply(<span class="predefined-type">BigInteger</span>.valueOf(i))); <span class="type">byte</span><span class="type">[]</span> b = <span class="predefined-type">String</span>.format(<span class="string"><span class="delimiter">"</span><span class="content">%016x</span><span class="delimiter">"</span></span>, key).getBytes(); splits[i] = b; } <span class="keyword">return</span> splits; }</code></pre> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="schema.versions"><a class="anchor" href="#schema.versions"></a>36. Number of Versions</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="schema.versions.max"><a class="anchor" href="#schema.versions.max"></a>36.1. Maximum Number of Versions</h3> <div class="paragraph"> <p>The maximum number of row versions to store is configured per column family via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a>. The default for max versions is 1. This is an important parameter because as described in <a href="#datamodel">Data Model</a> section HBase does <em>not</em> overwrite row values, but rather stores different values per row by time (and qualifier). Excess versions are removed during major compactions. The number of max versions may need to be increased or decreased depending on application needs.</p> </div> <div class="paragraph"> <p>It is not recommended setting the number of max versions to an exceedingly high level (e.g., hundreds or more) unless those old values are very dear to you because this will greatly increase StoreFile size.</p> </div> </div> <div class="sect2"> <h3 id="schema.minversions"><a class="anchor" href="#schema.minversions"></a>36.2. Minimum Number of Versions</h3> <div class="paragraph"> <p>Like maximum number of row versions, the minimum number of row versions to keep is configured per column family via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a>. The default for min versions is 0, which means the feature is disabled. The minimum number of row versions parameter is used together with the time-to-live parameter and can be combined with the number of row versions parameter to allow configurations such as "keep the last T minutes worth of data, at most N versions, <em>but keep at least M versions around</em>" (where M is the value for minimum number of row versions, M<N). This parameter should only be set when time-to-live is enabled for a column family and must be less than the number of row versions.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="supported.datatypes"><a class="anchor" href="#supported.datatypes"></a>37. Supported Datatypes</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase supports a "bytes-in/bytes-out" interface via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Put.html">Put</a> and <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Result.html">Result</a>, so anything that can be converted to an array of bytes can be stored as a value. Input could be strings, numbers, complex objects, or even images as long as they can rendered as bytes.</p> </div> <div class="paragraph"> <p>There are practical limits to the size of values (e.g., storing 10-50MB objects in HBase would probably be too much to ask); search the mailling list for conversations on this topic. All rows in HBase conform to the <a href="#datamodel">Data Model</a>, and that includes versioning. Take that into consideration when making your design, as well as block size for the ColumnFamily.</p> </div> <div class="sect2"> <h3 id="_counters"><a class="anchor" href="#_counters"></a>37.1. Counters</h3> <div class="paragraph"> <p>One supported datatype that deserves special mention are "counters" (i.e., the ability to do atomic increments of numbers). See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#increment%28org.apache.hadoop.hbase.client.Increment%29">Increment</a> in <code>Table</code>.</p> </div> <div class="paragraph"> <p>Synchronization on counters are done on the RegionServer, not in the client.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="schema.joins"><a class="anchor" href="#schema.joins"></a>38. Joins</h2> <div class="sectionbody"> <div class="paragraph"> <p>If you have multiple tables, don’t forget to factor in the potential for <a href="#joins">[joins]</a> into the schema design.</p> </div> </div> </div> <div class="sect1"> <h2 id="ttl"><a class="anchor" href="#ttl"></a>39. Time To Live (TTL)</h2> <div class="sectionbody"> <div class="paragraph"> <p>ColumnFamilies can set a TTL length in seconds, and HBase will automatically delete rows once the expiration time is reached. This applies to <em>all</em> versions of a row - even the current one. The TTL time encoded in the HBase for the row is specified in UTC.</p> </div> <div class="paragraph"> <p>Store files which contains only expired rows are deleted on minor compaction. Setting <code>hbase.store.delete.expired.storefile</code> to <code>false</code> disables this feature. Setting minimum number of versions to other than 0 also disables this.</p> </div> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a> for more information.</p> </div> <div class="paragraph"> <p>Recent versions of HBase also support setting time to live on a per cell basis. See <a href="https://issues.apache.org/jira/browse/HBASE-10560">HBASE-10560</a> for more information. Cell TTLs are submitted as an attribute on mutation requests (Appends, Increments, Puts, etc.) using Mutation#setTTL. If the TTL attribute is set, it will be applied to all cells updated on the server by the operation. There are two notable differences between cell TTL handling and ColumnFamily TTLs:</p> </div> <div class="ulist"> <ul> <li> <p>Cell TTLs are expressed in units of milliseconds instead of seconds.</p> </li> <li> <p>A cell TTLs cannot extend the effective lifetime of a cell beyond a ColumnFamily level TTL setting.</p> </li> </ul> </div> </div> </div> <div class="sect1"> <h2 id="cf.keep.deleted"><a class="anchor" href="#cf.keep.deleted"></a>40. Keeping Deleted Cells</h2> <div class="sectionbody"> <div class="paragraph"> <p>By default, delete markers extend back to the beginning of time. Therefore, <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</a> or <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</a> operations will not see a deleted cell (row or column), even when the Get or Scan operation indicates a time range before the delete marker was placed.</p> </div> <div class="paragraph"> <p>ColumnFamilies can optionally keep deleted cells. In this case, deleted cells can still be retrieved, as long as these operations specify a time range that ends before the timestamp of any delete that would affect the cells. This allows for point-in-time queries even in the presence of deletes.</p> </div> <div class="paragraph"> <p>Deleted cells are still subject to TTL and there will never be more than "maximum number of versions" deleted cells. A new "raw" scan options returns all deleted rows and the delete markers.</p> </div> <div class="exampleblock"> <div class="title">Example 18. Change the Value of <code>KEEP_DELETED_CELLS</code> Using HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> hbase> alter ‘t1′, NAME => ‘f1′, KEEP_DELETED_CELLS => true</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 19. Change the Value of <code>KEEP_DELETED_CELLS</code> Using the API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... HColumnDescriptor.setKeepDeletedCells(<span class="predefined-constant">true</span>); ...</code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>Let us illustrate the basic effect of setting the <code>KEEP_DELETED_CELLS</code> attribute on a table.</p> </div> <div class="paragraph"> <p>First, without:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">create <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {NAME=><span class="string"><span class="delimiter">'</span><span class="content">e</span><span class="delimiter">'</span></span>, VERSIONS=><span class="integer">2147483647</span>} put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, <span class="integer">10</span> put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, <span class="integer">12</span> put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, <span class="integer">14</span> delete <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="integer">11</span> hbase(main):<span class="octal">017</span>:<span class="integer">0</span>> scan <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {RAW=><span class="predefined-constant">true</span>, VERSIONS=><span class="integer">1000</span>} ROW COLUMN+CELL r1 column=e:c1, timestamp=<span class="integer">14</span>, value=value r1 column=e:c1, timestamp=<span class="integer">12</span>, value=value r1 column=e:c1, timestamp=<span class="integer">11</span>, type=DeleteColumn r1 column=e:c1, timestamp=<span class="integer">10</span>, value=value <span class="integer">1</span> row(s) in <span class="float">0.0120</span> seconds hbase(main):<span class="integer">018</span>:<span class="integer">0</span>> flush <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span> <span class="integer">0</span> row(s) in <span class="float">0.0350</span> seconds hbase(main):<span class="integer">019</span>:<span class="integer">0</span>> scan <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {RAW=><span class="predefined-constant">true</span>, VERSIONS=><span class="integer">1000</span>} ROW COLUMN+CELL r1 column=e:c1, timestamp=<span class="integer">14</span>, value=value r1 column=e:c1, timestamp=<span class="integer">12</span>, value=value r1 column=e:c1, timestamp=<span class="integer">11</span>, type=DeleteColumn <span class="integer">1</span> row(s) in <span class="float">0.0120</span> seconds hbase(main):<span class="octal">020</span>:<span class="integer">0</span>> major_compact <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span> <span class="integer">0</span> row(s) in <span class="float">0.0260</span> seconds hbase(main):<span class="octal">021</span>:<span class="integer">0</span>> scan <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {RAW=><span class="predefined-constant">true</span>, VERSIONS=><span class="integer">1000</span>} ROW COLUMN+CELL r1 column=e:c1, timestamp=<span class="integer">14</span>, value=value r1 column=e:c1, timestamp=<span class="integer">12</span>, value=value <span class="integer">1</span> row(s) in <span class="float">0.0120</span> seconds</code></pre> </div> </div> <div class="paragraph"> <p>Notice how delete cells are let go.</p> </div> <div class="paragraph"> <p>Now lets run the same test only with <code>KEEP_DELETED_CELLS</code> set on the table (you can do table or per-column-family):</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">005</span>:<span class="integer">0</span>> create <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {NAME=><span class="string"><span class="delimiter">'</span><span class="content">e</span><span class="delimiter">'</span></span>, VERSIONS=><span class="integer">2147483647</span>, KEEP_DELETED_CELLS => <span class="predefined-constant">true</span>} <span class="integer">0</span> row(s) in <span class="float">0.2160</span> seconds => Hbase::Table - test hbase(main):<span class="octal">006</span>:<span class="integer">0</span>> put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, <span class="integer">10</span> <span class="integer">0</span> row(s) in <span class="float">0.1070</span> seconds hbase(main):<span class="octal">007</span>:<span class="integer">0</span>> put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, <span class="integer">12</span> <span class="integer">0</span> row(s) in <span class="float">0.0140</span> seconds hbase(main):<span class="integer">008</span>:<span class="integer">0</span>> put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, <span class="integer">14</span> <span class="integer">0</span> row(s) in <span class="float">0.0160</span> seconds hbase(main):<span class="integer">009</span>:<span class="integer">0</span>> delete <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">r1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">e:c1</span><span class="delimiter">'</span></span>, <span class="integer">11</span> <span class="integer">0</span> row(s) in <span class="float">0.0290</span> seconds hbase(main):<span class="octal">010</span>:<span class="integer">0</span>> scan <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {RAW=><span class="predefined-constant">true</span>, VERSIONS=><span class="integer">1000</span>} ROW COLUMN+CELL r1 column=e:c1, timestamp=<span class="integer">14</span>, value=value r1 column=e:c1, timestamp=<span class="integer">12</span>, value=value r1 column=e:c1, timestamp=<span class="integer">11</span>, type=DeleteColumn r1 column=e:c1, timestamp=<span class="integer">10</span>, value=value <span class="integer">1</span> row(s) in <span class="float">0.0550</span> seconds hbase(main):<span class="octal">011</span>:<span class="integer">0</span>> flush <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span> <span class="integer">0</span> row(s) in <span class="float">0.2780</span> seconds hbase(main):<span class="octal">012</span>:<span class="integer">0</span>> scan <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {RAW=><span class="predefined-constant">true</span>, VERSIONS=><span class="integer">1000</span>} ROW COLUMN+CELL r1 column=e:c1, timestamp=<span class="integer">14</span>, value=value r1 column=e:c1, timestamp=<span class="integer">12</span>, value=value r1 column=e:c1, timestamp=<span class="integer">11</span>, type=DeleteColumn r1 column=e:c1, timestamp=<span class="integer">10</span>, value=value <span class="integer">1</span> row(s) in <span class="float">0.0620</span> seconds hbase(main):<span class="octal">013</span>:<span class="integer">0</span>> major_compact <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span> <span class="integer">0</span> row(s) in <span class="float">0.0530</span> seconds hbase(main):<span class="octal">014</span>:<span class="integer">0</span>> scan <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {RAW=><span class="predefined-constant">true</span>, VERSIONS=><span class="integer">1000</span>} ROW COLUMN+CELL r1 column=e:c1, timestamp=<span class="integer">14</span>, value=value r1 column=e:c1, timestamp=<span class="integer">12</span>, value=value r1 column=e:c1, timestamp=<span class="integer">11</span>, type=DeleteColumn r1 column=e:c1, timestamp=<span class="integer">10</span>, value=value <span class="integer">1</span> row(s) in <span class="float">0.0650</span> seconds</code></pre> </div> </div> <div class="paragraph"> <p>KEEP_DELETED_CELLS is to avoid removing Cells from HBase when the <em>only</em> reason to remove them is the delete marker. So with KEEP_DELETED_CELLS enabled deleted cells would get removed if either you write more versions than the configured max, or you have a TTL and Cells are in excess of the configured timeout, etc.</p> </div> </div> </div> <div class="sect1"> <h2 id="secondary.indexes"><a class="anchor" href="#secondary.indexes"></a>41. Secondary Indexes and Alternate Query Paths</h2> <div class="sectionbody"> <div class="paragraph"> <p>This section could also be titled "what if my table rowkey looks like <em>this</em> but I also want to query my table like <em>that</em>." A common example on the dist-list is where a row-key is of the format "user-timestamp" but there are reporting requirements on activity across users for certain time ranges. Thus, selecting by user is easy because it is in the lead position of the key, but time is not.</p> </div> <div class="paragraph"> <p>There is no single answer on the best way to handle this because it depends on…</p> </div> <div class="ulist"> <ul> <li> <p>Number of users</p> </li> <li> <p>Data size and data arrival rate</p> </li> <li> <p>Flexibility of reporting requirements (e.g., completely ad-hoc date selection vs. pre-configured ranges)</p> </li> <li> <p>Desired execution speed of query (e.g., 90 seconds may be reasonable to some for an ad-hoc report, whereas it may be too long for others)</p> </li> </ul> </div> <div class="paragraph"> <p>and solutions are also influenced by the size of the cluster and how much processing power you have to throw at the solution. Common techniques are in sub-sections below. This is a comprehensive, but not exhaustive, list of approaches.</p> </div> <div class="paragraph"> <p>It should not be a surprise that secondary indexes require additional cluster space and processing. This is precisely what happens in an RDBMS because the act of creating an alternate index requires both space and processing cycles to update. RDBMS products are more advanced in this regard to handle alternative index management out of the box. However, HBase scales better at larger data volumes, so this is a feature trade-off.</p> </div> <div class="paragraph"> <p>Pay attention to <a href="#performance">Apache HBase Performance Tuning</a> when implementing any of these approaches.</p> </div> <div class="paragraph"> <p>Additionally, see the David Butler response in this dist-list thread <a href="http://search-hadoop.com/m/nvbiBp2TDP/Stargate%252Bhbase&subj=Stargate+hbase">HBase, mail # user - Stargate+hbase</a></p> </div> <div class="sect2"> <h3 id="secondary.indexes.filter"><a class="anchor" href="#secondary.indexes.filter"></a>41.1. Filter Query</h3> <div class="paragraph"> <p>Depending on the case, it may be appropriate to use <a href="#client.filter">Client Request Filters</a>. In this case, no secondary index is created. However, don’t try a full-scan on a large table like this from an application (i.e., single-threaded client).</p> </div> </div> <div class="sect2"> <h3 id="secondary.indexes.periodic"><a class="anchor" href="#secondary.indexes.periodic"></a>41.2. Periodic-Update Secondary Index</h3> <div class="paragraph"> <p>A secondary index could be created in an other table which is periodically updated via a MapReduce job. The job could be executed intra-day, but depending on load-strategy it could still potentially be out of sync with the main data table.</p> </div> <div class="paragraph"> <p>See <a href="#mapreduce.example.readwrite">mapreduce.example.readwrite</a> for more information.</p> </div> </div> <div class="sect2"> <h3 id="secondary.indexes.dualwrite"><a class="anchor" href="#secondary.indexes.dualwrite"></a>41.3. Dual-Write Secondary Index</h3> <div class="paragraph"> <p>Another strategy is to build the secondary index while publishing data to the cluster (e.g., write to data table, write to index table). If this is approach is taken after a data table already exists, then bootstrapping will be needed for the secondary index with a MapReduce job (see <a href="#secondary.indexes.periodic">secondary.indexes.periodic</a>).</p> </div> </div> <div class="sect2"> <h3 id="secondary.indexes.summary"><a class="anchor" href="#secondary.indexes.summary"></a>41.4. Summary Tables</h3> <div class="paragraph"> <p>Where time-ranges are very wide (e.g., year-long report) and where the data is voluminous, summary tables are a common approach. These would be generated with MapReduce jobs into another table.</p> </div> <div class="paragraph"> <p>See <a href="#mapreduce.example.summary">mapreduce.example.summary</a> for more information.</p> </div> </div> <div class="sect2"> <h3 id="secondary.indexes.coproc"><a class="anchor" href="#secondary.indexes.coproc"></a>41.5. Coprocessor Secondary Index</h3> <div class="paragraph"> <p>Coprocessors act like RDBMS triggers. These were added in 0.92. For more information, see <a href="#coprocessors">coprocessors</a></p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_constraints"><a class="anchor" href="#_constraints"></a>42. Constraints</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase currently supports 'constraints' in traditional (SQL) database parlance. The advised usage for Constraints is in enforcing business rules for attributes in the table (e.g. make sure values are in the range 1-10). Constraints could also be used to enforce referential integrity, but this is strongly discouraged as it will dramatically decrease the write throughput of the tables where integrity checking is enabled. Extensive documentation on using Constraints can be found at: <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/constraint">Constraint</a> since version 0.94.</p> </div> </div> </div> <div class="sect1"> <h2 id="schema.casestudies"><a class="anchor" href="#schema.casestudies"></a>43. Schema Design Case Studies</h2> <div class="sectionbody"> <div class="paragraph"> <p>The following will describe some typical data ingestion use-cases with HBase, and how the rowkey design and construction can be approached. Note: this is just an illustration of potential approaches, not an exhaustive list. Know your data, and know your processing requirements.</p> </div> <div class="paragraph"> <p>It is highly recommended that you read the rest of the <a href="#schema">HBase and Schema Design</a> first, before reading these case studies.</p> </div> <div class="paragraph"> <p>The following case studies are described:</p> </div> <div class="ulist"> <ul> <li> <p>Log Data / Timeseries Data</p> </li> <li> <p>Log Data / Timeseries on Steroids</p> </li> <li> <p>Customer/Order</p> </li> <li> <p>Tall/Wide/Middle Schema Design</p> </li> <li> <p>List Data</p> </li> </ul> </div> <div class="sect2"> <h3 id="schema.casestudies.log_timeseries"><a class="anchor" href="#schema.casestudies.log_timeseries"></a>43.1. Case Study - Log Data and Timeseries Data</h3> <div class="paragraph"> <p>Assume that the following data elements are being collected.</p> </div> <div class="ulist"> <ul> <li> <p>Hostname</p> </li> <li> <p>Timestamp</p> </li> <li> <p>Log event</p> </li> <li> <p>Value/message</p> </li> </ul> </div> <div class="paragraph"> <p>We can store them in an HBase table called LOG_DATA, but what will the rowkey be? From these attributes the rowkey will be some combination of hostname, timestamp, and log-event - but what specifically?</p> </div> <div class="sect3"> <h4 id="schema.casestudies.log_timeseries.tslead"><a class="anchor" href="#schema.casestudies.log_timeseries.tslead"></a>43.1.1. Timestamp In The Rowkey Lead Position</h4> <div class="paragraph"> <p>The rowkey <code>[timestamp][hostname][log-event]</code> suffers from the monotonically increasing rowkey problem described in <a href="#timeseries">Monotonically Increasing Row Keys/Timeseries Data</a>.</p> </div> <div class="paragraph"> <p>There is another pattern frequently mentioned in the dist-lists about "bucketing" timestamps, by performing a mod operation on the timestamp. If time-oriented scans are important, this could be a useful approach. Attention must be paid to the number of buckets, because this will require the same number of scans to return results.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="type">long</span> bucket = timestamp % numBuckets;</code></pre> </div> </div> <div class="paragraph"> <p>to construct:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">[bucket][timestamp][hostname][log-event]</code></pre> </div> </div> <div class="paragraph"> <p>As stated above, to select data for a particular timerange, a Scan will need to be performed for each bucket. 100 buckets, for example, will provide a wide distribution in the keyspace but it will require 100 Scans to obtain data for a single timestamp, so there are trade-offs.</p> </div> </div> <div class="sect3"> <h4 id="schema.casestudies.log_timeseries.hostlead"><a class="anchor" href="#schema.casestudies.log_timeseries.hostlead"></a>43.1.2. Host In The Rowkey Lead Position</h4> <div class="paragraph"> <p>The rowkey <code>[hostname][log-event][timestamp]</code> is a candidate if there is a large-ish number of hosts to spread the writes and reads across the keyspace. This approach would be useful if scanning by hostname was a priority.</p> </div> </div> <div class="sect3"> <h4 id="schema.casestudies.log_timeseries.revts"><a class="anchor" href="#schema.casestudies.log_timeseries.revts"></a>43.1.3. Timestamp, or Reverse Timestamp?</h4> <div class="paragraph"> <p>If the most important access path is to pull most recent events, then storing the timestamps as reverse-timestamps (e.g., <code>timestamp = Long.MAX_VALUE – timestamp</code>) will create the property of being able to do a Scan on <code>[hostname][log-event]</code> to obtain the quickly obtain the most recently captured events.</p> </div> <div class="paragraph"> <p>Neither approach is wrong, it just depends on what is most appropriate for the situation.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Reverse Scan API</div> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-4811">HBASE-4811</a> implements an API to scan a table or a range within a table in reverse, reducing the need to optimize your schema for forward or reverse scanning. This feature is available in HBase 0.98 and later. See <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html#setReversed%28boolean" class="bare">https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html#setReversed%28boolean</a> for more information.</p> </div> </td> </tr> </table> </div> </div> <div class="sect3"> <h4 id="schema.casestudies.log_timeseries.varkeys"><a class="anchor" href="#schema.casestudies.log_timeseries.varkeys"></a>43.1.4. Variangle Length or Fixed Length Rowkeys?</h4> <div class="paragraph"> <p>It is critical to remember that rowkeys are stamped on every column in HBase. If the hostname is <code>a</code> and the event type is <code>e1</code> then the resulting rowkey would be quite small. However, what if the ingested hostname is <code>myserver1.mycompany.com</code> and the event type is <code>com.package1.subpackage2.subsubpackage3.ImportantService</code>?</p> </div> <div class="paragraph"> <p>It might make sense to use some substitution in the rowkey. There are at least two approaches: hashed and numeric. In the Hostname In The Rowkey Lead Position example, it might look like this:</p> </div> <div class="paragraph"> <p>Composite Rowkey With Hashes:</p> </div> <div class="ulist"> <ul> <li> <p>[MD5 hash of hostname] = 16 bytes</p> </li> <li> <p>[MD5 hash of event-type] = 16 bytes</p> </li> <li> <p>[timestamp] = 8 bytes</p> </li> </ul> </div> <div class="paragraph"> <p>Composite Rowkey With Numeric Substitution:</p> </div> <div class="paragraph"> <p>For this approach another lookup table would be needed in addition to LOG_DATA, called LOG_TYPES. The rowkey of LOG_TYPES would be:</p> </div> <div class="ulist"> <ul> <li> <p>[type] (e.g., byte indicating hostname vs. event-type)</p> </li> <li> <p>[bytes] variable length bytes for raw hostname or event-type.</p> </li> </ul> </div> <div class="paragraph"> <p>A column for this rowkey could be a long with an assigned number, which could be obtained by using an <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#incrementColumnValue%28byte" class="bare">http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#incrementColumnValue%28byte</a>,%20byte[],%20byte[],%20long%29[HBase counter].</p> </div> <div class="paragraph"> <p>So the resulting composite rowkey would be:</p> </div> <div class="ulist"> <ul> <li> <p>[substituted long for hostname] = 8 bytes</p> </li> <li> <p>[substituted long for event type] = 8 bytes</p> </li> <li> <p>[timestamp] = 8 bytes</p> </li> </ul> </div> <div class="paragraph"> <p>In either the Hash or Numeric substitution approach, the raw values for hostname and event-type can be stored as columns.</p> </div> </div> </div> <div class="sect2"> <h3 id="schema.casestudies.log_steroids"><a class="anchor" href="#schema.casestudies.log_steroids"></a>43.2. Case Study - Log Data and Timeseries Data on Steroids</h3> <div class="paragraph"> <p>This effectively is the OpenTSDB approach. What OpenTSDB does is re-write data and pack rows into columns for certain time-periods. For a detailed explanation, see: link:http://opentsdb.net/schema.html, and <a href="http://www.cloudera.com/content/cloudera/en/resources/library/hbasecon/video-hbasecon-2012-lessons-learned-from-opentsdb.html">Lessons Learned from OpenTSDB</a> from HBaseCon2012.</p> </div> <div class="paragraph"> <p>But this is how the general concept works: data is ingested, for example, in this manner…</p> </div> <div class="listingblock"> <div class="content"> <pre>[hostname][log-event][timestamp1] [hostname][log-event][timestamp2] [hostname][log-event][timestamp3]</pre> </div> </div> <div class="paragraph"> <p>with separate rowkeys for each detailed event, but is re-written like this…</p> </div> <div class="listingblock"> <div class="content"> <pre>[hostname][log-event][timerange]</pre> </div> </div> <div class="paragraph"> <p>and each of the above events are converted into columns stored with a time-offset relative to the beginning timerange (e.g., every 5 minutes). This is obviously a very advanced processing technique, but HBase makes this possible.</p> </div> </div> <div class="sect2"> <h3 id="schema.casestudies.custorder"><a class="anchor" href="#schema.casestudies.custorder"></a>43.3. Case Study - Customer/Order</h3> <div class="paragraph"> <p>Assume that HBase is used to store customer and order information. There are two core record-types being ingested: a Customer record type, and Order record type.</p> </div> <div class="paragraph"> <p>The Customer record type would include all the things that you’d typically expect:</p> </div> <div class="ulist"> <ul> <li> <p>Customer number</p> </li> <li> <p>Customer name</p> </li> <li> <p>Address (e.g., city, state, zip)</p> </li> <li> <p>Phone numbers, etc.</p> </li> </ul> </div> <div class="paragraph"> <p>The Order record type would include things like:</p> </div> <div class="ulist"> <ul> <li> <p>Customer number</p> </li> <li> <p>Order number</p> </li> <li> <p>Sales date</p> </li> <li> <p>A series of nested objects for shipping locations and line-items (see <a href="#schema.casestudies.custorder.obj">Order Object Design</a> for details)</p> </li> </ul> </div> <div class="paragraph"> <p>Assuming that the combination of customer number and sales order uniquely identify an order, these two attributes will compose the rowkey, and specifically a composite key such as:</p> </div> <div class="listingblock"> <div class="content"> <pre>[customer number][order number]</pre> </div> </div> <div class="paragraph"> <p>for a ORDER table. However, there are more design decisions to make: are the <em>raw</em> values the best choices for rowkeys?</p> </div> <div class="paragraph"> <p>The same design questions in the Log Data use-case confront us here. What is the keyspace of the customer number, and what is the format (e.g., numeric? alphanumeric?) As it is advantageous to use fixed-length keys in HBase, as well as keys that can support a reasonable spread in the keyspace, similar options appear:</p> </div> <div class="paragraph"> <p>Composite Rowkey With Hashes:</p> </div> <div class="ulist"> <ul> <li> <p>[MD5 of customer number] = 16 bytes</p> </li> <li> <p>[MD5 of order number] = 16 bytes</p> </li> </ul> </div> <div class="paragraph"> <p>Composite Numeric/Hash Combo Rowkey:</p> </div> <div class="ulist"> <ul> <li> <p>[substituted long for customer number] = 8 bytes</p> </li> <li> <p>[MD5 of order number] = 16 bytes</p> </li> </ul> </div> <div class="sect3"> <h4 id="schema.casestudies.custorder.tables"><a class="anchor" href="#schema.casestudies.custorder.tables"></a>43.3.1. Single Table? Multiple Tables?</h4> <div class="paragraph"> <p>A traditional design approach would have separate tables for CUSTOMER and SALES. Another option is to pack multiple record types into a single table (e.g., CUSTOMER++).</p> </div> <div class="paragraph"> <p>Customer Record Type Rowkey:</p> </div> <div class="ulist"> <ul> <li> <p>[customer-id]</p> </li> <li> <p>[type] = type indicating `1' for customer record type</p> </li> </ul> </div> <div class="paragraph"> <p>Order Record Type Rowkey:</p> </div> <div class="ulist"> <ul> <li> <p>[customer-id]</p> </li> <li> <p>[type] = type indicating `2' for order record type</p> </li> <li> <p>[order]</p> </li> </ul> </div> <div class="paragraph"> <p>The advantage of this particular CUSTOMER++ approach is that organizes many different record-types by customer-id (e.g., a single scan could get you everything about that customer). The disadvantage is that it’s not as easy to scan for a particular record-type.</p> </div> </div> <div class="sect3"> <h4 id="schema.casestudies.custorder.obj"><a class="anchor" href="#schema.casestudies.custorder.obj"></a>43.3.2. Order Object Design</h4> <div class="paragraph"> <p>Now we need to address how to model the Order object. Assume that the class structure is as follows:</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Order</dt> <dd> <p>(an Order can have multiple ShippingLocations</p> </dd> <dt class="hdlist1">LineItem</dt> <dd> <p>(a ShippingLocation can have multiple LineItems</p> </dd> </dl> </div> <div class="paragraph"> <p>there are multiple options on storing this data.</p> </div> <div class="sect4"> <h5 id="schema.casestudies.custorder.obj.norm"><a class="anchor" href="#schema.casestudies.custorder.obj.norm"></a>Completely Normalized</h5> <div class="paragraph"> <p>With this approach, there would be separate tables for ORDER, SHIPPING_LOCATION, and LINE_ITEM.</p> </div> <div class="paragraph"> <p>The ORDER table’s rowkey was described above: <a href="#schema.casestudies.custorder">schema.casestudies.custorder</a></p> </div> <div class="paragraph"> <p>The SHIPPING_LOCATION’s composite rowkey would be something like this:</p> </div> <div class="ulist"> <ul> <li> <p>[order-rowkey]</p> </li> <li> <p>[shipping location number] (e.g., 1st location, 2nd, etc.)</p> </li> </ul> </div> <div class="paragraph"> <p>The LINE_ITEM table’s composite rowkey would be something like this:</p> </div> <div class="ulist"> <ul> <li> <p>[order-rowkey]</p> </li> <li> <p>[shipping location number] (e.g., 1st location, 2nd, etc.)</p> </li> <li> <p>[line item number] (e.g., 1st lineitem, 2nd, etc.)</p> </li> </ul> </div> <div class="paragraph"> <p>Such a normalized model is likely to be the approach with an RDBMS, but that’s not your only option with HBase. The cons of such an approach is that to retrieve information about any Order, you will need:</p> </div> <div class="ulist"> <ul> <li> <p>Get on the ORDER table for the Order</p> </li> <li> <p>Scan on the SHIPPING_LOCATION table for that order to get the ShippingLocation instances</p> </li> <li> <p>Scan on the LINE_ITEM for each ShippingLocation</p> </li> </ul> </div> <div class="paragraph"> <p>granted, this is what an RDBMS would do under the covers anyway, but since there are no joins in HBase you’re just more aware of this fact.</p> </div> </div> <div class="sect4"> <h5 id="schema.casestudies.custorder.obj.rectype"><a class="anchor" href="#schema.casestudies.custorder.obj.rectype"></a>Single Table With Record Types</h5> <div class="paragraph"> <p>With this approach, there would exist a single table ORDER that would contain</p> </div> <div class="paragraph"> <p>The Order rowkey was described above: <a href="#schema.casestudies.custorder">schema.casestudies.custorder</a></p> </div> <div class="ulist"> <ul> <li> <p>[order-rowkey]</p> </li> <li> <p>[ORDER record type]</p> </li> </ul> </div> <div class="paragraph"> <p>The ShippingLocation composite rowkey would be something like this:</p> </div> <div class="ulist"> <ul> <li> <p>[order-rowkey]</p> </li> <li> <p>[SHIPPING record type]</p> </li> <li> <p>[shipping location number] (e.g., 1st location, 2nd, etc.)</p> </li> </ul> </div> <div class="paragraph"> <p>The LineItem composite rowkey would be something like this:</p> </div> <div class="ulist"> <ul> <li> <p>[order-rowkey]</p> </li> <li> <p>[LINE record type]</p> </li> <li> <p>[shipping location number] (e.g., 1st location, 2nd, etc.)</p> </li> <li> <p>[line item number] (e.g., 1st lineitem, 2nd, etc.)</p> </li> </ul> </div> </div> <div class="sect4"> <h5 id="schema.casestudies.custorder.obj.denorm"><a class="anchor" href="#schema.casestudies.custorder.obj.denorm"></a>Denormalized</h5> <div class="paragraph"> <p>A variant of the Single Table With Record Types approach is to denormalize and flatten some of the object hierarchy, such as collapsing the ShippingLocation attributes onto each LineItem instance.</p> </div> <div class="paragraph"> <p>The LineItem composite rowkey would be something like this:</p> </div> <div class="ulist"> <ul> <li> <p>[order-rowkey]</p> </li> <li> <p>[LINE record type]</p> </li> <li> <p>[line item number] (e.g., 1st lineitem, 2nd, etc., care must be taken that there are unique across the entire order)</p> </li> </ul> </div> <div class="paragraph"> <p>and the LineItem columns would be something like this:</p> </div> <div class="ulist"> <ul> <li> <p>itemNumber</p> </li> <li> <p>quantity</p> </li> <li> <p>price</p> </li> <li> <p>shipToLine1 (denormalized from ShippingLocation)</p> </li> <li> <p>shipToLine2 (denormalized from ShippingLocation)</p> </li> <li> <p>shipToCity (denormalized from ShippingLocation)</p> </li> <li> <p>shipToState (denormalized from ShippingLocation)</p> </li> <li> <p>shipToZip (denormalized from ShippingLocation)</p> </li> </ul> </div> <div class="paragraph"> <p>The pros of this approach include a less complex object hierarchy, but one of the cons is that updating gets more complicated in case any of this information changes.</p> </div> </div> <div class="sect4"> <h5 id="schema.casestudies.custorder.obj.singleobj"><a class="anchor" href="#schema.casestudies.custorder.obj.singleobj"></a>Object BLOB</h5> <div class="paragraph"> <p>With this approach, the entire Order object graph is treated, in one way or another, as a BLOB. For example, the ORDER table’s rowkey was described above: <a href="#schema.casestudies.custorder">schema.casestudies.custorder</a>, and a single column called "order" would contain an object that could be deserialized that contained a container Order, ShippingLocations, and LineItems.</p> </div> <div class="paragraph"> <p>There are many options here: JSON, XML, Java Serialization, Avro, Hadoop Writables, etc. All of them are variants of the same approach: encode the object graph to a byte-array. Care should be taken with this approach to ensure backward compatibilty in case the object model changes such that older persisted structures can still be read back out of HBase.</p> </div> <div class="paragraph"> <p>Pros are being able to manage complex object graphs with minimal I/O (e.g., a single HBase Get per Order in this example), but the cons include the aforementioned warning about backward compatiblity of serialization, language dependencies of serialization (e.g., Java Serialization only works with Java clients), the fact that you have to deserialize the entire object to get any piece of information inside the BLOB, and the difficulty in getting frameworks like Hive to work with custom objects like this.</p> </div> </div> </div> </div> <div class="sect2"> <h3 id="schema.smackdown"><a class="anchor" href="#schema.smackdown"></a>43.4. Case Study - "Tall/Wide/Middle" Schema Design Smackdown</h3> <div class="paragraph"> <p>This section will describe additional schema design questions that appear on the dist-list, specifically about tall and wide tables. These are general guidelines and not laws - each application must consider its own needs.</p> </div> <div class="sect3"> <h4 id="schema.smackdown.rowsversions"><a class="anchor" href="#schema.smackdown.rowsversions"></a>43.4.1. Rows vs. Versions</h4> <div class="paragraph"> <p>A common question is whether one should prefer rows or HBase’s built-in-versioning. The context is typically where there are "a lot" of versions of a row to be retained (e.g., where it is significantly above the HBase default of 1 max versions). The rows-approach would require storing a timestamp in some portion of the rowkey so that they would not overwite with each successive update.</p> </div> <div class="paragraph"> <p>Preference: Rows (generally speaking).</p> </div> </div> <div class="sect3"> <h4 id="schema.smackdown.rowscols"><a class="anchor" href="#schema.smackdown.rowscols"></a>43.4.2. Rows vs. Columns</h4> <div class="paragraph"> <p>Another common question is whether one should prefer rows or columns. The context is typically in extreme cases of wide tables, such as having 1 row with 1 million attributes, or 1 million rows with 1 columns apiece.</p> </div> <div class="paragraph"> <p>Preference: Rows (generally speaking). To be clear, this guideline is in the context is in extremely wide cases, not in the standard use-case where one needs to store a few dozen or hundred columns. But there is also a middle path between these two options, and that is "Rows as Columns."</p> </div> </div> <div class="sect3"> <h4 id="schema.smackdown.rowsascols"><a class="anchor" href="#schema.smackdown.rowsascols"></a>43.4.3. Rows as Columns</h4> <div class="paragraph"> <p>The middle path between Rows vs. Columns is packing data that would be a separate row into columns, for certain rows. OpenTSDB is the best example of this case where a single row represents a defined time-range, and then discrete events are treated as columns. This approach is often more complex, and may require the additional complexity of re-writing your data, but has the advantage of being I/O efficient. For an overview of this approach, see <a href="#schema.casestudies.log_steroids">schema.casestudies.log-steroids</a>.</p> </div> </div> </div> <div class="sect2"> <h3 id="casestudies.schema.listdata"><a class="anchor" href="#casestudies.schema.listdata"></a>43.5. Case Study - List Data</h3> <div class="paragraph"> <p>The following is an exchange from the user dist-list regarding a fairly common question: how to handle per-user list data in Apache HBase.</p> </div> <div class="ulist"> <ul> <li> <p>QUESTION <strong>*</strong></p> </li> </ul> </div> <div class="paragraph"> <p>We’re looking at how to store a large amount of (per-user) list data in HBase, and we were trying to figure out what kind of access pattern made the most sense. One option is store the majority of the data in a key, so we could have something like:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><FixedWidthUserName><FixedWidthValueId1>:<span class="string"><span class="delimiter">"</span><span class="delimiter">"</span></span> (no value) <FixedWidthUserName><FixedWidthValueId2>:<span class="string"><span class="delimiter">"</span><span class="delimiter">"</span></span> (no value) <FixedWidthUserName><FixedWidthValueId3>:<span class="string"><span class="delimiter">"</span><span class="delimiter">"</span></span> (no value)</code></pre> </div> </div> <div class="paragraph"> <p>The other option we had was to do this entirely using:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><FixedWidthUserName></span><span class="tag"><FixedWidthPageNum0></span>:<span class="tag"><FixedWidthLength></span><span class="tag"><FixedIdNextPageNum></span><span class="tag"><ValueId1></span><span class="tag"><ValueId2></span><span class="tag"><ValueId3></span>... <span class="tag"><FixedWidthUserName></span><span class="tag"><FixedWidthPageNum1></span>:<span class="tag"><FixedWidthLength></span><span class="tag"><FixedIdNextPageNum></span><span class="tag"><ValueId1></span><span class="tag"><ValueId2></span><span class="tag"><ValueId3></span>...</code></pre> </div> </div> <div class="paragraph"> <p>where each row would contain multiple values. So in one case reading the first thirty values would be:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">scan { STARTROW => <span class="string"><span class="delimiter">'</span><span class="content">FixedWidthUsername</span><span class="delimiter">'</span></span> LIMIT => <span class="integer">30</span>}</code></pre> </div> </div> <div class="paragraph"> <p>And in the second case it would be</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">get <span class="string"><span class="delimiter">'</span><span class="content">FixedWidthUserName</span><span class="content">\x00</span><span class="content">\x00</span><span class="content">\x00</span><span class="content">\x00</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>The general usage pattern would be to read only the first 30 values of these lists, with infrequent access reading deeper into the lists. Some users would have ⇐ 30 total values in these lists, and some users would have millions (i.e. power-law distribution)</p> </div> <div class="paragraph"> <p>The single-value format seems like it would take up more space on HBase, but would offer some improved retrieval / pagination flexibility. Would there be any significant performance advantages to be able to paginate via gets vs paginating with scans?</p> </div> <div class="paragraph"> <p>My initial understanding was that doing a scan should be faster if our paging size is unknown (and caching is set appropriately), but that gets should be faster if we’ll always need the same page size. I’ve ended up hearing different people tell me opposite things about performance. I assume the page sizes would be relatively consistent, so for most use cases we could guarantee that we only wanted one page of data in the fixed-page-length case. I would also assume that we would have infrequent updates, but may have inserts into the middle of these lists (meaning we’d need to update all subsequent rows).</p> </div> <div class="paragraph"> <p>Thanks for help / suggestions / follow-up questions.</p> </div> <div class="ulist"> <ul> <li> <p>ANSWER <strong>*</strong></p> </li> </ul> </div> <div class="paragraph"> <p>If I understand you correctly, you’re ultimately trying to store triples in the form "user, valueid, value", right? E.g., something like:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">user123, firstname, Paul</span><span class="delimiter">"</span></span>, <span class="string"><span class="delimiter">"</span><span class="content">user234, lastname, Smith</span><span class="delimiter">"</span></span></code></pre> </div> </div> <div class="paragraph"> <p>(But the usernames are fixed width, and the valueids are fixed width).</p> </div> <div class="paragraph"> <p>And, your access pattern is along the lines of: "for user X, list the next 30 values, starting with valueid Y". Is that right? And these values should be returned sorted by valueid?</p> </div> <div class="paragraph"> <p>The tl;dr version is that you should probably go with one row per user+value, and not build a complicated intra-row pagination scheme on your own unless you’re really sure it is needed.</p> </div> <div class="paragraph"> <p>Your two options mirror a common question people have when designing HBase schemas: should I go "tall" or "wide"? Your first schema is "tall": each row represents one value for one user, and so there are many rows in the table for each user; the row key is user + valueid, and there would be (presumably) a single column qualifier that means "the value". This is great if you want to scan over rows in sorted order by row key (thus my question above, about whether these ids are sorted correctly). You can start a scan at any user+valueid, read the next 30, and be done. What you’re giving up is the ability to have transactional guarantees around all the rows for one user, but it doesn’t sound like you need that. Doing it this way is generally recommended (see here link:http://hbase.apache.org/book.html#schema.smackdown).</p> </div> <div class="paragraph"> <p>Your second option is "wide": you store a bunch of values in one row, using different qualifiers (where the qualifier is the valueid). The simple way to do that would be to just store ALL values for one user in a single row. I’m guessing you jumped to the "paginated" version because you’re assuming that storing millions of columns in a single row would be bad for performance, which may or may not be true; as long as you’re not trying to do too much in a single request, or do things like scanning over and returning all of the cells in the row, it shouldn’t be fundamentally worse. The client has methods that allow you to get specific slices of columns.</p> </div> <div class="paragraph"> <p>Note that neither case fundamentally uses more disk space than the other; you’re just "shifting" part of the identifying information for a value either to the left (into the row key, in option one) or to the right (into the column qualifiers in option 2). Under the covers, every key/value still stores the whole row key, and column family name. (If this is a bit confusing, take an hour and watch Lars George’s excellent video about understanding HBase schema design: link:http://www.youtube.com/watch?v=_HLoH_PgrLk).</p> </div> <div class="paragraph"> <p>A manually paginated version has lots more complexities, as you note, like having to keep track of how many things are in each page, re-shuffling if new values are inserted, etc. That seems significantly more complex. It might have some slight speed advantages (or disadvantages!) at extremely high throughput, and the only way to really know that would be to try it out. If you don’t have time to build it both ways and compare, my advice would be to start with the simplest option (one row per user+value). Start simple and iterate! :)</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="schema.ops"><a class="anchor" href="#schema.ops"></a>44. Operational and Performance Configuration Options</h2> <div class="sectionbody"> <div class="paragraph"> <p>See the Performance section <a href="#perf.schema">perf.schema</a> for more information operational and performance schema design options, such as Bloom Filters, Table-configured regionsizes, compression, and blocksizes.</p> </div> </div> </div> <h1 id="mapreduce" class="sect0"><a class="anchor" href="#mapreduce"></a>HBase and MapReduce</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>Apache MapReduce is a software framework used to analyze large amounts of data, and is the framework used most often with <a href="http://hadoop.apache.org/">Apache Hadoop</a>. MapReduce itself is out of the scope of this document. A good place to get started with MapReduce is <a href="http://hadoop.apache.org/docs/r2.6.0/hadoop-mapreduce-client/hadoop-mapreduce-client-core/MapReduceTutorial.html" class="bare">http://hadoop.apache.org/docs/r2.6.0/hadoop-mapreduce-client/hadoop-mapreduce-client-core/MapReduceTutorial.html</a>. MapReduce version 2 (MR2)is now part of <a href="http://hadoop.apache.org/docs/r2.3.0/hadoop-yarn/hadoop-yarn-site/">YARN</a>.</p> </div> <div class="paragraph"> <p>This chapter discusses specific configuration steps you need to take to use MapReduce on data within HBase. In addition, it discusses other interactions and issues between HBase and MapReduce jobs.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title"><code>mapred</code> and <code>mapreduce</code></div> <div class="paragraph"> <p>There are two mapreduce packages in HBase as in MapReduce itself: <em>org.apache.hadoop.hbase.mapred</em> and <em>org.apache.hadoop.hbase.mapreduce</em>. The former does old-style API and the latter the new style. The latter has more facility though you can usually find an equivalent in the older package. Pick the package that goes with your MapReduce deploy. When in doubt or starting over, pick the <em>org.apache.hadoop.hbase.mapreduce</em>. In the notes below, we refer to o.a.h.h.mapreduce but replace with the o.a.h.h.mapred if that is what you are using.</p> </div> </td> </tr> </table> </div> </div> </div> <div class="sect1"> <h2 id="hbase.mapreduce.classpath"><a class="anchor" href="#hbase.mapreduce.classpath"></a>45. HBase, MapReduce, and the CLASSPATH</h2> <div class="sectionbody"> <div class="paragraph"> <p>By default, MapReduce jobs deployed to a MapReduce cluster do not have access to either the HBase configuration under <code>$HBASE_CONF_DIR</code> or the HBase classes.</p> </div> <div class="paragraph"> <p>To give the MapReduce jobs the access they need, you could add <em>hbase-site.xml</em> to <em>$HADOOP_HOME/conf</em> and add HBase jars to the <em>$HADOOP_HOME/lib</em> directory. You would then need to copy these changes across your cluster. Or you can edit <em>$HADOOP_HOME/conf/hadoop-env.sh</em> and add them to the <code>HADOOP_CLASSPATH</code> variable. However, this approach is not recommended because it will pollute your Hadoop install with HBase references. It also requires you to restart the Hadoop cluster before Hadoop can use the HBase data.</p> </div> <div class="paragraph"> <p>The recommended approach is to let HBase add its dependency jars itself and use <code>HADOOP_CLASSPATH</code> or <code>-libjars</code>.</p> </div> <div class="paragraph"> <p>Since HBase 0.90.x, HBase adds its dependency JARs to the job configuration itself. The dependencies only need to be available on the local <code>CLASSPATH</code>. The following example runs the bundled HBase <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/RowCounter.html">RowCounter</a> MapReduce job against a table named <code>usertable</code>. If you have not set the environment variables expected in the command (the parts prefixed by a <code>$</code> sign and surrounded by curly braces), you can use the actual system paths instead. Be sure to use the correct version of the HBase JAR for your system. The backticks (<code>`</code> symbols) cause ths shell to execute the sub-commands, setting the output of <code>hbase classpath</code> (the command to dump HBase CLASSPATH) to <code>HADOOP_CLASSPATH</code>. This example assumes you use a BASH-compatible shell.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_CLASSPATH=`${HBASE_HOME}/bin/hbase classpath` ${HADOOP_HOME}/bin/hadoop jar ${HBASE_HOME}/lib/hbase-server-VERSION.jar rowcounter usertable</code></pre> </div> </div> <div class="paragraph"> <p>When the command runs, internally, the HBase JAR finds the dependencies it needs and adds them to the MapReduce job configuration. See the source at <code>TableMapReduceUtil#addDependencyJars(org.apache.hadoop.mapreduce.Job)</code> for how this is done.</p> </div> <div class="paragraph"> <p>The command <code>hbase mapredcp</code> can also help you dump the CLASSPATH entries required by MapReduce, which are the same jars <code>TableMapReduceUtil#addDependencyJars</code> would add. You can add them together with HBase conf directory to <code>HADOOP_CLASSPATH</code>. For jobs that do not package their dependencies or call <code>TableMapReduceUtil#addDependencyJars</code>, the following command structure is necessary:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_CLASSPATH=`${HBASE_HOME}/bin/hbase mapredcp`:${HBASE_HOME}/conf hadoop jar MyApp.jar MyJobMainClass -libjars $(${HBASE_HOME}/bin/hbase mapredcp | tr ':' ',') ...</code></pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="paragraph"> <p>The example may not work if you are running HBase from its build directory rather than an installed location. You may see an error like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>java.lang.RuntimeException: java.lang.ClassNotFoundException: org.apache.hadoop.hbase.mapreduce.RowCounter$RowCounterMapper</pre> </div> </div> <div class="paragraph"> <p>If this occurs, try modifying the command as follows, so that it uses the HBase JARs from the <em>target/</em> directory within the build environment.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_CLASSPATH=${HBASE_BUILD_HOME}/hbase-server/target/hbase-server-VERSION-SNAPSHOT.jar:`${HBASE_BUILD_HOME}/bin/hbase classpath` ${HADOOP_HOME}/bin/hadoop jar ${HBASE_BUILD_HOME}/hbase-server/target/hbase-server-VERSION-SNAPSHOT.jar rowcounter usertable</code></pre> </div> </div> </td> </tr> </table> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> <div class="title">Notice to MapReduce users of HBase between 0.96.1 and 0.98.4</div> <div class="paragraph"> <p>Some MapReduce jobs that use HBase fail to launch. The symptom is an exception similar to the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>Exception in thread "main" java.lang.IllegalAccessError: class com.google.protobuf.ZeroCopyLiteralByteString cannot access its superclass com.google.protobuf.LiteralByteString at java.lang.ClassLoader.defineClass1(Native Method) at java.lang.ClassLoader.defineClass(ClassLoader.java:792) at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:142) at java.net.URLClassLoader.defineClass(URLClassLoader.java:449) at java.net.URLClassLoader.access$100(URLClassLoader.java:71) at java.net.URLClassLoader$1.run(URLClassLoader.java:361) at java.net.URLClassLoader$1.run(URLClassLoader.java:355) at java.security.AccessController.doPrivileged(Native Method) at java.net.URLClassLoader.findClass(URLClassLoader.java:354) at java.lang.ClassLoader.loadClass(ClassLoader.java:424) at java.lang.ClassLoader.loadClass(ClassLoader.java:357) at org.apache.hadoop.hbase.protobuf.ProtobufUtil.toScan(ProtobufUtil.java:818) at org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.convertScanToString(TableMapReduceUtil.java:433) at org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:186) at org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:147) at org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:270) at org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.initTableMapperJob(TableMapReduceUtil.java:100) ...</pre> </div> </div> <div class="paragraph"> <p>This is caused by an optimization introduced in <a href="https://issues.apache.org/jira/browse/HBASE-9867">HBASE-9867</a> that inadvertently introduced a classloader dependency.</p> </div> <div class="paragraph"> <p>This affects both jobs using the <code>-libjars</code> option and "fat jar," those which package their runtime dependencies in a nested <code>lib</code> folder.</p> </div> <div class="paragraph"> <p>In order to satisfy the new classloader requirements, <code>hbase-protocol.jar</code> must be included in Hadoop’s classpath. See <a href="#hbase.mapreduce.classpath">HBase, MapReduce, and the CLASSPATH</a> for current recommendations for resolving classpath errors. The following is included for historical purposes.</p> </div> <div class="paragraph"> <p>This can be resolved system-wide by including a reference to the <code>hbase-protocol.jar</code> in Hadoop’s lib directory, via a symlink or by copying the jar into the new location.</p> </div> <div class="paragraph"> <p>This can also be achieved on a per-job launch basis by including it in the <code>HADOOP_CLASSPATH</code> environment variable at job submission time. When launching jobs that package their dependencies, all three of the following job launching commands satisfy this requirement:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_CLASSPATH=/path/to/hbase-protocol.jar:/path/to/hbase/conf hadoop jar MyJob.jar MyJobMainClass $ HADOOP_CLASSPATH=$(hbase mapredcp):/path/to/hbase/conf hadoop jar MyJob.jar MyJobMainClass $ HADOOP_CLASSPATH=$(hbase classpath) hadoop jar MyJob.jar MyJobMainClass</code></pre> </div> </div> <div class="paragraph"> <p>For jars that do not package their dependencies, the following command structure is necessary:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_CLASSPATH=$(hbase mapredcp):/etc/hbase/conf hadoop jar MyApp.jar MyJobMainClass -libjars $(hbase mapredcp | tr ':' ',') ...</code></pre> </div> </div> <div class="paragraph"> <p>See also <a href="https://issues.apache.org/jira/browse/HBASE-10304">HBASE-10304</a> for further discussion of this issue.</p> </div> </td> </tr> </table> </div> </div> </div> <div class="sect1"> <h2 id="_mapreduce_scan_caching"><a class="anchor" href="#_mapreduce_scan_caching"></a>46. MapReduce Scan Caching</h2> <div class="sectionbody"> <div class="paragraph"> <p>TableMapReduceUtil now restores the option to set scanner caching (the number of rows which are cached before returning the result to the client) on the Scan object that is passed in. This functionality was lost due to a bug in HBase 0.95 (<a href="https://issues.apache.org/jira/browse/HBASE-11558">HBASE-11558</a>), which is fixed for HBase 0.98.5 and 0.96.3. The priority order for choosing the scanner caching is as follows:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Caching settings which are set on the scan object.</p> </li> <li> <p>Caching settings which are specified via the configuration option <code>hbase.client.scanner.caching</code>, which can either be set manually in <em>hbase-site.xml</em> or via the helper method <code>TableMapReduceUtil.setScannerCaching()</code>.</p> </li> <li> <p>The default value <code>HConstants.DEFAULT_HBASE_CLIENT_SCANNER_CACHING</code>, which is set to <code>100</code>.</p> </li> </ol> </div> <div class="paragraph"> <p>Optimizing the caching settings is a balance between the time the client waits for a result and the number of sets of results the client needs to receive. If the caching setting is too large, the client could end up waiting for a long time or the request could even time out. If the setting is too small, the scan needs to return results in several pieces. If you think of the scan as a shovel, a bigger cache setting is analogous to a bigger shovel, and a smaller cache setting is equivalent to more shoveling in order to fill the bucket.</p> </div> <div class="paragraph"> <p>The list of priorities mentioned above allows you to set a reasonable default, and override it for specific operations.</p> </div> <div class="paragraph"> <p>See the API documentation for <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</a> for more details.</p> </div> </div> </div> <div class="sect1"> <h2 id="_bundled_hbase_mapreduce_jobs"><a class="anchor" href="#_bundled_hbase_mapreduce_jobs"></a>47. Bundled HBase MapReduce Jobs</h2> <div class="sectionbody"> <div class="paragraph"> <p>The HBase JAR also serves as a Driver for some bundled MapReduce jobs. To learn about the bundled MapReduce jobs, run the following command.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ${HADOOP_HOME}/bin/hadoop jar ${HBASE_HOME}/hbase-server-VERSION.jar An example program must be given as the first argument. Valid program names are: copytable: Export a table from local cluster to peer cluster completebulkload: Complete a bulk data load. export: Write table data to HDFS. import: Import data written by Export. importtsv: Import data in TSV format. rowcounter: Count rows in HBase table</code></pre> </div> </div> <div class="paragraph"> <p>Each of the valid program names are bundled MapReduce jobs. To run one of the jobs, model your command after the following example.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ${HADOOP_HOME}/bin/hadoop jar ${HBASE_HOME}/hbase-server-VERSION.jar rowcounter myTable</code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="_hbase_as_a_mapreduce_job_data_source_and_data_sink"><a class="anchor" href="#_hbase_as_a_mapreduce_job_data_source_and_data_sink"></a>48. HBase as a MapReduce Job Data Source and Data Sink</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase can be used as a data source, <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html">TableInputFormat</a>, and data sink, <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html">TableOutputFormat</a> or <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/MultiTableOutputFormat.html">MultiTableOutputFormat</a>, for MapReduce jobs. Writing MapReduce jobs that read or write HBase, it is advisable to subclass <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableMapper.html">TableMapper</a> and/or <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableReducer.html">TableReducer</a>. See the do-nothing pass-through classes <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/IdentityTableMapper.html">IdentityTableMapper</a> and <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/IdentityTableReducer.html">IdentityTableReducer</a> for basic usage. For a more involved example, see <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/RowCounter.html">RowCounter</a> or review the <code>org.apache.hadoop.hbase.mapreduce.TestTableMapReduce</code> unit test.</p> </div> <div class="paragraph"> <p>If you run MapReduce jobs that use HBase as source or sink, need to specify source and sink table and column names in your configuration.</p> </div> <div class="paragraph"> <p>When you read from HBase, the <code>TableInputFormat</code> requests the list of regions from HBase and makes a map, which is either a <code>map-per-region</code> or <code>mapreduce.job.maps</code> map, whichever is smaller. If your job only has two maps, raise <code>mapreduce.job.maps</code> to a number greater than the number of regions. Maps will run on the adjacent TaskTracker/NodeManager if you are running a TaskTracer/NodeManager and RegionServer per node. When writing to HBase, it may make sense to avoid the Reduce step and write back into HBase from within your map. This approach works when your job does not need the sort and collation that MapReduce does on the map-emitted data. On insert, HBase 'sorts' so there is no point double-sorting (and shuffling data around your MapReduce cluster) unless you need to. If you do not need the Reduce, your map might emit counts of records processed for reporting at the end of the job, or set the number of Reduces to zero and use TableOutputFormat. If running the Reduce step makes sense in your case, you should typically use multiple reducers so that load is spread across the HBase cluster.</p> </div> <div class="paragraph"> <p>A new HBase partitioner, the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/HRegionPartitioner.html">HRegionPartitioner</a>, can run as many reducers the number of existing regions. The HRegionPartitioner is suitable when your table is large and your upload will not greatly alter the number of existing regions upon completion. Otherwise use the default partitioner.</p> </div> </div> </div> <div class="sect1"> <h2 id="_writing_hfiles_directly_during_bulk_import"><a class="anchor" href="#_writing_hfiles_directly_during_bulk_import"></a>49. Writing HFiles Directly During Bulk Import</h2> <div class="sectionbody"> <div class="paragraph"> <p>If you are importing into a new table, you can bypass the HBase API and write your content directly to the filesystem, formatted into HBase data files (HFiles). Your import will run faster, perhaps an order of magnitude faster. For more on how this mechanism works, see <a href="#arch.bulk.load">Bulk Loading</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_rowcounter_example"><a class="anchor" href="#_rowcounter_example"></a>50. RowCounter Example</h2> <div class="sectionbody"> <div class="paragraph"> <p>The included <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/RowCounter.html">RowCounter</a> MapReduce job uses <code>TableInputFormat</code> and does a count of all rows in the specified table. To run it, use the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ./bin/hadoop jar hbase-X.X.X.jar</code></pre> </div> </div> <div class="paragraph"> <p>This will invoke the HBase MapReduce Driver class. Select <code>rowcounter</code> from the choice of jobs offered. This will print rowcounter usage advice to standard output. Specify the tablename, column to count, and output directory. If you have classpath errors, see <a href="#hbase.mapreduce.classpath">HBase, MapReduce, and the CLASSPATH</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="splitter"><a class="anchor" href="#splitter"></a>51. Map-Task Splitting</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="splitter.default"><a class="anchor" href="#splitter.default"></a>51.1. The Default HBase MapReduce Splitter</h3> <div class="paragraph"> <p>When <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableInputFormat.html">TableInputFormat</a> is used to source an HBase table in a MapReduce job, its splitter will make a map task for each region of the table. Thus, if there are 100 regions in the table, there will be 100 map-tasks for the job - regardless of how many column families are selected in the Scan.</p> </div> </div> <div class="sect2"> <h3 id="splitter.custom"><a class="anchor" href="#splitter.custom"></a>51.2. Custom Splitters</h3> <div class="paragraph"> <p>For those interested in implementing custom splitters, see the method <code>getSplits</code> in <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.html">TableInputFormatBase</a>. That is where the logic for map-task assignment resides.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="mapreduce.example"><a class="anchor" href="#mapreduce.example"></a>52. HBase MapReduce Examples</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="mapreduce.example.read"><a class="anchor" href="#mapreduce.example.read"></a>52.1. HBase MapReduce Read Example</h3> <div class="paragraph"> <p>The following is an example of using HBase as a MapReduce source in read-only manner. Specifically, there is a Mapper instance but no Reducer, and nothing is being emitted from the Mapper. There job would be defined as follows…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); Job job = <span class="keyword">new</span> Job(config, <span class="string"><span class="delimiter">"</span><span class="content">ExampleRead</span><span class="delimiter">"</span></span>); job.setJarByClass(MyReadJob.class); <span class="comment">// class that contains mapper</span> Scan scan = <span class="keyword">new</span> Scan(); scan.setCaching(<span class="integer">500</span>); <span class="comment">// 1 is the default in Scan, which will be bad for MapReduce jobs</span> scan.setCacheBlocks(<span class="predefined-constant">false</span>); <span class="comment">// don't set to true for MR jobs</span> <span class="comment">// set other scan attrs</span> ... TableMapReduceUtil.initTableMapperJob( tableName, <span class="comment">// input HBase table name</span> scan, <span class="comment">// Scan instance to control CF and attribute selection</span> MyMapper.class, <span class="comment">// mapper</span> <span class="predefined-constant">null</span>, <span class="comment">// mapper output key</span> <span class="predefined-constant">null</span>, <span class="comment">// mapper output value</span> job); job.setOutputFormatClass(NullOutputFormat.class); <span class="comment">// because we aren't emitting anything from mapper</span> <span class="type">boolean</span> b = job.waitForCompletion(<span class="predefined-constant">true</span>); <span class="keyword">if</span> (!b) { <span class="keyword">throw</span> <span class="keyword">new</span> <span class="exception">IOException</span>(<span class="string"><span class="delimiter">"</span><span class="content">error with job!</span><span class="delimiter">"</span></span>); }</code></pre> </div> </div> <div class="paragraph"> <p>…and the mapper instance would extend <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableMapper.html">TableMapper</a>…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">class</span> <span class="class">MyMapper</span> <span class="directive">extends</span> TableMapper<Text, Text> { <span class="directive">public</span> <span class="type">void</span> map(ImmutableBytesWritable row, <span class="predefined-type">Result</span> value, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">InterruptedException</span>, <span class="exception">IOException</span> { <span class="comment">// process data for the row from the Result instance.</span> } }</code></pre> </div> </div> </div> <div class="sect2"> <h3 id="mapreduce.example.readwrite"><a class="anchor" href="#mapreduce.example.readwrite"></a>52.2. HBase MapReduce Read/Write Example</h3> <div class="paragraph"> <p>The following is an example of using HBase both as a source and as a sink with MapReduce. This example will simply copy data from one table to another.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); Job job = <span class="keyword">new</span> Job(config,<span class="string"><span class="delimiter">"</span><span class="content">ExampleReadWrite</span><span class="delimiter">"</span></span>); job.setJarByClass(MyReadWriteJob.class); <span class="comment">// class that contains mapper</span> Scan scan = <span class="keyword">new</span> Scan(); scan.setCaching(<span class="integer">500</span>); <span class="comment">// 1 is the default in Scan, which will be bad for MapReduce jobs</span> scan.setCacheBlocks(<span class="predefined-constant">false</span>); <span class="comment">// don't set to true for MR jobs</span> <span class="comment">// set other scan attrs</span> TableMapReduceUtil.initTableMapperJob( sourceTable, <span class="comment">// input table</span> scan, <span class="comment">// Scan instance to control CF and attribute selection</span> MyMapper.class, <span class="comment">// mapper class</span> <span class="predefined-constant">null</span>, <span class="comment">// mapper output key</span> <span class="predefined-constant">null</span>, <span class="comment">// mapper output value</span> job); TableMapReduceUtil.initTableReducerJob( targetTable, <span class="comment">// output table</span> <span class="predefined-constant">null</span>, <span class="comment">// reducer class</span> job); job.setNumReduceTasks(<span class="integer">0</span>); <span class="type">boolean</span> b = job.waitForCompletion(<span class="predefined-constant">true</span>); <span class="keyword">if</span> (!b) { <span class="keyword">throw</span> <span class="keyword">new</span> <span class="exception">IOException</span>(<span class="string"><span class="delimiter">"</span><span class="content">error with job!</span><span class="delimiter">"</span></span>); }</code></pre> </div> </div> <div class="paragraph"> <p>An explanation is required of what <code>TableMapReduceUtil</code> is doing, especially with the reducer. <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html">TableOutputFormat</a> is being used as the outputFormat class, and several parameters are being set on the config (e.g., <code>TableOutputFormat.OUTPUT_TABLE</code>), as well as setting the reducer output key to <code>ImmutableBytesWritable</code> and reducer value to <code>Writable</code>. These could be set by the programmer on the job and conf, but <code>TableMapReduceUtil</code> tries to make things easier.</p> </div> <div class="paragraph"> <p>The following is the example mapper, which will create a <code>Put</code> and matching the input <code>Result</code> and emit it. Note: this is what the CopyTable utility does.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">class</span> <span class="class">MyMapper</span> <span class="directive">extends</span> TableMapper<ImmutableBytesWritable, Put> { <span class="directive">public</span> <span class="type">void</span> map(ImmutableBytesWritable row, <span class="predefined-type">Result</span> value, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="comment">// this example is just copying the data from the source table...</span> context.write(row, resultToPut(row,value)); } <span class="directive">private</span> <span class="directive">static</span> Put resultToPut(ImmutableBytesWritable key, <span class="predefined-type">Result</span> result) <span class="directive">throws</span> <span class="exception">IOException</span> { Put put = <span class="keyword">new</span> Put(key.get()); <span class="keyword">for</span> (KeyValue kv : result.raw()) { put.add(kv); } <span class="keyword">return</span> put; } }</code></pre> </div> </div> <div class="paragraph"> <p>There isn’t actually a reducer step, so <code>TableOutputFormat</code> takes care of sending the <code>Put</code> to the target table.</p> </div> <div class="paragraph"> <p>This is just an example, developers could choose not to use <code>TableOutputFormat</code> and connect to the target table themselves.</p> </div> </div> <div class="sect2"> <h3 id="mapreduce.example.readwrite.multi"><a class="anchor" href="#mapreduce.example.readwrite.multi"></a>52.3. HBase MapReduce Read/Write Example With Multi-Table Output</h3> <div class="paragraph"> <p>TODO: example for <code>MultiTableOutputFormat</code>.</p> </div> </div> <div class="sect2"> <h3 id="mapreduce.example.summary"><a class="anchor" href="#mapreduce.example.summary"></a>52.4. HBase MapReduce Summary to HBase Example</h3> <div class="paragraph"> <p>The following example uses HBase as a MapReduce source and sink with a summarization step. This example will count the number of distinct instances of a value in a table and write those summarized counts in another table.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); Job job = <span class="keyword">new</span> Job(config,<span class="string"><span class="delimiter">"</span><span class="content">ExampleSummary</span><span class="delimiter">"</span></span>); job.setJarByClass(MySummaryJob.class); <span class="comment">// class that contains mapper and reducer</span> Scan scan = <span class="keyword">new</span> Scan(); scan.setCaching(<span class="integer">500</span>); <span class="comment">// 1 is the default in Scan, which will be bad for MapReduce jobs</span> scan.setCacheBlocks(<span class="predefined-constant">false</span>); <span class="comment">// don't set to true for MR jobs</span> <span class="comment">// set other scan attrs</span> TableMapReduceUtil.initTableMapperJob( sourceTable, <span class="comment">// input table</span> scan, <span class="comment">// Scan instance to control CF and attribute selection</span> MyMapper.class, <span class="comment">// mapper class</span> Text.class, <span class="comment">// mapper output key</span> IntWritable.class, <span class="comment">// mapper output value</span> job); TableMapReduceUtil.initTableReducerJob( targetTable, <span class="comment">// output table</span> MyTableReducer.class, <span class="comment">// reducer class</span> job); job.setNumReduceTasks(<span class="integer">1</span>); <span class="comment">// at least one, adjust as required</span> <span class="type">boolean</span> b = job.waitForCompletion(<span class="predefined-constant">true</span>); <span class="keyword">if</span> (!b) { <span class="keyword">throw</span> <span class="keyword">new</span> <span class="exception">IOException</span>(<span class="string"><span class="delimiter">"</span><span class="content">error with job!</span><span class="delimiter">"</span></span>); }</code></pre> </div> </div> <div class="paragraph"> <p>In this example mapper a column with a String-value is chosen as the value to summarize upon. This value is used as the key to emit from the mapper, and an <code>IntWritable</code> represents an instance counter.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">class</span> <span class="class">MyMapper</span> <span class="directive">extends</span> TableMapper<Text, IntWritable> { <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR1 = <span class="string"><span class="delimiter">"</span><span class="content">attr1</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">private</span> <span class="directive">final</span> IntWritable ONE = <span class="keyword">new</span> IntWritable(<span class="integer">1</span>); <span class="directive">private</span> Text text = <span class="keyword">new</span> Text(); <span class="directive">public</span> <span class="type">void</span> map(ImmutableBytesWritable row, <span class="predefined-type">Result</span> value, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="predefined-type">String</span> val = <span class="keyword">new</span> <span class="predefined-type">String</span>(value.getValue(CF, ATTR1)); text.set(val); <span class="comment">// we can only emit Writables...</span> context.write(text, ONE); } }</code></pre> </div> </div> <div class="paragraph"> <p>In the reducer, the "ones" are counted (just like any other MR example that does this), and then emits a <code>Put</code>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">class</span> <span class="class">MyTableReducer</span> <span class="directive">extends</span> TableReducer<Text, IntWritable, ImmutableBytesWritable> { <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> COUNT = <span class="string"><span class="delimiter">"</span><span class="content">count</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="type">void</span> reduce(Text key, <span class="predefined-type">Iterable</span><IntWritable> values, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="type">int</span> i = <span class="integer">0</span>; <span class="keyword">for</span> (IntWritable val : values) { i += val.get(); } Put put = <span class="keyword">new</span> Put(Bytes.toBytes(key.toString())); put.add(CF, COUNT, Bytes.toBytes(i)); context.write(<span class="predefined-constant">null</span>, put); } }</code></pre> </div> </div> </div> <div class="sect2"> <h3 id="mapreduce.example.summary.file"><a class="anchor" href="#mapreduce.example.summary.file"></a>52.5. HBase MapReduce Summary to File Example</h3> <div class="paragraph"> <p>This very similar to the summary example above, with exception that this is using HBase as a MapReduce source but HDFS as the sink. The differences are in the job setup and in the reducer. The mapper remains the same.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); Job job = <span class="keyword">new</span> Job(config,<span class="string"><span class="delimiter">"</span><span class="content">ExampleSummaryToFile</span><span class="delimiter">"</span></span>); job.setJarByClass(MySummaryFileJob.class); <span class="comment">// class that contains mapper and reducer</span> Scan scan = <span class="keyword">new</span> Scan(); scan.setCaching(<span class="integer">500</span>); <span class="comment">// 1 is the default in Scan, which will be bad for MapReduce jobs</span> scan.setCacheBlocks(<span class="predefined-constant">false</span>); <span class="comment">// don't set to true for MR jobs</span> <span class="comment">// set other scan attrs</span> TableMapReduceUtil.initTableMapperJob( sourceTable, <span class="comment">// input table</span> scan, <span class="comment">// Scan instance to control CF and attribute selection</span> MyMapper.class, <span class="comment">// mapper class</span> Text.class, <span class="comment">// mapper output key</span> IntWritable.class, <span class="comment">// mapper output value</span> job); job.setReducerClass(MyReducer.class); <span class="comment">// reducer class</span> job.setNumReduceTasks(<span class="integer">1</span>); <span class="comment">// at least one, adjust as required</span> FileOutputFormat.setOutputPath(job, <span class="keyword">new</span> Path(<span class="string"><span class="delimiter">"</span><span class="content">/tmp/mr/mySummaryFile</span><span class="delimiter">"</span></span>)); <span class="comment">// adjust directories as required</span> <span class="type">boolean</span> b = job.waitForCompletion(<span class="predefined-constant">true</span>); <span class="keyword">if</span> (!b) { <span class="keyword">throw</span> <span class="keyword">new</span> <span class="exception">IOException</span>(<span class="string"><span class="delimiter">"</span><span class="content">error with job!</span><span class="delimiter">"</span></span>); }</code></pre> </div> </div> <div class="paragraph"> <p>As stated above, the previous Mapper can run unchanged with this example. As for the Reducer, it is a "generic" Reducer instead of extending TableMapper and emitting Puts.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">class</span> <span class="class">MyReducer</span> <span class="directive">extends</span> Reducer<Text, IntWritable, Text, IntWritable> { <span class="directive">public</span> <span class="type">void</span> reduce(Text key, <span class="predefined-type">Iterable</span><IntWritable> values, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="type">int</span> i = <span class="integer">0</span>; <span class="keyword">for</span> (IntWritable val : values) { i += val.get(); } context.write(key, <span class="keyword">new</span> IntWritable(i)); } }</code></pre> </div> </div> </div> <div class="sect2"> <h3 id="mapreduce.example.summary.noreducer"><a class="anchor" href="#mapreduce.example.summary.noreducer"></a>52.6. HBase MapReduce Summary to HBase Without Reducer</h3> <div class="paragraph"> <p>It is also possible to perform summaries without a reducer - if you use HBase as the reducer.</p> </div> <div class="paragraph"> <p>An HBase target table would need to exist for the job summary. The Table method <code>incrementColumnValue</code> would be used to atomically increment values. From a performance perspective, it might make sense to keep a Map of values with their values to be incremented for each map-task, and make one update per key at during the <code>cleanup</code> method of the mapper. However, your mileage may vary depending on the number of rows to be processed and unique keys.</p> </div> <div class="paragraph"> <p>In the end, the summary results are in HBase.</p> </div> </div> <div class="sect2"> <h3 id="mapreduce.example.summary.rdbms"><a class="anchor" href="#mapreduce.example.summary.rdbms"></a>52.7. HBase MapReduce Summary to RDBMS</h3> <div class="paragraph"> <p>Sometimes it is more appropriate to generate summaries to an RDBMS. For these cases, it is possible to generate summaries directly to an RDBMS via a custom reducer. The <code>setup</code> method can connect to an RDBMS (the connection information can be passed via custom parameters in the context) and the cleanup method can close the connection.</p> </div> <div class="paragraph"> <p>It is critical to understand that number of reducers for the job affects the summarization implementation, and you’ll have to design this into your reducer. Specifically, whether it is designed to run as a singleton (one reducer) or multiple reducers. Neither is right or wrong, it depends on your use-case. Recognize that the more reducers that are assigned to the job, the more simultaneous connections to the RDBMS will be created - this will scale, but only to a point.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">class</span> <span class="class">MyRdbmsReducer</span> <span class="directive">extends</span> Reducer<Text, IntWritable, Text, IntWritable> { <span class="directive">private</span> <span class="predefined-type">Connection</span> c = <span class="predefined-constant">null</span>; <span class="directive">public</span> <span class="type">void</span> setup(<span class="predefined-type">Context</span> context) { <span class="comment">// create DB connection...</span> } <span class="directive">public</span> <span class="type">void</span> reduce(Text key, <span class="predefined-type">Iterable</span><IntWritable> values, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="comment">// do summarization</span> <span class="comment">// in this example the keys are Text, but this is just an example</span> } <span class="directive">public</span> <span class="type">void</span> cleanup(<span class="predefined-type">Context</span> context) { <span class="comment">// close db connection</span> } }</code></pre> </div> </div> <div class="paragraph"> <p>In the end, the summary results are written to your RDBMS table/s.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="mapreduce.htable.access"><a class="anchor" href="#mapreduce.htable.access"></a>53. Accessing Other HBase Tables in a MapReduce Job</h2> <div class="sectionbody"> <div class="paragraph"> <p>Although the framework currently allows one HBase table as input to a MapReduce job, other HBase tables can be accessed as lookup tables, etc., in a MapReduce job via creating an Table instance in the setup method of the Mapper.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">class</span> <span class="class">MyMapper</span> <span class="directive">extends</span> TableMapper<Text, LongWritable> { <span class="directive">private</span> Table myOtherTable; <span class="directive">public</span> <span class="type">void</span> setup(<span class="predefined-type">Context</span> context) { <span class="comment">// In here create a Connection to the cluster and save it or use the Connection</span> <span class="comment">// from the existing table</span> myOtherTable = connection.getTable(<span class="string"><span class="delimiter">"</span><span class="content">myOtherTable</span><span class="delimiter">"</span></span>); } <span class="directive">public</span> <span class="type">void</span> map(ImmutableBytesWritable row, <span class="predefined-type">Result</span> value, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="comment">// process Result...</span> <span class="comment">// use 'myOtherTable' for lookups</span> }</code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="mapreduce.specex"><a class="anchor" href="#mapreduce.specex"></a>54. Speculative Execution</h2> <div class="sectionbody"> <div class="paragraph"> <p>It is generally advisable to turn off speculative execution for MapReduce jobs that use HBase as a source. This can either be done on a per-Job basis through properties, on on the entire cluster. Especially for longer running jobs, speculative execution will create duplicate map-tasks which will double-write your data to HBase; this is probably not what you want.</p> </div> <div class="paragraph"> <p>See <a href="#spec.ex">spec.ex</a> for more information.</p> </div> </div> </div> <h1 id="security" class="sect0"><a class="anchor" href="#security"></a>Securing Apache HBase</h1> <div class="openblock partintro"> <div class="content"> <div class="admonitionblock important"> <table> <tr> <td class="icon"> <i class="fa icon-important" title="Important"></i> </td> <td class="content"> <div class="title">Reporting Security Bugs</div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> To protect existing HBase installations from exploitation, please <strong>do not</strong> use JIRA to report security-related bugs. Instead, send your report to the mailing list <a href="mailto:private@apache.org">private@apache.org</a>, which allows anyone to send messages, but restricts who can read them. Someone on that list will contact you to follow up on your report. </td> </tr> </table> </div> <div class="paragraph"> <p>HBase adheres to the Apache Software Foundation’s policy on reported vulnerabilities, available at <a href="http://apache.org/security/" class="bare">http://apache.org/security/</a>.</p> </div> <div class="paragraph"> <p>If you wish to send an encrypted report, you can use the GPG details provided for the general ASF security list. This will likely increase the response time to your report.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>HBase provides mechanisms to secure various components and aspects of HBase and how it relates to the rest of the Hadoop infrastructure, as well as clients and resources outside Hadoop.</p> </div> </div> </div> <div class="sect1"> <h2 id="_using_secure_http_https_for_the_web_ui"><a class="anchor" href="#_using_secure_http_https_for_the_web_ui"></a>55. Using Secure HTTP (HTTPS) for the Web UI</h2> <div class="sectionbody"> <div class="paragraph"> <p>A default HBase install uses insecure HTTP connections for Web UIs for the master and region servers. To enable secure HTTP (HTTPS) connections instead, set <code>hadoop.ssl.enabled</code> to <code>true</code> in <em>hbase-site.xml</em>. This does not change the port used by the Web UI. To change the port for the web UI for a given HBase component, configure that port’s setting in hbase-site.xml. These settings are:</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.master.info.port</code></p> </li> <li> <p><code>hbase.regionserver.info.port</code></p> </li> </ul> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">If you enable HTTPS, clients should avoid using the non-secure HTTP connection.</div> <div class="paragraph"> <p>If you enable secure HTTP, clients should connect to HBase using the <code>https://</code> URL. Clients using the <code>http://</code> URL will receive an HTTP response of <code>200</code>, but will not receive any data. The following exception is logged:</p> </div> <div class="listingblock"> <div class="content"> <pre>javax.net.ssl.SSLException: Unrecognized SSL message, plaintext connection?</pre> </div> </div> <div class="paragraph"> <p>This is because the same port is used for HTTP and HTTPS.</p> </div> <div class="paragraph"> <p>HBase uses Jetty for the Web UI. Without modifying Jetty itself, it does not seem possible to configure Jetty to redirect one port to another on the same host. See Nick Dimiduk’s contribution on this <a href="http://stackoverflow.com/questions/20611815/redirect-from-http-to-https-in-jetty">Stack Overflow</a> thread for more information. If you know how to fix this without opening a second port for HTTPS, patches are appreciated.</p> </div> </td> </tr> </table> </div> </div> </div> <div class="sect1"> <h2 id="hbase.secure.spnego.ui"><a class="anchor" href="#hbase.secure.spnego.ui"></a>56. Using SPNEGO for Kerberos authentication with Web UIs</h2> <div class="sectionbody"> <div class="paragraph"> <p>Kerberos-authentication to HBase Web UIs can be enabled via configuring SPNEGO with the <code>hbase.security.authentication.ui</code> property in <em>hbase-site.xml</em>. Enabling this authentication requires that HBase is also configured to use Kerberos authentication for RPCs (e.g <code>hbase.security.authentication</code> = <code>kerberos</code>).</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication.ui<span class="tag"></name></span> <span class="tag"><value></span>kerberos<span class="tag"></value></span> <span class="tag"><description></span>Controls what kind of authentication should be used for the HBase web UIs.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication<span class="tag"></name></span> <span class="tag"><value></span>kerberos<span class="tag"></value></span> <span class="tag"><description></span>The Kerberos keytab file to use for SPNEGO authentication by the web server.<span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>A number of properties exist to configure SPNEGO authentication for the web server:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication.spnego.kerberos.principal<span class="tag"></name></span> <span class="tag"><value></span>HTTP/_HOST@EXAMPLE.COM<span class="tag"></value></span> <span class="tag"><description></span>Required for SPNEGO, the Kerberos principal to use for SPNEGO authentication by the web server. The _HOST keyword will be automatically substituted with the node's hostname.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication.spnego.kerberos.keytab<span class="tag"></name></span> <span class="tag"><value></span>/etc/security/keytabs/spnego.service.keytab<span class="tag"></value></span> <span class="tag"><description></span>Required for SPNEGO, the Kerberos keytab file to use for SPNEGO authentication by the web server.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication.spnego.kerberos.name.rules<span class="tag"></name></span> <span class="tag"><value></span><span class="tag"></value></span> <span class="tag"><description></span>Optional, Hadoop-style `auth_to_local` rules which will be parsed and used in the handling of Kerberos principals<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication.signature.secret.file<span class="tag"></name></span> <span class="tag"><value></span><span class="tag"></value></span> <span class="tag"><description></span>Optional, a file whose contents will be used as a secret to sign the HTTP cookies as a part of the SPNEGO authentication handshake. If this is not provided, Java's `Random` library will be used for the secret.<span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbase.secure.configuration"><a class="anchor" href="#hbase.secure.configuration"></a>57. Secure Client Access to Apache HBase</h2> <div class="sectionbody"> <div class="paragraph"> <p>Newer releases of Apache HBase (>= 0.92) support optional SASL authentication of clients. See also Matteo Bertozzi’s article on <a href="http://www.cloudera.com/blog/2012/09/understanding-user-authentication-and-authorization-in-apache-hbase/">Understanding User Authentication and Authorization in Apache HBase</a>.</p> </div> <div class="paragraph"> <p>This describes how to set up Apache HBase and clients for connection to secure HBase resources.</p> </div> <div class="sect2"> <h3 id="security.prerequisites"><a class="anchor" href="#security.prerequisites"></a>57.1. Prerequisites</h3> <div class="dlist"> <dl> <dt class="hdlist1">Hadoop Authentication Configuration</dt> <dd> <p>To run HBase RPC with strong authentication, you must set <code>hbase.security.authentication</code> to <code>kerberos</code>. In this case, you must also set <code>hadoop.security.authentication</code> to <code>kerberos</code> in core-site.xml. Otherwise, you would be using strong authentication for HBase but not for the underlying HDFS, which would cancel out any benefit.</p> </dd> <dt class="hdlist1">Kerberos KDC</dt> <dd> <p>You need to have a working Kerberos KDC.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_server_side_configuration_for_secure_operation"><a class="anchor" href="#_server_side_configuration_for_secure_operation"></a>57.2. Server-side Configuration for Secure Operation</h3> <div class="paragraph"> <p>First, refer to <a href="#security.prerequisites">security.prerequisites</a> and ensure that your underlying HDFS configuration is secure.</p> </div> <div class="paragraph"> <p>Add the following to the <code>hbase-site.xml</code> file on every server machine in the cluster:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication<span class="tag"></name></span> <span class="tag"><value></span>kerberos<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.token.TokenProvider<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>A full shutdown and restart of HBase service is required when deploying these configuration changes.</p> </div> </div> <div class="sect2"> <h3 id="_client_side_configuration_for_secure_operation"><a class="anchor" href="#_client_side_configuration_for_secure_operation"></a>57.3. Client-side Configuration for Secure Operation</h3> <div class="paragraph"> <p>First, refer to <a href="#security.prerequisites">Prerequisites</a> and ensure that your underlying HDFS configuration is secure.</p> </div> <div class="paragraph"> <p>Add the following to the <code>hbase-site.xml</code> file on every client:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication<span class="tag"></name></span> <span class="tag"><value></span>kerberos<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>The client environment must be logged in to Kerberos from KDC or keytab via the <code>kinit</code> command before communication with the HBase cluster will be possible.</p> </div> <div class="paragraph"> <p>Be advised that if the <code>hbase.security.authentication</code> in the client- and server-side site files do not match, the client will not be able to communicate with the cluster.</p> </div> <div class="paragraph"> <p>Once HBase is configured for secure RPC it is possible to optionally configure encrypted communication. To do so, add the following to the <code>hbase-site.xml</code> file on every client:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rpc.protection<span class="tag"></name></span> <span class="tag"><value></span>privacy<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>This configuration property can also be set on a per-connection basis. Set it in the <code>Configuration</code> supplied to <code>Table</code>:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> conf = HBaseConfiguration.create(); <span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf); conf.set(<span class="string"><span class="delimiter">"</span><span class="content">hbase.rpc.protection</span><span class="delimiter">"</span></span>, <span class="string"><span class="delimiter">"</span><span class="content">privacy</span><span class="delimiter">"</span></span>); <span class="keyword">try</span> (<span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf)) { <span class="keyword">try</span> (Table table = connection.getTable(TableName.valueOf(tablename)) { .... do your stuff } }</code></pre> </div> </div> <div class="paragraph"> <p>Expect a ~10% performance penalty for encrypted communication.</p> </div> </div> <div class="sect2"> <h3 id="security.client.thrift"><a class="anchor" href="#security.client.thrift"></a>57.4. Client-side Configuration for Secure Operation - Thrift Gateway</h3> <div class="paragraph"> <p>Add the following to the <code>hbase-site.xml</code> file for every Thrift gateway:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.thrift.keytab.file<span class="tag"></name></span> <span class="tag"><value></span>/etc/hbase/conf/hbase.keytab<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.thrift.kerberos.principal<span class="tag"></name></span> <span class="tag"><value></span>$USER/_HOST@HADOOP.LOCALDOMAIN<span class="tag"></value></span> <span class="comment"></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Substitute the appropriate credential and keytab for <em>$USER</em> and <em>$KEYTAB</em> respectively.</p> </div> <div class="paragraph"> <p>In order to use the Thrift API principal to interact with HBase, it is also necessary to add the <code>hbase.thrift.kerberos.principal</code> to the <code><em>acl</em></code> table. For example, to give the Thrift API principal, <code>thrift_server</code>, administrative access, a command such as this one will suffice:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">grant</span> <span class="string"><span class="delimiter">'</span><span class="content">thrift_server</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">RWCA</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>For more information about ACLs, please see the <a href="#hbase.accesscontrol.configuration">Access Control Labels (ACLs)</a> section</p> </div> <div class="paragraph"> <p>The Thrift gateway will authenticate with HBase using the supplied credential. No authentication will be performed by the Thrift gateway itself. All client access via the Thrift gateway will use the Thrift gateway’s credential and have its privilege.</p> </div> </div> <div class="sect2"> <h3 id="security.gateway.thrift"><a class="anchor" href="#security.gateway.thrift"></a>57.5. Configure the Thrift Gateway to Authenticate on Behalf of the Client</h3> <div class="paragraph"> <p><a href="#security.client.thrift">Client-side Configuration for Secure Operation - Thrift Gateway</a> describes how to authenticate a Thrift client to HBase using a fixed user. As an alternative, you can configure the Thrift gateway to authenticate to HBase on the client’s behalf, and to access HBase using a proxy user. This was implemented in <a href="https://issues.apache.org/jira/browse/HBASE-11349">HBASE-11349</a> for Thrift 1, and <a href="https://issues.apache.org/jira/browse/HBASE-11474">HBASE-11474</a> for Thrift 2.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Limitations with Thrift Framed Transport</div> <div class="paragraph"> <p>If you use framed transport, you cannot yet take advantage of this feature, because SASL does not work with Thrift framed transport at this time.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>To enable it, do the following.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Be sure Thrift is running in secure mode, by following the procedure described in <a href="#security.client.thrift">Client-side Configuration for Secure Operation - Thrift Gateway</a>.</p> </li> <li> <p>Be sure that HBase is configured to allow proxy users, as described in <a href="#security.rest.gateway">REST Gateway Impersonation Configuration</a>.</p> </li> <li> <p>In <em>hbase-site.xml</em> for each cluster node running a Thrift gateway, set the property <code>hbase.thrift.security.qop</code> to one of the following three values:</p> <div class="ulist"> <ul> <li> <p><code>privacy</code> - authentication, integrity, and confidentiality checking.</p> </li> <li> <p><code>integrity</code> - authentication and integrity checking</p> </li> <li> <p><code>authentication</code> - authentication checking only</p> </li> </ul> </div> </li> <li> <p>Restart the Thrift gateway processes for the changes to take effect. If a node is running Thrift, the output of the <code>jps</code> command will list a <code>ThriftServer</code> process. To stop Thrift on a node, run the command <code>bin/hbase-daemon.sh stop thrift</code>. To start Thrift on a node, run the command <code>bin/hbase-daemon.sh start thrift</code>.</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="security.gateway.thrift.doas"><a class="anchor" href="#security.gateway.thrift.doas"></a>57.6. Configure the Thrift Gateway to Use the <code>doAs</code> Feature</h3> <div class="paragraph"> <p><a href="#security.gateway.thrift">Configure the Thrift Gateway to Authenticate on Behalf of the Client</a> describes how to configure the Thrift gateway to authenticate to HBase on the client’s behalf, and to access HBase using a proxy user. The limitation of this approach is that after the client is initialized with a particular set of credentials, it cannot change these credentials during the session. The <code>doAs</code> feature provides a flexible way to impersonate multiple principals using the same client. This feature was implemented in <a href="https://issues.apache.org/jira/browse/HBASE-12640">HBASE-12640</a> for Thrift 1, but is currently not available for Thrift 2.</p> </div> <div class="paragraph"> <p><strong>To allow proxy users</strong>, add the following to the <em>hbase-site.xml</em> file for every HBase node:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.proxyuser.$USER.groups<span class="tag"></name></span> <span class="tag"><value></span>$GROUPS<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.proxyuser.$USER.hosts<span class="tag"></name></span> <span class="tag"><value></span>$GROUPS<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p><strong>To enable the <code>doAs</code> feature</strong>, add the following to the <em>hbase-site.xml</em> file for every Thrift gateway:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.thrift.http<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.thrift.support.proxyuser<span class="tag"></name></span> <span class="tag"><value></span>true/value<span class="error">></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Take a look at the <a href="https://github.com/apache/hbase/blob/master/hbase-examples/src/main/java/org/apache/hadoop/hbase/thrift/HttpDoAsClient.java">demo client</a> to get an overall idea of how to use this feature in your client.</p> </div> </div> <div class="sect2"> <h3 id="_client_side_configuration_for_secure_operation_rest_gateway"><a class="anchor" href="#_client_side_configuration_for_secure_operation_rest_gateway"></a>57.7. Client-side Configuration for Secure Operation - REST Gateway</h3> <div class="paragraph"> <p>Add the following to the <code>hbase-site.xml</code> file for every REST gateway:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.keytab.file<span class="tag"></name></span> <span class="tag"><value></span>$KEYTAB<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.kerberos.principal<span class="tag"></name></span> <span class="tag"><value></span>$USER/_HOST@HADOOP.LOCALDOMAIN<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Substitute the appropriate credential and keytab for <em>$USER</em> and <em>$KEYTAB</em> respectively.</p> </div> <div class="paragraph"> <p>The REST gateway will authenticate with HBase using the supplied credential.</p> </div> <div class="paragraph"> <p>In order to use the REST API principal to interact with HBase, it is also necessary to add the <code>hbase.rest.kerberos.principal</code> to the <code><em>acl</em></code> table. For example, to give the REST API principal, <code>rest_server</code>, administrative access, a command such as this one will suffice:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">grant</span> <span class="string"><span class="delimiter">'</span><span class="content">rest_server</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">RWCA</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>For more information about ACLs, please see the <a href="#hbase.accesscontrol.configuration">Access Control Labels (ACLs)</a> section</p> </div> <div class="paragraph"> <p>HBase REST gateway supports <a href="http://hadoop.apache.org/docs/stable/hadoop-auth/index.html">SPNEGO HTTP authentication</a> for client access to the gateway. To enable REST gateway Kerberos authentication for client access, add the following to the <code>hbase-site.xml</code> file for every REST gateway.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.authentication.type<span class="tag"></name></span> <span class="tag"><value></span>kerberos<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.authentication.kerberos.principal<span class="tag"></name></span> <span class="tag"><value></span>HTTP/_HOST@HADOOP.LOCALDOMAIN<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.authentication.kerberos.keytab<span class="tag"></name></span> <span class="tag"><value></span>$KEYTAB<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Substitute the keytab for HTTP for <em>$KEYTAB</em>.</p> </div> <div class="paragraph"> <p>HBase REST gateway supports different 'hbase.rest.authentication.type': simple, kerberos. You can also implement a custom authentication by implemening Hadoop AuthenticationHandler, then specify the full class name as 'hbase.rest.authentication.type' value. For more information, refer to <a href="http://hadoop.apache.org/docs/stable/hadoop-auth/index.html">SPNEGO HTTP authentication</a>.</p> </div> </div> <div class="sect2"> <h3 id="security.rest.gateway"><a class="anchor" href="#security.rest.gateway"></a>57.8. REST Gateway Impersonation Configuration</h3> <div class="paragraph"> <p>By default, the REST gateway doesn’t support impersonation. It accesses the HBase on behalf of clients as the user configured as in the previous section. To the HBase server, all requests are from the REST gateway user. The actual users are unknown. You can turn on the impersonation support. With impersonation, the REST gateway user is a proxy user. The HBase server knows the acutal/real user of each request. So it can apply proper authorizations.</p> </div> <div class="paragraph"> <p>To turn on REST gateway impersonation, we need to configure HBase servers (masters and region servers) to allow proxy users; configure REST gateway to enable impersonation.</p> </div> <div class="paragraph"> <p>To allow proxy users, add the following to the <code>hbase-site.xml</code> file for every HBase server:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.proxyuser.$USER.groups<span class="tag"></name></span> <span class="tag"><value></span>$GROUPS<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.proxyuser.$USER.hosts<span class="tag"></name></span> <span class="tag"><value></span>$GROUPS<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Substitute the REST gateway proxy user for <em>$USER</em>, and the allowed group list for <em>$GROUPS</em>.</p> </div> <div class="paragraph"> <p>To enable REST gateway impersonation, add the following to the <code>hbase-site.xml</code> file for every REST gateway.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.authentication.type<span class="tag"></name></span> <span class="tag"><value></span>kerberos<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.authentication.kerberos.principal<span class="tag"></name></span> <span class="tag"><value></span>HTTP/_HOST@HADOOP.LOCALDOMAIN<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rest.authentication.kerberos.keytab<span class="tag"></name></span> <span class="tag"><value></span>$KEYTAB<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Substitute the keytab for HTTP for <em>$KEYTAB</em>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbase.secure.simpleconfiguration"><a class="anchor" href="#hbase.secure.simpleconfiguration"></a>58. Simple User Access to Apache HBase</h2> <div class="sectionbody"> <div class="paragraph"> <p>Newer releases of Apache HBase (>= 0.92) support optional SASL authentication of clients. See also Matteo Bertozzi’s article on <a href="http://www.cloudera.com/blog/2012/09/understanding-user-authentication-and-authorization-in-apache-hbase/">Understanding User Authentication and Authorization in Apache HBase</a>.</p> </div> <div class="paragraph"> <p>This describes how to set up Apache HBase and clients for simple user access to HBase resources.</p> </div> <div class="sect2"> <h3 id="_simple_versus_secure_access"><a class="anchor" href="#_simple_versus_secure_access"></a>58.1. Simple versus Secure Access</h3> <div class="paragraph"> <p>The following section shows how to set up simple user access. Simple user access is not a secure method of operating HBase. This method is used to prevent users from making mistakes. It can be used to mimic the Access Control using on a development system without having to set up Kerberos.</p> </div> <div class="paragraph"> <p>This method is not used to prevent malicious or hacking attempts. To make HBase secure against these types of attacks, you must configure HBase for secure operation. Refer to the section <a href="#hbase.secure.configuration">Secure Client Access to Apache HBase</a> and complete all of the steps described there.</p> </div> </div> <div class="sect2"> <h3 id="_prerequisites"><a class="anchor" href="#_prerequisites"></a>58.2. Prerequisites</h3> <div class="paragraph"> <p>None</p> </div> </div> <div class="sect2"> <h3 id="_server_side_configuration_for_simple_user_access_operation"><a class="anchor" href="#_server_side_configuration_for_simple_user_access_operation"></a>58.3. Server-side Configuration for Simple User Access Operation</h3> <div class="paragraph"> <p>Add the following to the <code>hbase-site.xml</code> file on every server machine in the cluster:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication<span class="tag"></name></span> <span class="tag"><value></span>simple<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.master.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.regionserver.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>For 0.94, add the following to the <code>hbase-site.xml</code> file on every server machine in the cluster:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rpc.engine<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.ipc.SecureRpcEngine<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.master.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>A full shutdown and restart of HBase service is required when deploying these configuration changes.</p> </div> </div> <div class="sect2"> <h3 id="_client_side_configuration_for_simple_user_access_operation"><a class="anchor" href="#_client_side_configuration_for_simple_user_access_operation"></a>58.4. Client-side Configuration for Simple User Access Operation</h3> <div class="paragraph"> <p>Add the following to the <code>hbase-site.xml</code> file on every client:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authentication<span class="tag"></name></span> <span class="tag"><value></span>simple<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>For 0.94, add the following to the <code>hbase-site.xml</code> file on every server machine in the cluster:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.rpc.engine<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.ipc.SecureRpcEngine<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Be advised that if the <code>hbase.security.authentication</code> in the client- and server-side site files do not match, the client will not be able to communicate with the cluster.</p> </div> <div class="sect3"> <h4 id="_client_side_configuration_for_simple_user_access_operation_thrift_gateway"><a class="anchor" href="#_client_side_configuration_for_simple_user_access_operation_thrift_gateway"></a>58.4.1. Client-side Configuration for Simple User Access Operation - Thrift Gateway</h4> <div class="paragraph"> <p>The Thrift gateway user will need access. For example, to give the Thrift API user, <code>thrift_server</code>, administrative access, a command such as this one will suffice:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">grant</span> <span class="string"><span class="delimiter">'</span><span class="content">thrift_server</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">RWCA</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>For more information about ACLs, please see the <a href="#hbase.accesscontrol.configuration">Access Control Labels (ACLs)</a> section</p> </div> <div class="paragraph"> <p>The Thrift gateway will authenticate with HBase using the supplied credential. No authentication will be performed by the Thrift gateway itself. All client access via the Thrift gateway will use the Thrift gateway’s credential and have its privilege.</p> </div> </div> <div class="sect3"> <h4 id="_client_side_configuration_for_simple_user_access_operation_rest_gateway"><a class="anchor" href="#_client_side_configuration_for_simple_user_access_operation_rest_gateway"></a>58.4.2. Client-side Configuration for Simple User Access Operation - REST Gateway</h4> <div class="paragraph"> <p>The REST gateway will authenticate with HBase using the supplied credential. No authentication will be performed by the REST gateway itself. All client access via the REST gateway will use the REST gateway’s credential and have its privilege.</p> </div> <div class="paragraph"> <p>The REST gateway user will need access. For example, to give the REST API user, <code>rest_server</code>, administrative access, a command such as this one will suffice:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">grant</span> <span class="string"><span class="delimiter">'</span><span class="content">rest_server</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">RWCA</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>For more information about ACLs, please see the <a href="#hbase.accesscontrol.configuration">Access Control Labels (ACLs)</a> section</p> </div> <div class="paragraph"> <p>It should be possible for clients to authenticate with the HBase cluster through the REST gateway in a pass-through manner via SPNEGO HTTP authentication. This is future work.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="_securing_access_to_hdfs_and_zookeeper"><a class="anchor" href="#_securing_access_to_hdfs_and_zookeeper"></a>59. Securing Access to HDFS and ZooKeeper</h2> <div class="sectionbody"> <div class="paragraph"> <p>Secure HBase requires secure ZooKeeper and HDFS so that users cannot access and/or modify the metadata and data from under HBase. HBase uses HDFS (or configured file system) to keep its data files as well as write ahead logs (WALs) and other data. HBase uses ZooKeeper to store some metadata for operations (master address, table locks, recovery state, etc).</p> </div> <div class="sect2"> <h3 id="_securing_zookeeper_data"><a class="anchor" href="#_securing_zookeeper_data"></a>59.1. Securing ZooKeeper Data</h3> <div class="paragraph"> <p>ZooKeeper has a pluggable authentication mechanism to enable access from clients using different methods. ZooKeeper even allows authenticated and un-authenticated clients at the same time. The access to znodes can be restricted by providing Access Control Lists (ACLs) per znode. An ACL contains two components, the authentication method and the principal. ACLs are NOT enforced hierarchically. See <a href="https://zookeeper.apache.org/doc/r3.3.6/zookeeperProgrammers.html#sc_ZooKeeperPluggableAuthentication">ZooKeeper Programmers Guide</a> for details.</p> </div> <div class="paragraph"> <p>HBase daemons authenticate to ZooKeeper via SASL and kerberos (See <a href="#zk.sasl.auth">SASL Authentication with ZooKeeper</a>). HBase sets up the znode ACLs so that only the HBase user and the configured hbase superuser (<code>hbase.superuser</code>) can access and modify the data. In cases where ZooKeeper is used for service discovery or sharing state with the client, the znodes created by HBase will also allow anyone (regardless of authentication) to read these znodes (clusterId, master address, meta location, etc), but only the HBase user can modify them.</p> </div> </div> <div class="sect2"> <h3 id="_securing_file_system_hdfs_data"><a class="anchor" href="#_securing_file_system_hdfs_data"></a>59.2. Securing File System (HDFS) Data</h3> <div class="paragraph"> <p>All of the data under management is kept under the root directory in the file system (<code>hbase.rootdir</code>). Access to the data and WAL files in the filesystem should be restricted so that users cannot bypass the HBase layer, and peek at the underlying data files from the file system. HBase assumes the filesystem used (HDFS or other) enforces permissions hierarchically. If sufficient protection from the file system (both authorization and authentication) is not provided, HBase level authorization control (ACLs, visibility labels, etc) is meaningless since the user can always access the data from the file system.</p> </div> <div class="paragraph"> <p>HBase enforces the posix-like permissions 700 (<code>rwx------</code>) to its root directory. It means that only the HBase user can read or write the files in FS. The default setting can be changed by configuring <code>hbase.rootdir.perms</code> in hbase-site.xml. A restart of the active master is needed so that it changes the used permissions. For versions before 1.2.0, you can check whether HBASE-13780 is committed, and if not, you can manually set the permissions for the root directory if needed. Using HDFS, the command would be:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">sudo -u hdfs hadoop fs -chmod 700 /hbase</code></pre> </div> </div> <div class="paragraph"> <p>You should change <code>/hbase</code> if you are using a different <code>hbase.rootdir</code>.</p> </div> <div class="paragraph"> <p>In secure mode, SecureBulkLoadEndpoint should be configured and used for properly handing of users files created from MR jobs to the HBase daemons and HBase user. The staging directory in the distributed file system used for bulk load (<code>hbase.bulkload.staging.dir</code>, defaults to <code>/tmp/hbase-staging</code>) should have (mode 711, or <code>rwx—x—x</code>) so that users can access the staging directory created under that parent directory, but cannot do any other operation. See <a href="#hbase.secure.bulkload">Secure Bulk Load</a> for how to configure SecureBulkLoadEndPoint.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_securing_access_to_your_data"><a class="anchor" href="#_securing_access_to_your_data"></a>60. Securing Access To Your Data</h2> <div class="sectionbody"> <div class="paragraph"> <p>After you have configured secure authentication between HBase client and server processes and gateways, you need to consider the security of your data itself. HBase provides several strategies for securing your data:</p> </div> <div class="ulist"> <ul> <li> <p>Role-based Access Control (RBAC) controls which users or groups can read and write to a given HBase resource or execute a coprocessor endpoint, using the familiar paradigm of roles.</p> </li> <li> <p>Visibility Labels which allow you to label cells and control access to labelled cells, to further restrict who can read or write to certain subsets of your data. Visibility labels are stored as tags. See <a href="#hbase.tags">hbase.tags</a> for more information.</p> </li> <li> <p>Transparent encryption of data at rest on the underlying filesystem, both in HFiles and in the WAL. This protects your data at rest from an attacker who has access to the underlying filesystem, without the need to change the implementation of the client. It can also protect against data leakage from improperly disposed disks, which can be important for legal and regulatory compliance.</p> </li> </ul> </div> <div class="paragraph"> <p>Server-side configuration, administration, and implementation details of each of these features are discussed below, along with any performance trade-offs. An example security configuration is given at the end, to show these features all used together, as they might be in a real-world scenario.</p> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> All aspects of security in HBase are in active development and evolving rapidly. Any strategy you employ for security of your data should be thoroughly tested. In addition, some of these features are still in the experimental stage of development. To take advantage of many of these features, you must be running HBase 0.98+ and using the HFile v3 file format. </td> </tr> </table> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="title">Protecting Sensitive Files</div> <div class="paragraph"> <p>Several procedures in this section require you to copy files between cluster nodes. When copying keys, configuration files, or other files containing sensitive strings, use a secure method, such as <code>ssh</code>, to avoid leaking sensitive data.</p> </div> </td> </tr> </table> </div> <div class="olist arabic"> <div class="title">Procedure: Basic Server-Side Configuration</div> <ol class="arabic"> <li> <p>Enable HFile v3, by setting <code>hfile.format.version</code> to 3 in <em>hbase-site.xml</em>. This is the default for HBase 1.0 and newer.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hfile.format.version<span class="tag"></name></span> <span class="tag"><value></span>3<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </li> <li> <p>Enable SASL and Kerberos authentication for RPC and ZooKeeper, as described in <a href="#security.prerequisites">security.prerequisites</a> and <a href="#zk.sasl.auth">SASL Authentication with ZooKeeper</a>.</p> </li> </ol> </div> <div class="sect2"> <h3 id="hbase.tags"><a class="anchor" href="#hbase.tags"></a>60.1. Tags</h3> <div class="paragraph"> <p><em class="firstterm">Tags</em> are a feature of HFile v3. A tag is a piece of metadata which is part of a cell, separate from the key, value, and version. Tags are an implementation detail which provides a foundation for other security-related features such as cell-level ACLs and visibility labels. Tags are stored in the HFiles themselves. It is possible that in the future, tags will be used to implement other HBase features. You don’t need to know a lot about tags in order to use the security features they enable.</p> </div> <div class="sect3"> <h4 id="_implementation_details"><a class="anchor" href="#_implementation_details"></a>60.1.1. Implementation Details</h4> <div class="paragraph"> <p>Every cell can have zero or more tags. Every tag has a type and the actual tag byte array.</p> </div> <div class="paragraph"> <p>Just as row keys, column families, qualifiers and values can be encoded (see <a href="#data.block.encoding.types">data.block.encoding.types</a>), tags can also be encoded as well. You can enable or disable tag encoding at the level of the column family, and it is enabled by default. Use the <code>HColumnDescriptor#setCompressionTags(boolean compressTags)</code> method to manage encoding settings on a column family. You also need to enable the DataBlockEncoder for the column family, for encoding of tags to take effect.</p> </div> <div class="paragraph"> <p>You can enable compression of each tag in the WAL, if WAL compression is also enabled, by setting the value of <code>hbase.regionserver.wal.tags.enablecompression</code> to <code>true</code> in <em>hbase-site.xml</em>. Tag compression uses dictionary encoding.</p> </div> <div class="paragraph"> <p>Tag compression is not supported when using WAL encryption.</p> </div> </div> </div> <div class="sect2"> <h3 id="hbase.accesscontrol.configuration"><a class="anchor" href="#hbase.accesscontrol.configuration"></a>60.2. Access Control Labels (ACLs)</h3> <div class="sect3"> <h4 id="_how_it_works"><a class="anchor" href="#_how_it_works"></a>60.2.1. How It Works</h4> <div class="paragraph"> <p>ACLs in HBase are based upon a user’s membership in or exclusion from groups, and a given group’s permissions to access a given resource. ACLs are implemented as a coprocessor called AccessController.</p> </div> <div class="paragraph"> <p>HBase does not maintain a private group mapping, but relies on a <em class="firstterm">Hadoop group mapper</em>, which maps between entities in a directory such as LDAP or Active Directory, and HBase users. Any supported Hadoop group mapper will work. Users are then granted specific permissions (Read, Write, Execute, Create, Admin) against resources (global, namespaces, tables, cells, or endpoints).</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> With Kerberos and Access Control enabled, client access to HBase is authenticated and user data is private unless access has been explicitly granted. </td> </tr> </table> </div> <div class="paragraph"> <p>HBase has a simpler security model than relational databases, especially in terms of client operations. No distinction is made between an insert (new record) and update (of existing record), for example, as both collapse down into a Put.</p> </div> <div class="sect4"> <h5 id="_understanding_access_levels"><a class="anchor" href="#_understanding_access_levels"></a>Understanding Access Levels</h5> <div class="paragraph"> <p>HBase access levels are granted independently of each other and allow for different types of operations at a given scope.</p> </div> <div class="ulist"> <ul> <li> <p><em>Read (R)</em> - can read data at the given scope</p> </li> <li> <p><em>Write (W)</em> - can write data at the given scope</p> </li> <li> <p><em>Execute (X)</em> - can execute coprocessor endpoints at the given scope</p> </li> <li> <p><em>Create (C)</em> - can create tables or drop tables (even those they did not create) at the given scope</p> </li> <li> <p><em>Admin (A)</em> - can perform cluster operations such as balancing the cluster or assigning regions at the given scope</p> </li> </ul> </div> <div class="paragraph"> <p>The possible scopes are:</p> </div> <div class="ulist"> <ul> <li> <p><em>Superuser</em> - superusers can perform any operation available in HBase, to any resource. The user who runs HBase on your cluster is a superuser, as are any principals assigned to the configuration property <code>hbase.superuser</code> in <em>hbase-site.xml</em> on the HMaster.</p> </li> <li> <p><em>Global</em> - permissions granted at <em>global</em> scope allow the admin to operate on all tables of the cluster.</p> </li> <li> <p><em>Namespace</em> - permissions granted at <em>namespace</em> scope apply to all tables within a given namespace.</p> </li> <li> <p><em>Table</em> - permissions granted at <em>table</em> scope apply to data or metadata within a given table.</p> </li> <li> <p><em>ColumnFamily</em> - permissions granted at <em>ColumnFamily</em> scope apply to cells within that ColumnFamily.</p> </li> <li> <p><em>Cell</em> - permissions granted at <em>cell</em> scope apply to that exact cell coordinate (key, value, timestamp). This allows for policy evolution along with data.</p> <div class="paragraph"> <p>To change an ACL on a specific cell, write an updated cell with new ACL to the precise coordinates of the original.</p> </div> <div class="paragraph"> <p>If you have a multi-versioned schema and want to update ACLs on all visible versions, you need to write new cells for all visible versions. The application has complete control over policy evolution.</p> </div> <div class="paragraph"> <p>The exception to the above rule is <code>append</code> and <code>increment</code> processing. Appends and increments can carry an ACL in the operation. If one is included in the operation, then it will be applied to the result of the <code>append</code> or <code>increment</code>. Otherwise, the ACL of the existing cell you are appending to or incrementing is preserved.</p> </div> </li> </ul> </div> <div class="paragraph"> <p>The combination of access levels and scopes creates a matrix of possible access levels that can be granted to a user. In a production environment, it is useful to think of access levels in terms of what is needed to do a specific job. The following list describes appropriate access levels for some common types of HBase users. It is important not to grant more access than is required for a given user to perform their required tasks.</p> </div> <div class="ulist"> <ul> <li> <p><em>Superusers</em> - In a production system, only the HBase user should have superuser access. In a development environment, an administrator may need superuser access in order to quickly control and manage the cluster. However, this type of administrator should usually be a Global Admin rather than a superuser.</p> </li> <li> <p><em>Global Admins</em> - A global admin can perform tasks and access every table in HBase. In a typical production environment, an admin should not have Read or Write permissions to data within tables.</p> </li> <li> <p>A global admin with Admin permissions can perform cluster-wide operations on the cluster, such as balancing, assigning or unassigning regions, or calling an explicit major compaction. This is an operations role.</p> </li> <li> <p>A global admin with Create permissions can create or drop any table within HBase. This is more of a DBA-type role.</p> <div class="paragraph"> <p>In a production environment, it is likely that different users will have only one of Admin and Create permissions.</p> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="paragraph"> <p>In the current implementation, a Global Admin with <code>Admin</code> permission can grant himself <code>Read</code> and <code>Write</code> permissions on a table and gain access to that table’s data. For this reason, only grant <code>Global Admin</code> permissions to trusted user who actually need them.</p> </div> <div class="paragraph"> <p>Also be aware that a <code>Global Admin</code> with <code>Create</code> permission can perform a <code>Put</code> operation on the ACL table, simulating a <code>grant</code> or <code>revoke</code> and circumventing the authorization check for <code>Global Admin</code> permissions.</p> </div> <div class="paragraph"> <p>Due to these issues, be cautious with granting <code>Global Admin</code> privileges.</p> </div> </td> </tr> </table> </div> </li> <li> <p><em>Namespace Admins</em> - a namespace admin with <code>Create</code> permissions can create or drop tables within that namespace, and take and restore snapshots. A namespace admin with <code>Admin</code> permissions can perform operations such as splits or major compactions on tables within that namespace.</p> </li> <li> <p><em>Table Admins</em> - A table admin can perform administrative operations only on that table. A table admin with <code>Create</code> permissions can create snapshots from that table or restore that table from a snapshot. A table admin with <code>Admin</code> permissions can perform operations such as splits or major compactions on that table.</p> </li> <li> <p><em>Users</em> - Users can read or write data, or both. Users can also execute coprocessor endpoints, if given <code>Executable</code> permissions.</p> </li> </ul> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 7. Real-World Example of Access Levels</caption> <colgroup> <col style="width: 25%;"> <col style="width: 25%;"> <col style="width: 25%;"> <col style="width: 25%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Job Title</th> <th class="tableblock halign-left valign-top">Scope</th> <th class="tableblock halign-left valign-top">Permissions</th> <th class="tableblock halign-left valign-top">Description</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Senior Administrator</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Global</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Access, Create</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Manages the cluster and gives access to Junior Administrators.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Junior Administrator</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Global</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Create</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Creates tables and gives access to Table Administrators.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Table Administrator</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Table</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Access</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Maintains a table from an operations point of view.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Data Analyst</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Table</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Read</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Creates reports from HBase data.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Web Application</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Table</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Read, Write</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Puts data into HBase and uses HBase data to perform operations.</p></td> </tr> </tbody> </table> <div class="paragraph"> <div class="title">ACL Matrix</div> <p>For more details on how ACLs map to specific HBase operations and tasks, see <a href="#appendix_acl_matrix">appendix acl matrix</a>.</p> </div> </div> <div class="sect4"> <h5 id="_implementation_details_2"><a class="anchor" href="#_implementation_details_2"></a>Implementation Details</h5> <div class="paragraph"> <p>Cell-level ACLs are implemented using tags (see <a href="#hbase.tags">Tags</a>). In order to use cell-level ACLs, you must be using HFile v3 and HBase 0.98 or newer.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Files created by HBase are owned by the operating system user running the HBase process. To interact with HBase files, you should use the API or bulk load facility.</p> </li> <li> <p>HBase does not model "roles" internally in HBase. Instead, group names can be granted permissions. This allows external modeling of roles via group membership. Groups are created and manipulated externally to HBase, via the Hadoop group mapping service.</p> </li> </ol> </div> </div> <div class="sect4"> <h5 id="_server_side_configuration"><a class="anchor" href="#_server_side_configuration"></a>Server-Side Configuration</h5> <div class="olist arabic"> <ol class="arabic"> <li> <p>As a prerequisite, perform the steps in <a href="#security.data.basic.server.side">[security.data.basic.server.side]</a>.</p> </li> <li> <p>Install and configure the AccessController coprocessor, by setting the following properties in <em>hbase-site.xml</em>. These properties take a list of classes.</p> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> If you use the AccessController along with the VisibilityController, the AccessController must come first in the list, because with both components active, the VisibilityController will delegate access control on its system tables to the AccessController. For an example of using both together, see <a href="#security.example.config">Security Configuration Example</a>. </td> </tr> </table> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController, org.apache.hadoop.hbase.security.token.TokenProvider<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.master.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.regionserver.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.exec.permission.checks<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Optionally, you can enable transport security, by setting <code>hbase.rpc.protection</code> to <code>privacy</code>. This requires HBase 0.98.4 or newer.</p> </div> </li> <li> <p>Set up the Hadoop group mapper in the Hadoop namenode’s <em>core-site.xml</em>. This is a Hadoop file, not an HBase file. Customize it to your site’s needs. Following is an example.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.security.LdapGroupsMapping<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.url<span class="tag"></name></span> <span class="tag"><value></span>ldap://server<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.bind.user<span class="tag"></name></span> <span class="tag"><value></span>Administrator@example-ad.local<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.bind.password<span class="tag"></name></span> <span class="tag"><value></span>****<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.base<span class="tag"></name></span> <span class="tag"><value></span>dc=example-ad,dc=local<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.filter.user<span class="tag"></name></span> <span class="tag"><value></span>(<span class="entity">&amp;</span>(objectClass=user)(sAMAccountName={0}))<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.filter.group<span class="tag"></name></span> <span class="tag"><value></span>(objectClass=group)<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.attr.member<span class="tag"></name></span> <span class="tag"><value></span>member<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.attr.group.name<span class="tag"></name></span> <span class="tag"><value></span>cn<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </li> <li> <p>Optionally, enable the early-out evaluation strategy. Prior to HBase 0.98.0, if a user was not granted access to a column family, or at least a column qualifier, an AccessDeniedException would be thrown. HBase 0.98.0 removed this exception in order to allow cell-level exceptional grants. To restore the old behavior in HBase 0.98.0-0.98.6, set <code>hbase.security.access.early_out</code> to <code>true</code> in <em>hbase-site.xml</em>. In HBase 0.98.6, the default has been returned to <code>true</code>.</p> </li> <li> <p>Distribute your configuration and restart your cluster for changes to take effect.</p> </li> <li> <p>To test your configuration, log into HBase Shell as a given user and use the <code>whoami</code> command to report the groups your user is part of. In this example, the user is reported as being a member of the <code>services</code> group.</p> <div class="listingblock"> <div class="content"> <pre>hbase> whoami service (auth:KERBEROS) groups: services</pre> </div> </div> </li> </ol> </div> </div> <div class="sect4"> <h5 id="_administration"><a class="anchor" href="#_administration"></a>Administration</h5> <div class="paragraph"> <p>Administration tasks can be performed from HBase Shell or via an API.</p> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> <div class="title">API Examples</div> <div class="paragraph"> <p>Many of the API examples below are taken from source files <em>hbase-server/src/test/java/org/apache/hadoop/hbase/security/access/TestAccessController.java</em> and <em>hbase-server/src/test/java/org/apache/hadoop/hbase/security/access/SecureTestUtil.java</em>.</p> </div> <div class="paragraph"> <p>Neither the examples, nor the source files they are taken from, are part of the public HBase API, and are provided for illustration only. Refer to the official API for usage instructions.</p> </div> </td> </tr> </table> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>User and Group Administration</p> <div class="paragraph"> <p>Users and groups are maintained external to HBase, in your directory.</p> </div> </li> <li> <p>Granting Access To A Namespace, Table, Column Family, or Cell</p> <div class="paragraph"> <p>There are a few different types of syntax for grant statements. The first, and most familiar, is as follows, with the table and column family being optional:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">grant</span> <span class="string"><span class="delimiter">'</span><span class="content">user</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">RWXCA</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">TABLE</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">CF</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">CQ</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>Groups and users are granted access in the same way, but groups are prefixed with an <code>@</code> symbol. In the same way, tables and namespaces are specified in the same way, but namespaces are prefixed with an <code>@</code> symbol.</p> </div> <div class="paragraph"> <p>It is also possible to grant multiple permissions against the same resource in a single statement, as in this example. The first sub-clause maps users to ACLs and the second sub-clause specifies the resource.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> HBase Shell support for granting and revoking access at the cell level is for testing and verification support, and should not be employed for production use because it won’t apply the permissions to cells that don’t exist yet. The correct way to apply cell level permissions is to do so in the application code when storing the values. </td> </tr> </table> </div> <div class="paragraph"> <div class="title">ACL Granularity and Evaluation Order</div> <p>ACLs are evaluated from least granular to most granular, and when an ACL is reached that grants permission, evaluation stops. This means that cell ACLs do not override ACLs at less granularity.</p> </div> <div class="exampleblock"> <div class="title">Example 20. HBase Shell</div> <div class="content"> <div class="ulist"> <ul> <li> <p>Global:</p> <div class="listingblock"> <div class="content"> <pre>hbase> grant '@admins', 'RWXCA'</pre> </div> </div> </li> <li> <p>Namespace:</p> <div class="listingblock"> <div class="content"> <pre>hbase> grant 'service', 'RWXCA', '@test-NS'</pre> </div> </div> </li> <li> <p>Table:</p> <div class="listingblock"> <div class="content"> <pre>hbase> grant 'service', 'RWXCA', 'user'</pre> </div> </div> </li> <li> <p>Column Family:</p> <div class="listingblock"> <div class="content"> <pre>hbase> grant '@developers', 'RW', 'user', 'i'</pre> </div> </div> </li> <li> <p>Column Qualifier:</p> <div class="listingblock"> <div class="content"> <pre>hbase> grant 'service, 'RW', 'user', 'i', 'foo'</pre> </div> </div> </li> <li> <p>Cell:</p> <div class="paragraph"> <p>The syntax for granting cell ACLs uses the following syntax:</p> </div> <div class="listingblock"> <div class="content"> <pre>grant <table>, \ { '<user-or-group>' => \ '<permissions>', ... }, \ { <scanner-specification> }</pre> </div> </div> </li> <li> <p><em><user-or-group></em> is the user or group name, prefixed with <code>@</code> in the case of a group.</p> </li> <li> <p><em><permissions></em> is a string containing any or all of "RWXCA", though only R and W are meaningful at cell scope.</p> </li> <li> <p><em><scanner-specification></em> is the scanner specification syntax and conventions used by the 'scan' shell command. For some examples of scanner specifications, issue the following HBase Shell command.</p> <div class="listingblock"> <div class="content"> <pre>hbase> help "scan"</pre> </div> </div> <div class="paragraph"> <p>This example grants read access to the 'testuser' user and read/write access to the 'developers' group, on cells in the 'pii' column which match the filter.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> grant 'user', \ { '@developers' => 'RW', 'testuser' => 'R' }, \ { COLUMNS => 'pii', FILTER => "(PrefixFilter ('test'))" }</pre> </div> </div> <div class="paragraph"> <p>The shell will run a scanner with the given criteria, rewrite the found cells with new ACLs, and store them back to their exact coordinates.</p> </div> </li> </ul> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 21. API</div> <div class="content"> <div class="paragraph"> <p>The following example shows how to grant access at the table level.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> grantOnTable(<span class="directive">final</span> HBaseTestingUtility util, <span class="directive">final</span> <span class="predefined-type">String</span> user, <span class="directive">final</span> TableName table, <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> family, <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> qualifier, <span class="directive">final</span> <span class="predefined-type">Permission</span>.Action... actions) <span class="directive">throws</span> <span class="exception">Exception</span> { SecureTestUtil.updateACLs(util, <span class="keyword">new</span> <span class="predefined-type">Callable</span><<span class="predefined-type">Void</span>>() { <span class="annotation">@Override</span> <span class="directive">public</span> <span class="predefined-type">Void</span> call() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">Configuration</span> conf = HBaseConfiguration.create(); <span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf); <span class="keyword">try</span> (<span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf)) { <span class="keyword">try</span> (Table table = connection.getTable(TableName.valueOf(tablename)) { AccessControlLists.ACL_TABLE_NAME); <span class="keyword">try</span> { BlockingRpcChannel service = acl.coprocessorService(HConstants.EMPTY_START_ROW); AccessControlService.BlockingInterface protocol = AccessControlService.newBlockingStub(service); ProtobufUtil.grant(protocol, user, table, family, qualifier, actions); } <span class="keyword">finally</span> { acl.close(); } <span class="keyword">return</span> <span class="predefined-constant">null</span>; } } } } }</code></pre> </div> </div> <div class="paragraph"> <p>To grant permissions at the cell level, you can use the <code>Mutation.setACL</code> method:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Mutation.setACL(<span class="predefined-type">String</span> user, <span class="predefined-type">Permission</span> perms) Mutation.setACL(<span class="predefined-type">Map</span><<span class="predefined-type">String</span>, <span class="predefined-type">Permission</span>> perms)</code></pre> </div> </div> <div class="paragraph"> <p>Specifically, this example provides read permission to a user called <code>user1</code> on any cells contained in a particular Put operation:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">put.setACL(<span class="error">“</span>user1<span class="error">”</span>, <span class="keyword">new</span> <span class="predefined-type">Permission</span>(<span class="predefined-type">Permission</span>.Action.READ))</code></pre> </div> </div> </div> </div> </li> <li> <p>Revoking Access Control From a Namespace, Table, Column Family, or Cell</p> <div class="paragraph"> <p>The <code>revoke</code> command and API are twins of the grant command and API, and the syntax is exactly the same. The only exception is that you cannot revoke permissions at the cell level. You can only revoke access that has previously been granted, and a <code>revoke</code> statement is not the same thing as explicit denial to a resource.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> HBase Shell support for granting and revoking access is for testing and verification support, and should not be employed for production use because it won’t apply the permissions to cells that don’t exist yet. The correct way to apply cell-level permissions is to do so in the application code when storing the values. </td> </tr> </table> </div> <div class="exampleblock"> <div class="title">Example 22. Revoking Access To a Table</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> revokeFromTable(<span class="directive">final</span> HBaseTestingUtility util, <span class="directive">final</span> <span class="predefined-type">String</span> user, <span class="directive">final</span> TableName table, <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> family, <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> qualifier, <span class="directive">final</span> <span class="predefined-type">Permission</span>.Action... actions) <span class="directive">throws</span> <span class="exception">Exception</span> { SecureTestUtil.updateACLs(util, <span class="keyword">new</span> <span class="predefined-type">Callable</span><<span class="predefined-type">Void</span>>() { <span class="annotation">@Override</span> <span class="directive">public</span> <span class="predefined-type">Void</span> call() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">Configuration</span> conf = HBaseConfiguration.create(); <span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf); Table acl = connection.getTable(util.getConfiguration(), AccessControlLists.ACL_TABLE_NAME); <span class="keyword">try</span> { BlockingRpcChannel service = acl.coprocessorService(HConstants.EMPTY_START_ROW); AccessControlService.BlockingInterface protocol = AccessControlService.newBlockingStub(service); ProtobufUtil.revoke(protocol, user, table, family, qualifier, actions); } <span class="keyword">finally</span> { acl.close(); } <span class="keyword">return</span> <span class="predefined-constant">null</span>; } }); }</code></pre> </div> </div> </div> </div> </li> <li> <p>Showing a User’s Effective Permissions</p> <div class="exampleblock"> <div class="title">Example 23. HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> user_permission 'user' hbase> user_permission '.*' hbase> user_permission JAVA_REGEX</pre> </div> </div> </div> </div> </li> </ol> </div> <div class="exampleblock"> <div class="title">Example 24. API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> verifyAllowed(User user, AccessTestAction action, <span class="type">int</span> count) <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="keyword">try</span> { <span class="predefined-type">Object</span> obj = user.runAs(action); <span class="keyword">if</span> (obj != <span class="predefined-constant">null</span> && obj <span class="keyword">instanceof</span> <span class="predefined-type">List</span>&lt;?&gt;) { <span class="predefined-type">List</span>&lt;?&gt; results = (<span class="predefined-type">List</span>&lt;?&gt;) obj; <span class="keyword">if</span> (results != <span class="predefined-constant">null</span> && results.isEmpty()) { fail(<span class="string"><span class="delimiter">"</span><span class="content">Empty non null results from action for user '</span><span class="delimiter">"</span></span> <span class="error">`</span> user.getShortName() <span class="error">`</span> <span class="string"><span class="delimiter">"</span><span class="content">'</span><span class="delimiter">"</span></span>); } assertEquals(count, results.size()); } } <span class="keyword">catch</span> (AccessDeniedException ade) { fail(<span class="string"><span class="delimiter">"</span><span class="content">Expected action to pass for user '</span><span class="delimiter">"</span></span> <span class="error">`</span> user.getShortName() <span class="error">`</span> <span class="string"><span class="delimiter">"</span><span class="content">' but was denied</span><span class="delimiter">"</span></span>); } }</code></pre> </div> </div> </div> </div> </div> </div> </div> <div class="sect2"> <h3 id="_visibility_labels"><a class="anchor" href="#_visibility_labels"></a>60.3. Visibility Labels</h3> <div class="paragraph"> <p>Visibility labels control can be used to only permit users or principals associated with a given label to read or access cells with that label. For instance, you might label a cell <code>top-secret</code>, and only grant access to that label to the <code>managers</code> group. Visibility labels are implemented using Tags, which are a feature of HFile v3, and allow you to store metadata on a per-cell basis. A label is a string, and labels can be combined into expressions by using logical operators (&, |, or !), and using parentheses for grouping. HBase does not do any kind of validation of expressions beyond basic well-formedness. Visibility labels have no meaning on their own, and may be used to denote sensitivity level, privilege level, or any other arbitrary semantic meaning.</p> </div> <div class="paragraph"> <p>If a user’s labels do not match a cell’s label or expression, the user is denied access to the cell.</p> </div> <div class="paragraph"> <p>In HBase 0.98.6 and newer, UTF-8 encoding is supported for visibility labels and expressions. When creating labels using the <code>addLabels(conf, labels)</code> method provided by the <code>org.apache.hadoop.hbase.security.visibility.VisibilityClient</code> class and passing labels in Authorizations via Scan or Get, labels can contain UTF-8 characters, as well as the logical operators normally used in visibility labels, with normal Java notations, without needing any escaping method. However, when you pass a CellVisibility expression via a Mutation, you must enclose the expression with the <code>CellVisibility.quote()</code> method if you use UTF-8 characters or logical operators. See <code>TestExpressionParser</code> and the source file <em>hbase-client/src/test/java/org/apache/hadoop/hbase/client/TestScan.java</em>.</p> </div> <div class="paragraph"> <p>A user adds visibility expressions to a cell during a Put operation. In the default configuration, the user does not need to have access to a label in order to label cells with it. This behavior is controlled by the configuration option <code>hbase.security.visibility.mutations.checkauths</code>. If you set this option to <code>true</code>, the labels the user is modifying as part of the mutation must be associated with the user, or the mutation will fail. Whether a user is authorized to read a labelled cell is determined during a Get or Scan, and results which the user is not allowed to read are filtered out. This incurs the same I/O penalty as if the results were returned, but reduces load on the network.</p> </div> <div class="paragraph"> <p>Visibility labels can also be specified during Delete operations. For details about visibility labels and Deletes, see <a href="https://issues.apache.org/jira/browse/HBASE-10885">HBASE-10885</a>.</p> </div> <div class="paragraph"> <p>The user’s effective label set is built in the RPC context when a request is first received by the RegionServer. The way that users are associated with labels is pluggable. The default plugin passes through labels specified in Authorizations added to the Get or Scan and checks those against the calling user’s authenticated labels list. When the client passes labels for which the user is not authenticated, the default plugin drops them. You can pass a subset of user authenticated labels via the <code>Get#setAuthorizations(Authorizations(String,…))</code> and <code>Scan#setAuthorizations(Authorizations(String,…));</code> methods.</p> </div> <div class="paragraph"> <p>Groups can be granted visibility labels the same way as users. Groups are prefixed with an @ symbol. When checking visibility labels of a user, the server will include the visibility labels of the groups of which the user is a member, together with the user’s own labels. When the visibility labels are retrieved using API <code>VisibilityClient#getAuths</code> or Shell command <code>get_auths</code> for a user, we will return labels added specifically for that user alone, not the group level labels.</p> </div> <div class="paragraph"> <p>Visibility label access checking is performed by the VisibilityController coprocessor. You can use interface <code>VisibilityLabelService</code> to provide a custom implementation and/or control the way that visibility labels are stored with cells. See the source file <em>hbase-server/src/test/java/org/apache/hadoop/hbase/security/visibility/TestVisibilityLabelsWithCustomVisLabService.java</em> for one example.</p> </div> <div class="paragraph"> <p>Visibility labels can be used in conjunction with ACLs.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> The labels have to be explicitly defined before they can be used in visibility labels. See below for an example of how this can be done. </td> </tr> </table> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> There is currently no way to determine which labels have been applied to a cell. See <a href="https://issues.apache.org/jira/browse/HBASE-12470">HBASE-12470</a> for details. </td> </tr> </table> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Visibility labels are not currently applied for superusers. </td> </tr> </table> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 8. Examples of Visibility Expressions</caption> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Expression</th> <th class="tableblock halign-left valign-top">Interpretation</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><div class="literal"><pre>fulltime</pre></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Allow accesss to users associated with the fulltime label.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><div class="literal"><pre>!public</pre></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Allow access to users not associated with the public label.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><div class="literal"><pre>( secret | topsecret ) & !probationary</pre></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Allow access to users associated with either the secret or topsecret label and not associated with the probationary label.</p></td> </tr> </tbody> </table> <div class="sect3"> <h4 id="_server_side_configuration_2"><a class="anchor" href="#_server_side_configuration_2"></a>60.3.1. Server-Side Configuration</h4> <div class="olist arabic"> <ol class="arabic"> <li> <p>As a prerequisite, perform the steps in <a href="#security.data.basic.server.side">[security.data.basic.server.side]</a>.</p> </li> <li> <p>Install and configure the VisibilityController coprocessor by setting the following properties in <em>hbase-site.xml</em>. These properties take a list of class names.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.visibility.VisibilityController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.master.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.visibility.VisibilityController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> If you use the AccessController and VisibilityController coprocessors together, the AccessController must come first in the list, because with both components active, the VisibilityController will delegate access control on its system tables to the AccessController. </td> </tr> </table> </div> </li> <li> <p>Adjust Configuration</p> <div class="paragraph"> <p>By default, users can label cells with any label, including labels they are not associated with, which means that a user can Put data that he cannot read. For example, a user could label a cell with the (hypothetical) 'topsecret' label even if the user is not associated with that label. If you only want users to be able to label cells with labels they are associated with, set <code>hbase.security.visibility.mutations.checkauths</code> to <code>true</code>. In that case, the mutation will fail if it makes use of labels the user is not associated with.</p> </div> </li> <li> <p>Distribute your configuration and restart your cluster for changes to take effect.</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="_administration_2"><a class="anchor" href="#_administration_2"></a>60.3.2. Administration</h4> <div class="paragraph"> <p>Administration tasks can be performed using the HBase Shell or the Java API. For defining the list of visibility labels and associating labels with users, the HBase Shell is probably simpler.</p> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> <div class="title">API Examples</div> <div class="paragraph"> <p>Many of the Java API examples in this section are taken from the source file <em>hbase-server/src/test/java/org/apache/hadoop/hbase/security/visibility/TestVisibilityLabels.java</em>. Refer to that file or the API documentation for more context.</p> </div> <div class="paragraph"> <p>Neither these examples, nor the source file they were taken from, are part of the public HBase API, and are provided for illustration only. Refer to the official API for usage instructions.</p> </div> </td> </tr> </table> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Define the List of Visibility Labels</p> <div class="exampleblock"> <div class="title">Example 25. HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> add_labels [ 'admin', 'service', 'developer', 'test' ]</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 26. Java API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> addLabels() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">PrivilegedExceptionAction</span><VisibilityLabelsResponse> action = <span class="keyword">new</span> <span class="predefined-type">PrivilegedExceptionAction</span><VisibilityLabelsResponse>() { <span class="directive">public</span> VisibilityLabelsResponse run() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">String</span><span class="type">[]</span> labels = { SECRET, TOPSECRET, CONFIDENTIAL, PUBLIC, PRIVATE, COPYRIGHT, ACCENT, UNICODE_VIS_TAG, UC1, UC2 }; <span class="keyword">try</span> { VisibilityClient.addLabels(conf, labels); } <span class="keyword">catch</span> (<span class="predefined-type">Throwable</span> t) { <span class="keyword">throw</span> <span class="keyword">new</span> <span class="exception">IOException</span>(t); } <span class="keyword">return</span> <span class="predefined-constant">null</span>; } }; SUPERUSER.runAs(action); }</code></pre> </div> </div> </div> </div> </li> <li> <p>Associate Labels with Users</p> <div class="exampleblock"> <div class="title">Example 27. HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> set_auths 'service', [ 'service' ]</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> set_auths 'testuser', [ 'test' ]</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> set_auths 'qa', [ 'test', 'developer' ]</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> set_auths '@qagroup', [ 'test' ]</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 28. Java API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">void</span> testSetAndGetUserAuths() <span class="directive">throws</span> <span class="predefined-type">Throwable</span> { <span class="directive">final</span> <span class="predefined-type">String</span> user = <span class="string"><span class="delimiter">"</span><span class="content">user1</span><span class="delimiter">"</span></span>; <span class="predefined-type">PrivilegedExceptionAction</span><<span class="predefined-type">Void</span>> action = <span class="keyword">new</span> <span class="predefined-type">PrivilegedExceptionAction</span><<span class="predefined-type">Void</span>>() { <span class="directive">public</span> <span class="predefined-type">Void</span> run() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">String</span><span class="type">[]</span> auths = { SECRET, CONFIDENTIAL }; <span class="keyword">try</span> { VisibilityClient.setAuths(conf, auths, user); } <span class="keyword">catch</span> (<span class="predefined-type">Throwable</span> e) { } <span class="keyword">return</span> <span class="predefined-constant">null</span>; } ...</code></pre> </div> </div> </div> </div> </li> <li> <p>Clear Labels From Users</p> <div class="exampleblock"> <div class="title">Example 29. HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> clear_auths 'service', [ 'service' ]</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> clear_auths 'testuser', [ 'test' ]</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> clear_auths 'qa', [ 'test', 'developer' ]</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> clear_auths '@qagroup', [ 'test', 'developer' ]</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 30. Java API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... auths = <span class="keyword">new</span> <span class="predefined-type">String</span><span class="type">[]</span> { SECRET, PUBLIC, CONFIDENTIAL }; VisibilityLabelsResponse response = <span class="predefined-constant">null</span>; <span class="keyword">try</span> { response = VisibilityClient.clearAuths(conf, auths, user); } <span class="keyword">catch</span> (<span class="predefined-type">Throwable</span> e) { fail(<span class="string"><span class="delimiter">"</span><span class="content">Should not have failed</span><span class="delimiter">"</span></span>); ... }</code></pre> </div> </div> </div> </div> </li> <li> <p>Apply a Label or Expression to a Cell</p> <div class="paragraph"> <p>The label is only applied when data is written. The label is associated with a given version of the cell.</p> </div> <div class="exampleblock"> <div class="title">Example 31. HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> set_visibility 'user', 'admin|service|developer', { COLUMNS => 'i' }</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> set_visibility 'user', 'admin|service', { COLUMNS => 'pii' }</pre> </div> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> set_visibility 'user', 'test', { COLUMNS => [ 'i', 'pii' ], FILTER => "(PrefixFilter ('test'))" }</pre> </div> </div> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> HBase Shell support for applying labels or permissions to cells is for testing and verification support, and should not be employed for production use because it won’t apply the labels to cells that don’t exist yet. The correct way to apply cell level labels is to do so in the application code when storing the values. </td> </tr> </table> </div> <div class="exampleblock"> <div class="title">Example 32. Java API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">static</span> Table createTableAndWriteDataWithLabels(TableName tableName, <span class="predefined-type">String</span>... labelExps) <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">Configuration</span> conf = HBaseConfiguration.create(); <span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf); Table table = NULL; <span class="keyword">try</span> { table = TEST_UTIL.createTable(tableName, fam); <span class="type">int</span> i = <span class="integer">1</span>; <span class="predefined-type">List</span><Put> puts = <span class="keyword">new</span> <span class="predefined-type">ArrayList</span><Put>(); <span class="keyword">for</span> (<span class="predefined-type">String</span> labelExp : labelExps) { Put put = <span class="keyword">new</span> Put(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">row</span><span class="delimiter">"</span></span> + i)); put.add(fam, qual, HConstants.LATEST_TIMESTAMP, value); put.setCellVisibility(<span class="keyword">new</span> CellVisibility(labelExp)); puts.add(put); i++; } table.put(puts); } <span class="keyword">finally</span> { <span class="keyword">if</span> (table != <span class="predefined-constant">null</span>) { table.flushCommits(); } }</code></pre> </div> </div> </div> </div> </li> </ol> </div> <div class="paragraph"> <p><a href="#reading_cells_with_labels">[reading_cells_with_labels]</a> ==== Reading Cells with Labels When you issue a Scan or Get, HBase uses your default set of authorizations to filter out cells that you do not have access to. A superuser can set the default set of authorizations for a given user by using the <code>set_auths</code> HBase Shell command or the <a href="http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/security/visibility/VisibilityClient.html#setAuths(org.apache.hadoop.conf.Configuration,%20java.lang.String\">],%20java.lang.String)[VisibilityClient.setAuths()</a> method.</p> </div> <div class="paragraph"> <p>You can specify a different authorization during the Scan or Get, by passing the AUTHORIZATIONS option in HBase Shell, or the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html#setAuthorizations%28org.apache.hadoop.hbase.security.visibility.Authorizations%29">setAuthorizations()</a> method if you use the API. This authorization will be combined with your default set as an additional filter. It will further filter your results, rather than giving you additional authorization.</p> </div> <div class="exampleblock"> <div class="title">Example 33. HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> get_auths 'myUser' hbase> scan 'table1', AUTHORIZATIONS => ['private']</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 34. Java API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... public <span class="predefined-type">Void</span> run() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="predefined-type">String</span><span class="type">[]</span> auths1 = { SECRET, CONFIDENTIAL }; GetAuthsResponse authsResponse = <span class="predefined-constant">null</span>; <span class="keyword">try</span> { VisibilityClient.setAuths(conf, auths1, user); <span class="keyword">try</span> { authsResponse = VisibilityClient.getAuths(conf, user); } <span class="keyword">catch</span> (<span class="predefined-type">Throwable</span> e) { fail(<span class="string"><span class="delimiter">"</span><span class="content">Should not have failed</span><span class="delimiter">"</span></span>); } } <span class="keyword">catch</span> (<span class="predefined-type">Throwable</span> e) { } <span class="predefined-type">List</span><<span class="predefined-type">String</span>> authsList = <span class="keyword">new</span> <span class="predefined-type">ArrayList</span><<span class="predefined-type">String</span>>(); <span class="keyword">for</span> (ByteString authBS : authsResponse.getAuthList()) { authsList.add(Bytes.toString(authBS.toByteArray())); } assertEquals(<span class="integer">2</span>, authsList.size()); assertTrue(authsList.contains(SECRET)); assertTrue(authsList.contains(CONFIDENTIAL)); <span class="keyword">return</span> <span class="predefined-constant">null</span>; } ...</code></pre> </div> </div> </div> </div> </div> <div class="sect3"> <h4 id="_implementing_your_own_visibility_label_algorithm"><a class="anchor" href="#_implementing_your_own_visibility_label_algorithm"></a>60.3.3. Implementing Your Own Visibility Label Algorithm</h4> <div class="paragraph"> <p>Interpreting the labels authenticated for a given get/scan request is a pluggable algorithm.</p> </div> <div class="paragraph"> <p>You can specify a custom plugin or plugins by using the property <code>hbase.regionserver.scan.visibility.label.generator.class</code>. The output for the first <code>ScanLabelGenerator</code> will be the input for the next one, until the end of the list.</p> </div> <div class="paragraph"> <p>The default implementation, which was implemented in <a href="https://issues.apache.org/jira/browse/HBASE-12466">HBASE-12466</a>, loads two plugins, <code>FeedUserAuthScanLabelGenerator</code> and <code>DefinedSetFilterScanLabelGenerator</code>. See <a href="#reading_cells_with_labels">[reading_cells_with_labels]</a>.</p> </div> </div> <div class="sect3"> <h4 id="_replicating_visibility_tags_as_strings"><a class="anchor" href="#_replicating_visibility_tags_as_strings"></a>60.3.4. Replicating Visibility Tags as Strings</h4> <div class="paragraph"> <p>As mentioned in the above sections, the interface <code>VisibilityLabelService</code> could be used to implement a different way of storing the visibility expressions in the cells. Clusters with replication enabled also must replicate the visibility expressions to the peer cluster. If <code>DefaultVisibilityLabelServiceImpl</code> is used as the implementation for <code>VisibilityLabelService</code>, all the visibility expression are converted to the corresponding expression based on the ordinals for each visibility label stored in the labels table. During replication, visible cells are also replicated with the ordinal-based expression intact. The peer cluster may not have the same <code>labels</code> table with the same ordinal mapping for the visibility labels. In that case, replicating the ordinals makes no sense. It would be better if the replication occurred with the visibility expressions transmitted as strings. To replicate the visibility expression as strings to the peer cluster, create a <code>RegionServerObserver</code> configuration which works based on the implementation of the <code>VisibilityLabelService</code> interface. The configuration below enables replication of visibility expressions to peer clusters as strings. See <a href="https://issues.apache.org/jira/browse/HBASE-11639">HBASE-11639</a> for more details.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.regionserver.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.visibility.VisibilityController$VisibilityReplication<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="hbase.encryption.server"><a class="anchor" href="#hbase.encryption.server"></a>60.4. Transparent Encryption of Data At Rest</h3> <div class="paragraph"> <p>HBase provides a mechanism for protecting your data at rest, in HFiles and the WAL, which reside within HDFS or another distributed filesystem. A two-tier architecture is used for flexible and non-intrusive key rotation. "Transparent" means that no implementation changes are needed on the client side. When data is written, it is encrypted. When it is read, it is decrypted on demand.</p> </div> <div class="sect3"> <h4 id="_how_it_works_2"><a class="anchor" href="#_how_it_works_2"></a>60.4.1. How It Works</h4> <div class="paragraph"> <p>The administrator provisions a master key for the cluster, which is stored in a key provider accessible to every trusted HBase process, including the HMaster, RegionServers, and clients (such as HBase Shell) on administrative workstations. The default key provider is integrated with the Java KeyStore API and any key management systems with support for it. Other custom key provider implementations are possible. The key retrieval mechanism is configured in the <em>hbase-site.xml</em> configuration file. The master key may be stored on the cluster servers, protected by a secure KeyStore file, or on an external keyserver, or in a hardware security module. This master key is resolved as needed by HBase processes through the configured key provider.</p> </div> <div class="paragraph"> <p>Next, encryption use can be specified in the schema, per column family, by creating or modifying a column descriptor to include two additional attributes: the name of the encryption algorithm to use (currently only "AES" is supported), and optionally, a data key wrapped (encrypted) with the cluster master key. If a data key is not explicitly configured for a ColumnFamily, HBase will create a random data key per HFile. This provides an incremental improvement in security over the alternative. Unless you need to supply an explicit data key, such as in a case where you are generating encrypted HFiles for bulk import with a given data key, only specify the encryption algorithm in the ColumnFamily schema metadata and let HBase create data keys on demand. Per Column Family keys facilitate low impact incremental key rotation and reduce the scope of any external leak of key material. The wrapped data key is stored in the ColumnFamily schema metadata, and in each HFile for the Column Family, encrypted with the cluster master key. After the Column Family is configured for encryption, any new HFiles will be written encrypted. To ensure encryption of all HFiles, trigger a major compaction after enabling this feature.</p> </div> <div class="paragraph"> <p>When the HFile is opened, the data key is extracted from the HFile, decrypted with the cluster master key, and used for decryption of the remainder of the HFile. The HFile will be unreadable if the master key is not available. If a remote user somehow acquires access to the HFile data because of some lapse in HDFS permissions, or from inappropriately discarded media, it will not be possible to decrypt either the data key or the file data.</p> </div> <div class="paragraph"> <p>It is also possible to encrypt the WAL. Even though WALs are transient, it is necessary to encrypt the WALEdits to avoid circumventing HFile protections for encrypted column families, in the event that the underlying filesystem is compromised. When WAL encryption is enabled, all WALs are encrypted, regardless of whether the relevant HFiles are encrypted.</p> </div> </div> <div class="sect3"> <h4 id="_server_side_configuration_3"><a class="anchor" href="#_server_side_configuration_3"></a>60.4.2. Server-Side Configuration</h4> <div class="paragraph"> <p>This procedure assumes you are using the default Java keystore implementation. If you are using a custom implementation, check its documentation and adjust accordingly.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Create a secret key of appropriate length for AES encryption, using the <code>keytool</code> utility.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ keytool -keystore /path/to/hbase/conf/hbase.jks \ -storetype jceks -storepass **** \ -genseckey -keyalg AES -keysize 128 \ -alias <alias></code></pre> </div> </div> <div class="paragraph"> <p>Replace <em class="replaceable">****</em> with the password for the keystore file and <alias> with the username of the HBase service account, or an arbitrary string. If you use an arbitrary string, you will need to configure HBase to use it, and that is covered below. Specify a keysize that is appropriate. Do not specify a separate password for the key, but press <kbd>Return</kbd> when prompted.</p> </div> </li> <li> <p>Set appropriate permissions on the keyfile and distribute it to all the HBase servers.</p> <div class="paragraph"> <p>The previous command created a file called <em>hbase.jks</em> in the HBase <em>conf/</em> directory. Set the permissions and ownership on this file such that only the HBase service account user can read the file, and securely distribute the key to all HBase servers.</p> </div> </li> <li> <p>Configure the HBase daemons.</p> <div class="paragraph"> <p>Set the following properties in <em>hbase-site.xml</em> on the region servers, to configure HBase daemons to use a key provider backed by the KeyStore file or retrieving the cluster master key. In the example below, replace <em class="replaceable">****</em> with the password.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.keyprovider<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.io.crypto.KeyStoreKeyProvider<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.keyprovider.parameters<span class="tag"></name></span> <span class="tag"><value></span>jceks:///path/to/hbase/conf/hbase.jks?password=****<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>By default, the HBase service account name will be used to resolve the cluster master key. However, you can store it with an arbitrary alias (in the <code>keytool</code> command). In that case, set the following property to the alias you used.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.master.key.name<span class="tag"></name></span> <span class="tag"><value></span>my-alias<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>You also need to be sure your HFiles use HFile v3, in order to use transparent encryption. This is the default configuration for HBase 1.0 onward. For previous versions, set the following property in your <em>hbase-site.xml</em> file.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hfile.format.version<span class="tag"></name></span> <span class="tag"><value></span>3<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Optionally, you can use a different cipher provider, either a Java Cryptography Encryption (JCE) algorithm provider or a custom HBase cipher implementation.</p> </div> <div class="ulist"> <ul> <li> <p>JCE:</p> <div class="ulist"> <ul> <li> <p>Install a signed JCE provider (supporting <code>AES/CTR/NoPadding</code> mode with 128 bit keys)</p> </li> <li> <p>Add it with highest preference to the JCE site configuration file <em>$JAVA_HOME/lib/security/java.security</em>.</p> </li> <li> <p>Update <code>hbase.crypto.algorithm.aes.provider</code> and <code>hbase.crypto.algorithm.rng.provider</code> options in <em class="path">hbase-site.xml</em>.</p> </li> </ul> </div> </li> <li> <p>Custom HBase Cipher:</p> <div class="ulist"> <ul> <li> <p>Implement <code>org.apache.hadoop.hbase.io.crypto.CipherProvider</code>.</p> </li> <li> <p>Add the implementation to the server classpath.</p> </li> <li> <p>Update <code>hbase.crypto.cipherprovider</code> in <em>hbase-site.xml</em>.</p> </li> </ul> </div> </li> </ul> </div> </li> <li> <p>Configure WAL encryption.</p> <div class="paragraph"> <p>Configure WAL encryption in every RegionServer’s <em>hbase-site.xml</em>, by setting the following properties. You can include these in the HMaster’s <em>hbase-site.xml</em> as well, but the HMaster does not have a WAL and will not use them.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.hlog.reader.impl<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.regionserver.wal.SecureProtobufLogReader<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.hlog.writer.impl<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.regionserver.wal.SecureProtobufLogWriter<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.wal.encryption<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </li> <li> <p>Configure permissions on the <em>hbase-site.xml</em> file.</p> <div class="paragraph"> <p>Because the keystore password is stored in the hbase-site.xml, you need to ensure that only the HBase user can read the <em>hbase-site.xml</em> file, using file ownership and permissions.</p> </div> </li> <li> <p>Restart your cluster.</p> <div class="paragraph"> <p>Distribute the new configuration file to all nodes and restart your cluster.</p> </div> </li> </ol> </div> </div> <div class="sect3"> <h4 id="_administration_3"><a class="anchor" href="#_administration_3"></a>60.4.3. Administration</h4> <div class="paragraph"> <p>Administrative tasks can be performed in HBase Shell or the Java API.</p> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> <div class="title">Java API</div> <div class="paragraph"> <p>Java API examples in this section are taken from the source file <em>hbase-server/src/test/java/org/apache/hadoop/hbase/util/TestHBaseFsckEncryption.java</em>. .</p> </div> <div class="paragraph"> <p>Neither these examples, nor the source files they are taken from, are part of the public HBase API, and are provided for illustration only. Refer to the official API for usage instructions.</p> </div> </td> </tr> </table> </div> <div class="dlist"> <dl> <dt class="hdlist1">Enable Encryption on a Column Family</dt> <dd> <p>To enable encryption on a column family, you can either use HBase Shell or the Java API. After enabling encryption, trigger a major compaction. When the major compaction completes, the HFiles will be encrypted.</p> </dd> <dt class="hdlist1">Rotate the Data Key</dt> <dd> <p>To rotate the data key, first change the ColumnFamily key in the column descriptor, then trigger a major compaction. When compaction is complete, all HFiles will be re-encrypted using the new data key. Until the compaction completes, the old HFiles will still be readable using the old key.</p> </dd> <dt class="hdlist1">Switching Between Using a Random Data Key and Specifying A Key</dt> <dd> <p>If you configured a column family to use a specific key and you want to return to the default behavior of using a randomly-generated key for that column family, use the Java API to alter the <code>HColumnDescriptor</code> so that no value is sent with the key <code>ENCRYPTION_KEY</code>.</p> </dd> <dt class="hdlist1">Rotate the Master Key</dt> <dd> <p>To rotate the master key, first generate and distribute the new key. Then update the KeyStore to contain a new master key, and keep the old master key in the KeyStore using a different alias. Next, configure fallback to the old master key in the <em>hbase-site.xml</em> file.</p> </dd> <dt class="hdlist1"></dt> </dl> </div> </div> </div> <div class="sect2"> <h3 id="hbase.secure.bulkload"><a class="anchor" href="#hbase.secure.bulkload"></a>60.5. Secure Bulk Load</h3> <div class="paragraph"> <p>Bulk loading in secure mode is a bit more involved than normal setup, since the client has to transfer the ownership of the files generated from the MapReduce job to HBase. Secure bulk loading is implemented by a coprocessor, named <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/security/access/SecureBulkLoadEndpoint.html">SecureBulkLoadEndpoint</a>, which uses a staging directory configured by the configuration property <code>hbase.bulkload.staging.dir</code>, which defaults to <em>/tmp/hbase-staging/</em>.</p> </div> <div class="ulist"> <div class="title">Secure Bulk Load Algorithm</div> <ul> <li> <p>One time only, create a staging directory which is world-traversable and owned by the user which runs HBase (mode 711, or <code>rwx—x—x</code>). A listing of this directory will look similar to the following:</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ls -ld /tmp/hbase-staging drwx--x--x 2 hbase hbase 68 3 Sep 14:54 /tmp/hbase-staging</code></pre> </div> </div> </li> <li> <p>A user writes out data to a secure output directory owned by that user. For example, <em>/user/foo/data</em>.</p> </li> <li> <p>Internally, HBase creates a secret staging directory which is globally readable/writable (<code>-rwxrwxrwx, 777</code>). For example, <em>/tmp/hbase-staging/averylongandrandomdirectoryname</em>. The name and location of this directory is not exposed to the user. HBase manages creation and deletion of this directory.</p> </li> <li> <p>The user makes the data world-readable and world-writable, moves it into the random staging directory, then calls the <code>SecureBulkLoadClient#bulkLoadHFiles</code> method.</p> </li> </ul> </div> <div class="paragraph"> <p>The strength of the security lies in the length and randomness of the secret directory.</p> </div> <div class="paragraph"> <p>To enable secure bulk load, add the following properties to <em>hbase-site.xml</em>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.bulkload.staging.dir<span class="tag"></name></span> <span class="tag"><value></span>/tmp/hbase-staging<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.token.TokenProvider, org.apache.hadoop.hbase.security.access.AccessController,org.apache.hadoop.hbase.security.access.SecureBulkLoadEndpoint<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="security.example.config"><a class="anchor" href="#security.example.config"></a>61. Security Configuration Example</h2> <div class="sectionbody"> <div class="paragraph"> <p>This configuration example includes support for HFile v3, ACLs, Visibility Labels, and transparent encryption of data at rest and the WAL. All options have been discussed separately in the sections above.</p> </div> <div class="exampleblock"> <div class="title">Example 35. Example Security Settings in <em>hbase-site.xml</em></div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hfile.format.version<span class="tag"></name></span> <span class="tag"><value></span>3<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.superuser<span class="tag"></name></span> <span class="tag"><value></span>hbase, admin<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController, org.apache.hadoop.hbase.security.visibility.VisibilityController, org.apache.hadoop.hbase.security.token.TokenProvider<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.master.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.access.AccessController, org.apache.hadoop.hbase.security.visibility.VisibilityController<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.regionserver.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop/hbase.security.access.AccessController, org.apache.hadoop.hbase.security.access.VisibilityController<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.exec.permission.checks<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.visibility.mutations.checkauth<span class="tag"></name></span> <span class="tag"><value></span>false<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.rpc.protection<span class="tag"></name></span> <span class="tag"><value></span>privacy<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.keyprovider<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.io.crypto.KeyStoreKeyProvider<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.keyprovider.parameters<span class="tag"></name></span> <span class="tag"><value></span>jceks:///path/to/hbase/conf/hbase.jks?password=***<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.master.key.name<span class="tag"></name></span> <span class="tag"><value></span>hbase<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.hlog.reader.impl<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.regionserver.wal.SecureProtobufLogReader<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.hlog.writer.impl<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.regionserver.wal.SecureProtobufLogWriter<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.wal.encryption<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.crypto.master.alternate.key.name<span class="tag"></name></span> <span class="tag"><value></span>hbase.old<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.bulkload.staging.dir<span class="tag"></name></span> <span class="tag"><value></span>/tmp/hbase-staging<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.security.token.TokenProvider, org.apache.hadoop.hbase.security.access.AccessController,org.apache.hadoop.hbase.security.access.SecureBulkLoadEndpoint<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.security.authorization<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 36. Example Group Mapper in Hadoop <em>core-site.xml</em></div> <div class="content"> <div class="paragraph"> <p>Adjust these settings to suit your environment.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.security.LdapGroupsMapping<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.url<span class="tag"></name></span> <span class="tag"><value></span>ldap://server<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.bind.user<span class="tag"></name></span> <span class="tag"><value></span>Administrator@example-ad.local<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.bind.password<span class="tag"></name></span> <span class="tag"><value></span>****<span class="tag"></value></span> <span class="comment"></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.base<span class="tag"></name></span> <span class="tag"><value></span>dc=example-ad,dc=local<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.filter.user<span class="tag"></name></span> <span class="tag"><value></span>(<span class="entity">&amp;</span>(objectClass=user)(sAMAccountName={0}))<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.filter.group<span class="tag"></name></span> <span class="tag"><value></span>(objectClass=group)<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.attr.member<span class="tag"></name></span> <span class="tag"><value></span>member<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hadoop.security.group.mapping.ldap.search.attr.group.name<span class="tag"></name></span> <span class="tag"><value></span>cn<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> </div> </div> <h1 id="_architecture" class="sect0"><a class="anchor" href="#_architecture"></a>Architecture</h1> <div class="sect1"> <h2 id="arch.overview"><a class="anchor" href="#arch.overview"></a>62. Overview</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="arch.overview.nosql"><a class="anchor" href="#arch.overview.nosql"></a>62.1. NoSQL?</h3> <div class="paragraph"> <p>HBase is a type of "NoSQL" database. "NoSQL" is a general term meaning that the database isn’t an RDBMS which supports SQL as its primary access language, but there are many types of NoSQL databases: BerkeleyDB is an example of a local NoSQL database, whereas HBase is very much a distributed database. Technically speaking, HBase is really more a "Data Store" than "Data Base" because it lacks many of the features you find in an RDBMS, such as typed columns, secondary indexes, triggers, and advanced query languages, etc.</p> </div> <div class="paragraph"> <p>However, HBase has many features which supports both linear and modular scaling. HBase clusters expand by adding RegionServers that are hosted on commodity class servers. If a cluster expands from 10 to 20 RegionServers, for example, it doubles both in terms of storage and as well as processing capacity. RDBMS can scale well, but only up to a point - specifically, the size of a single database server - and for the best performance requires specialized hardware and storage devices. HBase features of note are:</p> </div> <div class="ulist"> <ul> <li> <p>Strongly consistent reads/writes: HBase is not an "eventually consistent" DataStore. This makes it very suitable for tasks such as high-speed counter aggregation.</p> </li> <li> <p>Automatic sharding: HBase tables are distributed on the cluster via regions, and regions are automatically split and re-distributed as your data grows.</p> </li> <li> <p>Automatic RegionServer failover</p> </li> <li> <p>Hadoop/HDFS Integration: HBase supports HDFS out of the box as its distributed file system.</p> </li> <li> <p>MapReduce: HBase supports massively parallelized processing via MapReduce for using HBase as both source and sink.</p> </li> <li> <p>Java Client API: HBase supports an easy to use Java API for programmatic access.</p> </li> <li> <p>Thrift/REST API: HBase also supports Thrift and REST for non-Java front-ends.</p> </li> <li> <p>Block Cache and Bloom Filters: HBase supports a Block Cache and Bloom Filters for high volume query optimization.</p> </li> <li> <p>Operational Management: HBase provides build-in web-pages for operational insight as well as JMX metrics.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="arch.overview.when"><a class="anchor" href="#arch.overview.when"></a>62.2. When Should I Use HBase?</h3> <div class="paragraph"> <p>HBase isn’t suitable for every problem.</p> </div> <div class="paragraph"> <p>First, make sure you have enough data. If you have hundreds of millions or billions of rows, then HBase is a good candidate. If you only have a few thousand/million rows, then using a traditional RDBMS might be a better choice due to the fact that all of your data might wind up on a single node (or two) and the rest of the cluster may be sitting idle.</p> </div> <div class="paragraph"> <p>Second, make sure you can live without all the extra features that an RDBMS provides (e.g., typed columns, secondary indexes, transactions, advanced query languages, etc.) An application built against an RDBMS cannot be "ported" to HBase by simply changing a JDBC driver, for example. Consider moving from an RDBMS to HBase as a complete redesign as opposed to a port.</p> </div> <div class="paragraph"> <p>Third, make sure you have enough hardware. Even HDFS doesn’t do well with anything less than 5 DataNodes (due to things such as HDFS block replication which has a default of 3), plus a NameNode.</p> </div> <div class="paragraph"> <p>HBase can run quite well stand-alone on a laptop - but this should be considered a development configuration only.</p> </div> </div> <div class="sect2"> <h3 id="arch.overview.hbasehdfs"><a class="anchor" href="#arch.overview.hbasehdfs"></a>62.3. What Is The Difference Between HBase and Hadoop/HDFS?</h3> <div class="paragraph"> <p><a href="http://hadoop.apache.org/hdfs/">HDFS</a> is a distributed file system that is well suited for the storage of large files. Its documentation states that it is not, however, a general purpose file system, and does not provide fast individual record lookups in files. HBase, on the other hand, is built on top of HDFS and provides fast record lookups (and updates) for large tables. This can sometimes be a point of conceptual confusion. HBase internally puts your data in indexed "StoreFiles" that exist on HDFS for high-speed lookups. See the <a href="#datamodel">Data Model</a> and the rest of this chapter for more information on how HBase achieves its goals.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="arch.catalog"><a class="anchor" href="#arch.catalog"></a>63. Catalog Tables</h2> <div class="sectionbody"> <div class="paragraph"> <p>The catalog table <code>hbase:meta</code> exists as an HBase table and is filtered out of the HBase shell’s <code>list</code> command, but is in fact a table just like any other.</p> </div> <div class="sect2"> <h3 id="arch.catalog.root"><a class="anchor" href="#arch.catalog.root"></a>63.1. -ROOT-</h3> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> The <code>-ROOT-</code> table was removed in HBase 0.96.0. Information here should be considered historical. </td> </tr> </table> </div> <div class="paragraph"> <p>The <code>-ROOT-</code> table kept track of the location of the <code>.META</code> table (the previous name for the table now called <code>hbase:meta</code>) prior to HBase 0.96. The <code>-ROOT-</code> table structure was as follows:</p> </div> <div class="ulist"> <div class="title">Key</div> <ul> <li> <p>.META. region key (<code>.META.,,1</code>)</p> </li> </ul> </div> <div class="ulist"> <div class="title">Values</div> <ul> <li> <p><code>info:regioninfo</code> (serialized <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HRegionInfo.html">HRegionInfo</a> instance of <code>hbase:meta</code>)</p> </li> <li> <p><code>info:server</code> (server:port of the RegionServer holding <code>hbase:meta</code>)</p> </li> <li> <p><code>info:serverstartcode</code> (start-time of the RegionServer process holding <code>hbase:meta</code>)</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="arch.catalog.meta"><a class="anchor" href="#arch.catalog.meta"></a>63.2. hbase:meta</h3> <div class="paragraph"> <p>The <code>hbase:meta</code> table (previously called <code>.META.</code>) keeps a list of all regions in the system. The location of <code>hbase:meta</code> was previously tracked within the <code>-ROOT-</code> table, but is now stored in ZooKeeper.</p> </div> <div class="paragraph"> <p>The <code>hbase:meta</code> table structure is as follows:</p> </div> <div class="ulist"> <div class="title">Key</div> <ul> <li> <p>Region key of the format (<code>[table],[region start key],[region id]</code>)</p> </li> </ul> </div> <div class="ulist"> <div class="title">Values</div> <ul> <li> <p><code>info:regioninfo</code> (serialized <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HRegionInfo.html">HRegionInfo</a> instance for this region)</p> </li> <li> <p><code>info:server</code> (server:port of the RegionServer containing this region)</p> </li> <li> <p><code>info:serverstartcode</code> (start-time of the RegionServer process containing this region)</p> </li> </ul> </div> <div class="paragraph"> <p>When a table is in the process of splitting, two other columns will be created, called <code>info:splitA</code> and <code>info:splitB</code>. These columns represent the two daughter regions. The values for these columns are also serialized HRegionInfo instances. After the region has been split, eventually this row will be deleted.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Note on HRegionInfo</div> <div class="paragraph"> <p>The empty key is used to denote table start and table end. A region with an empty start key is the first region in a table. If a region has both an empty start and an empty end key, it is the only region in the table</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>In the (hopefully unlikely) event that programmatic processing of catalog metadata is required, see the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/util/Writables.html#getHRegionInfo%28byte" class="bare">http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/util/Writables.html#getHRegionInfo%28byte</a>%29[Writables] utility.</p> </div> </div> <div class="sect2"> <h3 id="arch.catalog.startup"><a class="anchor" href="#arch.catalog.startup"></a>63.3. Startup Sequencing</h3> <div class="paragraph"> <p>First, the location of <code>hbase:meta</code> is looked up in ZooKeeper. Next, <code>hbase:meta</code> is updated with server and startcode values.</p> </div> <div class="paragraph"> <p>For information on region-RegionServer assignment, see <a href="#regions.arch.assignment">Region-RegionServer Assignment</a>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="architecture.client"><a class="anchor" href="#architecture.client"></a>64. Client</h2> <div class="sectionbody"> <div class="paragraph"> <p>The HBase client finds the RegionServers that are serving the particular row range of interest. It does this by querying the <code>hbase:meta</code> table. See <a href="#arch.catalog.meta">hbase:meta</a> for details. After locating the required region(s), the client contacts the RegionServer serving that region, rather than going through the master, and issues the read or write request. This information is cached in the client so that subsequent requests need not go through the lookup process. Should a region be reassigned either by the master load balancer or because a RegionServer has died, the client will requery the catalog tables to determine the new location of the user region.</p> </div> <div class="paragraph"> <p>See <a href="#master.runtime">Runtime Impact</a> for more information about the impact of the Master on HBase Client communication.</p> </div> <div class="paragraph"> <p>Administrative functions are done via an instance of <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Admin.html">Admin</a></p> </div> <div class="sect2"> <h3 id="client.connections"><a class="anchor" href="#client.connections"></a>64.1. Cluster Connections</h3> <div class="paragraph"> <p>The API changed in HBase 1.0. For connection configuration information, see <a href="#client_dependencies">Client configuration and dependencies connecting to an HBase cluster</a>.</p> </div> <div class="sect3"> <h4 id="_api_as_of_hbase_1_0_0"><a class="anchor" href="#_api_as_of_hbase_1_0_0"></a>64.1.1. API as of HBase 1.0.0</h4> <div class="paragraph"> <p>Its been cleaned up and users are returned Interfaces to work against rather than particular types. In HBase 1.0, obtain a <code>Connection</code> object from <code>ConnectionFactory</code> and thereafter, get from it instances of <code>Table</code>, <code>Admin</code>, and <code>RegionLocator</code> on an as-need basis. When done, close the obtained instances. Finally, be sure to cleanup your <code>Connection</code> instance before exiting. <code>Connections</code> are heavyweight objects but thread-safe so you can create one for your application and keep the instance around. <code>Table</code>, <code>Admin</code> and <code>RegionLocator</code> instances are lightweight. Create as you go and then let go as soon as you are done by closing them. See the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/package-summary.html">Client Package Javadoc Description</a> for example usage of the new HBase 1.0 API.</p> </div> </div> <div class="sect3"> <h4 id="_api_before_hbase_1_0_0"><a class="anchor" href="#_api_before_hbase_1_0_0"></a>64.1.2. API before HBase 1.0.0</h4> <div class="paragraph"> <p>Instances of <code>HTable</code> are the way to interact with an HBase cluster earlier than 1.0.0. <em><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html">Table</a> instances are not thread-safe</em>. Only one thread can use an instance of Table at any given time. When creating Table instances, it is advisable to use the same <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HBaseConfiguration">HBaseConfiguration</a> instance. This will ensure sharing of ZooKeeper and socket instances to the RegionServers which is usually what you want. For example, this is preferred:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HBaseConfiguration conf = HBaseConfiguration.create(); HTable table1 = <span class="keyword">new</span> HTable(conf, <span class="string"><span class="delimiter">"</span><span class="content">myTable</span><span class="delimiter">"</span></span>); HTable table2 = <span class="keyword">new</span> HTable(conf, <span class="string"><span class="delimiter">"</span><span class="content">myTable</span><span class="delimiter">"</span></span>);</code></pre> </div> </div> <div class="paragraph"> <p>as opposed to this:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HBaseConfiguration conf1 = HBaseConfiguration.create(); HTable table1 = <span class="keyword">new</span> HTable(conf1, <span class="string"><span class="delimiter">"</span><span class="content">myTable</span><span class="delimiter">"</span></span>); HBaseConfiguration conf2 = HBaseConfiguration.create(); HTable table2 = <span class="keyword">new</span> HTable(conf2, <span class="string"><span class="delimiter">"</span><span class="content">myTable</span><span class="delimiter">"</span></span>);</code></pre> </div> </div> <div class="paragraph"> <p>For more information about how connections are handled in the HBase client, see <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/ConnectionFactory.html">ConnectionFactory</a>.</p> </div> <div class="sect4"> <h5 id="client.connection.pooling"><a class="anchor" href="#client.connection.pooling"></a>Connection Pooling</h5> <div class="paragraph"> <p>For applications which require high-end multithreaded access (e.g., web-servers or application servers that may serve many application threads in a single JVM), you can pre-create a <code>Connection</code>, as shown in the following example:</p> </div> <div class="exampleblock"> <div class="title">Example 37. Pre-Creating a <code>Connection</code></div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="comment">// Create a connection to the cluster.</span> <span class="predefined-type">Configuration</span> conf = HBaseConfiguration.create(); <span class="keyword">try</span> (<span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(conf)) { <span class="keyword">try</span> (Table table = connection.getTable(TableName.valueOf(tablename)) { <span class="comment">// use table as needed, the table returned is lightweight</span> } }</code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>Constructing HTableInterface implementation is very lightweight and resources are controlled.</p> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="title"><code>HTablePool</code> is Deprecated</div> <div class="paragraph"> <p>Previous versions of this guide discussed <code>HTablePool</code>, which was deprecated in HBase 0.94, 0.95, and 0.96, and removed in 0.98.1, by <a href="https://issues.apache.org/jira/browse/HBASE-6580">HBASE-6500</a>, or <code>HConnection</code>, which is deprecated in HBase 1.0 by <code>Connection</code>. Please use <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Connection.html">Connection</a> instead.</p> </div> </td> </tr> </table> </div> </div> </div> </div> <div class="sect2"> <h3 id="client.writebuffer"><a class="anchor" href="#client.writebuffer"></a>64.2. WriteBuffer and Batch Methods</h3> <div class="paragraph"> <p>In HBase 1.0 and later, <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/HTable.html">HTable</a> is deprecated in favor of <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html">Table</a>. <code>Table</code> does not use autoflush. To do buffered writes, use the BufferedMutator class.</p> </div> <div class="paragraph"> <p>Before a <code>Table</code> or <code>HTable</code> instance is discarded, invoke either <code>close()</code> or <code>flushCommits()</code>, so `Put`s will not be lost.</p> </div> <div class="paragraph"> <p>For additional information on write durability, review the <a href="../acid-semantics.html">ACID semantics</a> page.</p> </div> <div class="paragraph"> <p>For fine-grained control of batching of <code>Put</code>s or <code>Delete</code>s, see the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#batch%28java.util.List%29">batch</a> methods on Table.</p> </div> </div> <div class="sect2"> <h3 id="client.external"><a class="anchor" href="#client.external"></a>64.3. External Clients</h3> <div class="paragraph"> <p>Information on non-Java clients and custom protocols is covered in <a href="#external_apis">Apache HBase External APIs</a></p> </div> </div> </div> </div> <div class="sect1"> <h2 id="client.filter"><a class="anchor" href="#client.filter"></a>65. Client Request Filters</h2> <div class="sectionbody"> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Get.html">Get</a> and <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</a> instances can be optionally configured with <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/Filter.html">filters</a> which are applied on the RegionServer.</p> </div> <div class="paragraph"> <p>Filters can be confusing because there are many different types, and it is best to approach them by understanding the groups of Filter functionality.</p> </div> <div class="sect2"> <h3 id="client.filter.structural"><a class="anchor" href="#client.filter.structural"></a>65.1. Structural</h3> <div class="paragraph"> <p>Structural Filters contain other Filters.</p> </div> <div class="sect3"> <h4 id="client.filter.structural.fl"><a class="anchor" href="#client.filter.structural.fl"></a>65.1.1. FilterList</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FilterList.html">FilterList</a> represents a list of Filters with a relationship of <code>FilterList.Operator.MUST_PASS_ALL</code> or <code>FilterList.Operator.MUST_PASS_ONE</code> between the Filters. The following example shows an 'or' between two Filters (checking for either 'my value' or 'my other value' on the same attribute).</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">FilterList list = <span class="keyword">new</span> FilterList(FilterList.Operator.MUST_PASS_ONE); SingleColumnValueFilter filter1 = <span class="keyword">new</span> SingleColumnValueFilter( cf, column, CompareOp.EQUAL, Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">my value</span><span class="delimiter">"</span></span>) ); list.add(filter1); SingleColumnValueFilter filter2 = <span class="keyword">new</span> SingleColumnValueFilter( cf, column, CompareOp.EQUAL, Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">my other value</span><span class="delimiter">"</span></span>) ); list.add(filter2); scan.setFilter(list);</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="client.filter.cv"><a class="anchor" href="#client.filter.cv"></a>65.2. Column Value</h3> <div class="sect3"> <h4 id="client.filter.cv.scvf"><a class="anchor" href="#client.filter.cv.scvf"></a>65.2.1. SingleColumnValueFilter</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/SingleColumnValueFilter.html">SingleColumnValueFilter</a> can be used to test column values for equivalence (<code>CompareOp.EQUAL</code>), inequality (<code>CompareOp.NOT_EQUAL</code>), or ranges (e.g., <code>CompareOp.GREATER</code>). The following is example of testing equivalence a column to a String value "my value"…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">SingleColumnValueFilter filter = <span class="keyword">new</span> SingleColumnValueFilter( cf, column, CompareOp.EQUAL, Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">my value</span><span class="delimiter">"</span></span>) ); scan.setFilter(filter);</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="client.filter.cvp"><a class="anchor" href="#client.filter.cvp"></a>65.3. Column Value Comparators</h3> <div class="paragraph"> <p>There are several Comparator classes in the Filter package that deserve special mention. These Comparators are used in concert with other Filters, such as <a href="#client.filter.cv.scvf">SingleColumnValueFilter</a>.</p> </div> <div class="sect3"> <h4 id="client.filter.cvp.rcs"><a class="anchor" href="#client.filter.cvp.rcs"></a>65.3.1. RegexStringComparator</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/RegexStringComparator.html">RegexStringComparator</a> supports regular expressions for value comparisons.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">RegexStringComparator comp = <span class="keyword">new</span> RegexStringComparator(<span class="string"><span class="delimiter">"</span><span class="content">my.</span><span class="delimiter">"</span></span>); <span class="comment">// any value that starts with 'my'</span> SingleColumnValueFilter filter = <span class="keyword">new</span> SingleColumnValueFilter( cf, column, CompareOp.EQUAL, comp ); scan.setFilter(filter);</code></pre> </div> </div> <div class="paragraph"> <p>See the Oracle JavaDoc for <a href="http://download.oracle.com/javase/6/docs/api/java/util/regex/Pattern.html">supported RegEx patterns in Java</a>.</p> </div> </div> <div class="sect3"> <h4 id="client.filter.cvp.substringcomparator"><a class="anchor" href="#client.filter.cvp.substringcomparator"></a>65.3.2. SubstringComparator</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/SubstringComparator.html">SubstringComparator</a> can be used to determine if a given substring exists in a value. The comparison is case-insensitive.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">SubstringComparator comp = <span class="keyword">new</span> SubstringComparator(<span class="string"><span class="delimiter">"</span><span class="content">y val</span><span class="delimiter">"</span></span>); <span class="comment">// looking for 'my value'</span> SingleColumnValueFilter filter = <span class="keyword">new</span> SingleColumnValueFilter( cf, column, CompareOp.EQUAL, comp ); scan.setFilter(filter);</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="client.filter.cvp.bfp"><a class="anchor" href="#client.filter.cvp.bfp"></a>65.3.3. BinaryPrefixComparator</h4> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/BinaryPrefixComparator.html">BinaryPrefixComparator</a>.</p> </div> </div> <div class="sect3"> <h4 id="client.filter.cvp.bc"><a class="anchor" href="#client.filter.cvp.bc"></a>65.3.4. BinaryComparator</h4> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/BinaryComparator.html">BinaryComparator</a>.</p> </div> </div> </div> <div class="sect2"> <h3 id="client.filter.kvm"><a class="anchor" href="#client.filter.kvm"></a>65.4. KeyValue Metadata</h3> <div class="paragraph"> <p>As HBase stores data internally as KeyValue pairs, KeyValue Metadata Filters evaluate the existence of keys (i.e., ColumnFamily:Column qualifiers) for a row, as opposed to values the previous section.</p> </div> <div class="sect3"> <h4 id="client.filter.kvm.ff"><a class="anchor" href="#client.filter.kvm.ff"></a>65.4.1. FamilyFilter</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FamilyFilter.html">FamilyFilter</a> can be used to filter on the ColumnFamily. It is generally a better idea to select ColumnFamilies in the Scan than to do it with a Filter.</p> </div> </div> <div class="sect3"> <h4 id="client.filter.kvm.qf"><a class="anchor" href="#client.filter.kvm.qf"></a>65.4.2. QualifierFilter</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/QualifierFilter.html">QualifierFilter</a> can be used to filter based on Column (aka Qualifier) name.</p> </div> </div> <div class="sect3"> <h4 id="client.filter.kvm.cpf"><a class="anchor" href="#client.filter.kvm.cpf"></a>65.4.3. ColumnPrefixFilter</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/ColumnPrefixFilter.html">ColumnPrefixFilter</a> can be used to filter based on the lead portion of Column (aka Qualifier) names.</p> </div> <div class="paragraph"> <p>A ColumnPrefixFilter seeks ahead to the first column matching the prefix in each row and for each involved column family. It can be used to efficiently get a subset of the columns in very wide rows.</p> </div> <div class="paragraph"> <p>Note: The same column qualifier can be used in different column families. This filter returns all matching columns.</p> </div> <div class="paragraph"> <p>Example: Find all columns in a row and family that start with "abc"</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HTableInterface t = ...; <span class="type">byte</span><span class="type">[]</span> row = ...; <span class="type">byte</span><span class="type">[]</span> family = ...; <span class="type">byte</span><span class="type">[]</span> prefix = Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">abc</span><span class="delimiter">"</span></span>); Scan scan = <span class="keyword">new</span> Scan(row, row); <span class="comment">// (optional) limit to one row</span> scan.addFamily(family); <span class="comment">// (optional) limit to one family</span> <span class="predefined-type">Filter</span> f = <span class="keyword">new</span> ColumnPrefixFilter(prefix); scan.setFilter(f); scan.setBatch(<span class="integer">10</span>); <span class="comment">// set this if there could be many columns returned</span> ResultScanner rs = t.getScanner(scan); <span class="keyword">for</span> (<span class="predefined-type">Result</span> r = rs.next(); r != <span class="predefined-constant">null</span>; r = rs.next()) { <span class="keyword">for</span> (KeyValue kv : r.raw()) { <span class="comment">// each kv represents a column</span> } } rs.close();</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="client.filter.kvm.mcpf"><a class="anchor" href="#client.filter.kvm.mcpf"></a>65.4.4. MultipleColumnPrefixFilter</h4> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/MultipleColumnPrefixFilter.html">MultipleColumnPrefixFilter</a> behaves like ColumnPrefixFilter but allows specifying multiple prefixes.</p> </div> <div class="paragraph"> <p>Like ColumnPrefixFilter, MultipleColumnPrefixFilter efficiently seeks ahead to the first column matching the lowest prefix and also seeks past ranges of columns between prefixes. It can be used to efficiently get discontinuous sets of columns from very wide rows.</p> </div> <div class="paragraph"> <p>Example: Find all columns in a row and family that start with "abc" or "xyz"</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HTableInterface t = ...; <span class="type">byte</span><span class="type">[]</span> row = ...; <span class="type">byte</span><span class="type">[]</span> family = ...; <span class="type">byte</span><span class="type">[]</span><span class="type">[]</span> prefixes = <span class="keyword">new</span> <span class="type">byte</span><span class="type">[]</span><span class="type">[]</span> {Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">abc</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">xyz</span><span class="delimiter">"</span></span>)}; Scan scan = <span class="keyword">new</span> Scan(row, row); <span class="comment">// (optional) limit to one row</span> scan.addFamily(family); <span class="comment">// (optional) limit to one family</span> <span class="predefined-type">Filter</span> f = <span class="keyword">new</span> MultipleColumnPrefixFilter(prefixes); scan.setFilter(f); scan.setBatch(<span class="integer">10</span>); <span class="comment">// set this if there could be many columns returned</span> ResultScanner rs = t.getScanner(scan); <span class="keyword">for</span> (<span class="predefined-type">Result</span> r = rs.next(); r != <span class="predefined-constant">null</span>; r = rs.next()) { <span class="keyword">for</span> (KeyValue kv : r.raw()) { <span class="comment">// each kv represents a column</span> } } rs.close();</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="client.filter.kvm.crf"><a class="anchor" href="#client.filter.kvm.crf"></a>65.4.5. ColumnRangeFilter</h4> <div class="paragraph"> <p>A <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/ColumnRangeFilter.html">ColumnRangeFilter</a> allows efficient intra row scanning.</p> </div> <div class="paragraph"> <p>A ColumnRangeFilter can seek ahead to the first matching column for each involved column family. It can be used to efficiently get a 'slice' of the columns of a very wide row. i.e. you have a million columns in a row but you only want to look at columns bbbb-bbdd.</p> </div> <div class="paragraph"> <p>Note: The same column qualifier can be used in different column families. This filter returns all matching columns.</p> </div> <div class="paragraph"> <p>Example: Find all columns in a row and family between "bbbb" (inclusive) and "bbdd" (inclusive)</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HTableInterface t = ...; <span class="type">byte</span><span class="type">[]</span> row = ...; <span class="type">byte</span><span class="type">[]</span> family = ...; <span class="type">byte</span><span class="type">[]</span> startColumn = Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">bbbb</span><span class="delimiter">"</span></span>); <span class="type">byte</span><span class="type">[]</span> endColumn = Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">bbdd</span><span class="delimiter">"</span></span>); Scan scan = <span class="keyword">new</span> Scan(row, row); <span class="comment">// (optional) limit to one row</span> scan.addFamily(family); <span class="comment">// (optional) limit to one family</span> <span class="predefined-type">Filter</span> f = <span class="keyword">new</span> ColumnRangeFilter(startColumn, <span class="predefined-constant">true</span>, endColumn, <span class="predefined-constant">true</span>); scan.setFilter(f); scan.setBatch(<span class="integer">10</span>); <span class="comment">// set this if there could be many columns returned</span> ResultScanner rs = t.getScanner(scan); <span class="keyword">for</span> (<span class="predefined-type">Result</span> r = rs.next(); r != <span class="predefined-constant">null</span>; r = rs.next()) { <span class="keyword">for</span> (KeyValue kv : r.raw()) { <span class="comment">// each kv represents a column</span> } } rs.close();</code></pre> </div> </div> <div class="paragraph"> <p>Note: Introduced in HBase 0.92</p> </div> </div> </div> <div class="sect2"> <h3 id="client.filter.row"><a class="anchor" href="#client.filter.row"></a>65.5. RowKey</h3> <div class="sect3"> <h4 id="client.filter.row.rf"><a class="anchor" href="#client.filter.row.rf"></a>65.5.1. RowFilter</h4> <div class="paragraph"> <p>It is generally a better idea to use the startRow/stopRow methods on Scan for row selection, however <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/RowFilter.html">RowFilter</a> can also be used.</p> </div> </div> </div> <div class="sect2"> <h3 id="client.filter.utility"><a class="anchor" href="#client.filter.utility"></a>65.6. Utility</h3> <div class="sect3"> <h4 id="client.filter.utility.fkof"><a class="anchor" href="#client.filter.utility.fkof"></a>65.6.1. FirstKeyOnlyFilter</h4> <div class="paragraph"> <p>This is primarily used for rowcount jobs. See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FirstKeyOnlyFilter.html">FirstKeyOnlyFilter</a>.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="_master"><a class="anchor" href="#_master"></a>66. Master</h2> <div class="sectionbody"> <div class="paragraph"> <p><code>HMaster</code> is the implementation of the Master Server. The Master server is responsible for monitoring all RegionServer instances in the cluster, and is the interface for all metadata changes. In a distributed cluster, the Master typically runs on the <a href="#arch.hdfs.nn">NameNode</a>. J Mohamed Zahoor goes into some more detail on the Master Architecture in this blog posting, <a href="http://blog.zahoor.in/2012/08/hbase-hmaster-architecture/">HBase HMaster Architecture </a>.</p> </div> <div class="sect2"> <h3 id="master.startup"><a class="anchor" href="#master.startup"></a>66.1. Startup Behavior</h3> <div class="paragraph"> <p>If run in a multi-Master environment, all Masters compete to run the cluster. If the active Master loses its lease in ZooKeeper (or the Master shuts down), then the remaining Masters jostle to take over the Master role.</p> </div> </div> <div class="sect2"> <h3 id="master.runtime"><a class="anchor" href="#master.runtime"></a>66.2. Runtime Impact</h3> <div class="paragraph"> <p>A common dist-list question involves what happens to an HBase cluster when the Master goes down. Because the HBase client talks directly to the RegionServers, the cluster can still function in a "steady state". Additionally, per <a href="#arch.catalog">Catalog Tables</a>, <code>hbase:meta</code> exists as an HBase table and is not resident in the Master. However, the Master controls critical functions such as RegionServer failover and completing region splits. So while the cluster can still run for a short time without the Master, the Master should be restarted as soon as possible.</p> </div> </div> <div class="sect2"> <h3 id="master.api"><a class="anchor" href="#master.api"></a>66.3. Interface</h3> <div class="paragraph"> <p>The methods exposed by <code>HMasterInterface</code> are primarily metadata-oriented methods:</p> </div> <div class="ulist"> <ul> <li> <p>Table (createTable, modifyTable, removeTable, enable, disable)</p> </li> <li> <p>ColumnFamily (addColumn, modifyColumn, removeColumn)</p> </li> <li> <p>Region (move, assign, unassign) For example, when the <code>Admin</code> method <code>disableTable</code> is invoked, it is serviced by the Master server.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="master.processes"><a class="anchor" href="#master.processes"></a>66.4. Processes</h3> <div class="paragraph"> <p>The Master runs several background threads:</p> </div> <div class="sect3"> <h4 id="master.processes.loadbalancer"><a class="anchor" href="#master.processes.loadbalancer"></a>66.4.1. LoadBalancer</h4> <div class="paragraph"> <p>Periodically, and when there are no regions in transition, a load balancer will run and move regions around to balance the cluster’s load. See <a href="#balancer_config">Balancer</a> for configuring this property.</p> </div> <div class="paragraph"> <p>See <a href="#regions.arch.assignment">Region-RegionServer Assignment</a> for more information on region assignment.</p> </div> </div> <div class="sect3"> <h4 id="master.processes.catalog"><a class="anchor" href="#master.processes.catalog"></a>66.4.2. CatalogJanitor</h4> <div class="paragraph"> <p>Periodically checks and cleans up the <code>hbase:meta</code> table. See <arch.catalog.meta>> for more information on the meta table.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="regionserver.arch"><a class="anchor" href="#regionserver.arch"></a>67. RegionServer</h2> <div class="sectionbody"> <div class="paragraph"> <p><code>HRegionServer</code> is the RegionServer implementation. It is responsible for serving and managing regions. In a distributed cluster, a RegionServer runs on a <a href="#arch.hdfs.dn">DataNode</a>.</p> </div> <div class="sect2"> <h3 id="regionserver.arch.api"><a class="anchor" href="#regionserver.arch.api"></a>67.1. Interface</h3> <div class="paragraph"> <p>The methods exposed by <code>HRegionRegionInterface</code> contain both data-oriented and region-maintenance methods:</p> </div> <div class="ulist"> <ul> <li> <p>Data (get, put, delete, next, etc.)</p> </li> <li> <p>Region (splitRegion, compactRegion, etc.) For example, when the <code>Admin</code> method <code>majorCompact</code> is invoked on a table, the client is actually iterating through all regions for the specified table and requesting a major compaction directly to each region.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="regionserver.arch.processes"><a class="anchor" href="#regionserver.arch.processes"></a>67.2. Processes</h3> <div class="paragraph"> <p>The RegionServer runs a variety of background threads:</p> </div> <div class="sect3"> <h4 id="regionserver.arch.processes.compactsplit"><a class="anchor" href="#regionserver.arch.processes.compactsplit"></a>67.2.1. CompactSplitThread</h4> <div class="paragraph"> <p>Checks for splits and handle minor compactions.</p> </div> </div> <div class="sect3"> <h4 id="regionserver.arch.processes.majorcompact"><a class="anchor" href="#regionserver.arch.processes.majorcompact"></a>67.2.2. MajorCompactionChecker</h4> <div class="paragraph"> <p>Checks for major compactions.</p> </div> </div> <div class="sect3"> <h4 id="regionserver.arch.processes.memstore"><a class="anchor" href="#regionserver.arch.processes.memstore"></a>67.2.3. MemStoreFlusher</h4> <div class="paragraph"> <p>Periodically flushes in-memory writes in the MemStore to StoreFiles.</p> </div> </div> <div class="sect3"> <h4 id="regionserver.arch.processes.log"><a class="anchor" href="#regionserver.arch.processes.log"></a>67.2.4. LogRoller</h4> <div class="paragraph"> <p>Periodically checks the RegionServer’s WAL.</p> </div> </div> </div> <div class="sect2"> <h3 id="_coprocessors"><a class="anchor" href="#_coprocessors"></a>67.3. Coprocessors</h3> <div class="paragraph"> <p>Coprocessors were added in 0.92. There is a thorough <a href="https://blogs.apache.org/hbase/entry/coprocessor_introduction">Blog Overview of CoProcessors</a> posted. Documentation will eventually move to this reference guide, but the blog is the most current information available at this time.</p> </div> </div> <div class="sect2"> <h3 id="block.cache"><a class="anchor" href="#block.cache"></a>67.4. Block Cache</h3> <div class="paragraph"> <p>HBase provides two different BlockCache implementations: the default on-heap <code>LruBlockCache</code> and the <code>BucketCache</code>, which is (usually) off-heap. This section discusses benefits and drawbacks of each implementation, how to choose the appropriate option, and configuration options for each.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Block Cache Reporting: UI</div> <div class="paragraph"> <p>See the RegionServer UI for detail on caching deploy. Since HBase 0.98.4, the Block Cache detail has been significantly extended showing configurations, sizings, current usage, time-in-the-cache, and even detail on block counts and types.</p> </div> </td> </tr> </table> </div> <div class="sect3"> <h4 id="_cache_choices"><a class="anchor" href="#_cache_choices"></a>67.4.1. Cache Choices</h4> <div class="paragraph"> <p><code>LruBlockCache</code> is the original implementation, and is entirely within the Java heap. <code>BucketCache</code> is mainly intended for keeping block cache data off-heap, although <code>BucketCache</code> can also keep data on-heap and serve from a file-backed cache.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">BucketCache is production ready as of HBase 0.98.6</div> <div class="paragraph"> <p>To run with BucketCache, you need HBASE-11678. This was included in 0.98.6.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>Fetching will always be slower when fetching from BucketCache, as compared to the native on-heap LruBlockCache. However, latencies tend to be less erratic across time, because there is less garbage collection when you use BucketCache since it is managing BlockCache allocations, not the GC. If the BucketCache is deployed in off-heap mode, this memory is not managed by the GC at all. This is why you’d use BucketCache, so your latencies are less erratic and to mitigate GCs and heap fragmentation. See Nick Dimiduk’s <a href="http://www.n10k.com/blog/blockcache-101/">BlockCache 101</a> for comparisons running on-heap vs off-heap tests. Also see <a href="http://people.apache.org/~stack/bc/">Comparing BlockCache Deploys</a> which finds that if your dataset fits inside your LruBlockCache deploy, use it otherwise if you are experiencing cache churn (or you want your cache to exist beyond the vagaries of java GC), use BucketCache.</p> </div> <div class="paragraph"> <p>When you enable BucketCache, you are enabling a two tier caching system, an L1 cache which is implemented by an instance of LruBlockCache and an off-heap L2 cache which is implemented by BucketCache. Management of these two tiers and the policy that dictates how blocks move between them is done by <code>CombinedBlockCache</code>. It keeps all DATA blocks in the L2 BucketCache and meta blocks — INDEX and BLOOM blocks — on-heap in the L1 <code>LruBlockCache</code>. See <a href="#offheap.blockcache">Off-heap Block Cache</a> for more detail on going off-heap.</p> </div> </div> <div class="sect3"> <h4 id="cache.configurations"><a class="anchor" href="#cache.configurations"></a>67.4.2. General Cache Configurations</h4> <div class="paragraph"> <p>Apart from the cache implementation itself, you can set some general configuration options to control how the cache performs. See <a href="http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/io/hfile/CacheConfig.html" class="bare">http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/io/hfile/CacheConfig.html</a>. After setting any of these options, restart or rolling restart your cluster for the configuration to take effect. Check logs for errors or unexpected behavior.</p> </div> <div class="paragraph"> <p>See also <a href="#blockcache.prefetch">Prefetch Option for Blockcache</a>, which discusses a new option introduced in <a href="https://issues.apache.org/jira/browse/HBASE-9857">HBASE-9857</a>.</p> </div> </div> <div class="sect3"> <h4 id="block.cache.design"><a class="anchor" href="#block.cache.design"></a>67.4.3. LruBlockCache Design</h4> <div class="paragraph"> <p>The LruBlockCache is an LRU cache that contains three levels of block priority to allow for scan-resistance and in-memory ColumnFamilies:</p> </div> <div class="ulist"> <ul> <li> <p>Single access priority: The first time a block is loaded from HDFS it normally has this priority and it will be part of the first group to be considered during evictions. The advantage is that scanned blocks are more likely to get evicted than blocks that are getting more usage.</p> </li> <li> <p>Multi access priority: If a block in the previous priority group is accessed again, it upgrades to this priority. It is thus part of the second group considered during evictions.</p> </li> <li> <p>In-memory access priority: If the block’s family was configured to be "in-memory", it will be part of this priority disregarding the number of times it was accessed. Catalog tables are configured like this. This group is the last one considered during evictions.</p> <div class="paragraph"> <p>To mark a column family as in-memory, call</p> </div> </li> </ul> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HColumnDescriptor.setInMemory(<span class="predefined-constant">true</span>);</code></pre> </div> </div> <div class="paragraph"> <p>if creating a table from java, or set <code>IN_MEMORY ⇒ true</code> when creating or altering a table in the shell: e.g.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">003</span>:<span class="integer">0</span>> create <span class="string"><span class="delimiter">'</span><span class="content">t</span><span class="delimiter">'</span></span>, {NAME => <span class="string"><span class="delimiter">'</span><span class="content">f</span><span class="delimiter">'</span></span>, IN_MEMORY => <span class="string"><span class="delimiter">'</span><span class="content">true</span><span class="delimiter">'</span></span>}</code></pre> </div> </div> <div class="paragraph"> <p>For more information, see the <a href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/io/hfile/LruBlockCache.html">LruBlockCache source</a></p> </div> </div> <div class="sect3"> <h4 id="block.cache.usage"><a class="anchor" href="#block.cache.usage"></a>67.4.4. LruBlockCache Usage</h4> <div class="paragraph"> <p>Block caching is enabled by default for all the user tables which means that any read operation will load the LRU cache. This might be good for a large number of use cases, but further tunings are usually required in order to achieve better performance. An important concept is the <a href="http://en.wikipedia.org/wiki/Working_set_size">working set size</a>, or WSS, which is: "the amount of memory needed to compute the answer to a problem". For a website, this would be the data that’s needed to answer the queries over a short amount of time.</p> </div> <div class="paragraph"> <p>The way to calculate how much memory is available in HBase for caching is:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">number of region servers * heap size * hfile.block.cache.size * <span class="float">0.99</span></code></pre> </div> </div> <div class="paragraph"> <p>The default value for the block cache is 0.25 which represents 25% of the available heap. The last value (99%) is the default acceptable loading factor in the LRU cache after which eviction is started. The reason it is included in this equation is that it would be unrealistic to say that it is possible to use 100% of the available memory since this would make the process blocking from the point where it loads new blocks. Here are some examples:</p> </div> <div class="ulist"> <ul> <li> <p>One region server with the heap size set to 1 GB and the default block cache size will have 253 MB of block cache available.</p> </li> <li> <p>20 region servers with the heap size set to 8 GB and a default block cache size will have 39.6 of block cache.</p> </li> <li> <p>100 region servers with the heap size set to 24 GB and a block cache size of 0.5 will have about 1.16 TB of block cache.</p> </li> </ul> </div> <div class="paragraph"> <p>Your data is not the only resident of the block cache. Here are others that you may have to take into account:</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Catalog Tables</dt> <dd> <p>The <code>-ROOT-</code> (prior to HBase 0.96, see <a href="#arch.catalog.root">arch.catalog.root</a>) and <code>hbase:meta</code> tables are forced into the block cache and have the in-memory priority which means that they are harder to evict. The former never uses more than a few hundreds bytes while the latter can occupy a few MBs (depending on the number of regions).</p> </dd> <dt class="hdlist1">HFiles Indexes</dt> <dd> <p>An <em>HFile</em> is the file format that HBase uses to store data in HDFS. It contains a multi-layered index which allows HBase to seek to the data without having to read the whole file. The size of those indexes is a factor of the block size (64KB by default), the size of your keys and the amount of data you are storing. For big data sets it’s not unusual to see numbers around 1GB per region server, although not all of it will be in cache because the LRU will evict indexes that aren’t used.</p> </dd> <dt class="hdlist1">Keys</dt> <dd> <p>The values that are stored are only half the picture, since each value is stored along with its keys (row key, family qualifier, and timestamp). See <a href="#keysize">Try to minimize row and column sizes</a>.</p> </dd> <dt class="hdlist1">Bloom Filters</dt> <dd> <p>Just like the HFile indexes, those data structures (when enabled) are stored in the LRU.</p> </dd> </dl> </div> <div class="paragraph"> <p>Currently the recommended way to measure HFile indexes and bloom filters sizes is to look at the region server web UI and checkout the relevant metrics. For keys, sampling can be done by using the HFile command line tool and look for the average key size metric. Since HBase 0.98.3, you can view details on BlockCache stats and metrics in a special Block Cache section in the UI.</p> </div> <div class="paragraph"> <p>It’s generally bad to use block caching when the WSS doesn’t fit in memory. This is the case when you have for example 40GB available across all your region servers' block caches but you need to process 1TB of data. One of the reasons is that the churn generated by the evictions will trigger more garbage collections unnecessarily. Here are two use cases:</p> </div> <div class="ulist"> <ul> <li> <p>Fully random reading pattern: This is a case where you almost never access the same row twice within a short amount of time such that the chance of hitting a cached block is close to 0. Setting block caching on such a table is a waste of memory and CPU cycles, more so that it will generate more garbage to pick up by the JVM. For more information on monitoring GC, see <a href="#trouble.log.gc">JVM Garbage Collection Logs</a>.</p> </li> <li> <p>Mapping a table: In a typical MapReduce job that takes a table in input, every row will be read only once so there’s no need to put them into the block cache. The Scan object has the option of turning this off via the setCaching method (set it to false). You can still keep block caching turned on on this table if you need fast random read access. An example would be counting the number of rows in a table that serves live traffic, caching every block of that table would create massive churn and would surely evict data that’s currently in use.</p> </li> </ul> </div> <div class="sect4"> <h5 id="data.blocks.in.fscache"><a class="anchor" href="#data.blocks.in.fscache"></a>Caching META blocks only (DATA blocks in fscache)</h5> <div class="paragraph"> <p>An interesting setup is one where we cache META blocks only and we read DATA blocks in on each access. If the DATA blocks fit inside fscache, this alternative may make sense when access is completely random across a very large dataset. To enable this setup, alter your table and for each column family set <code>BLOCKCACHE ⇒ 'false'</code>. You are 'disabling' the BlockCache for this column family only. You can never disable the caching of META blocks. Since <a href="https://issues.apache.org/jira/browse/HBASE-4683">HBASE-4683 Always cache index and bloom blocks</a>, we will cache META blocks even if the BlockCache is disabled.</p> </div> </div> </div> <div class="sect3"> <h4 id="offheap.blockcache"><a class="anchor" href="#offheap.blockcache"></a>67.4.5. Off-heap Block Cache</h4> <div class="sect4"> <h5 id="enable.bucketcache"><a class="anchor" href="#enable.bucketcache"></a>How to Enable BucketCache</h5> <div class="paragraph"> <p>The usual deploy of BucketCache is via a managing class that sets up two caching tiers: an L1 on-heap cache implemented by LruBlockCache and a second L2 cache implemented with BucketCache. The managing class is <a href="http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/io/hfile/CombinedBlockCache.html">CombinedBlockCache</a> by default. The previous link describes the caching 'policy' implemented by CombinedBlockCache. In short, it works by keeping meta blocks — INDEX and BLOOM in the L1, on-heap LruBlockCache tier — and DATA blocks are kept in the L2, BucketCache tier. It is possible to amend this behavior in HBase since version 1.0 and ask that a column family have both its meta and DATA blocks hosted on-heap in the L1 tier by setting <code>cacheDataInL1</code> via <code>(HColumnDescriptor.setCacheDataInL1(true)</code> or in the shell, creating or amending column families setting <code>CACHE_DATA_IN_L1</code> to true: e.g.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">003</span>:<span class="integer">0</span>> create <span class="string"><span class="delimiter">'</span><span class="content">t</span><span class="delimiter">'</span></span>, {NAME => <span class="string"><span class="delimiter">'</span><span class="content">t</span><span class="delimiter">'</span></span>, CONFIGURATION => {CACHE_DATA_IN_L1 => <span class="string"><span class="delimiter">'</span><span class="content">true</span><span class="delimiter">'</span></span>}}</code></pre> </div> </div> <div class="paragraph"> <p>The BucketCache Block Cache can be deployed on-heap, off-heap, or file based. You set which via the <code>hbase.bucketcache.ioengine</code> setting. Setting it to <code>heap</code> will have BucketCache deployed inside the allocated Java heap. Setting it to <code>offheap</code> will have BucketCache make its allocations off-heap, and an ioengine setting of <code>file:PATH_TO_FILE</code> will direct BucketCache to use a file caching (Useful in particular if you have some fast I/O attached to the box such as SSDs).</p> </div> <div class="paragraph"> <p>It is possible to deploy an L1+L2 setup where we bypass the CombinedBlockCache policy and have BucketCache working as a strict L2 cache to the L1 LruBlockCache. For such a setup, set <code>CacheConfig.BUCKET_CACHE_COMBINED_KEY</code> to <code>false</code>. In this mode, on eviction from L1, blocks go to L2. When a block is cached, it is cached first in L1. When we go to look for a cached block, we look first in L1 and if none found, then search L2. Let us call this deploy format, <em>Raw L1+L2</em>.</p> </div> <div class="paragraph"> <p>Other BucketCache configs include: specifying a location to persist cache to across restarts, how many threads to use writing the cache, etc. See the <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/io/hfile/CacheConfig.html">CacheConfig.html</a> class for configuration options and descriptions.</p> </div> <div class="sect5"> <h6 id="_bucketcache_example_configuration"><a class="anchor" href="#_bucketcache_example_configuration"></a>BucketCache Example Configuration</h6> <div class="paragraph"> <p>This sample provides a configuration for a 4 GB off-heap BucketCache with a 1 GB on-heap cache.</p> </div> <div class="paragraph"> <p>Configuration is performed on the RegionServer.</p> </div> <div class="paragraph"> <p>Setting <code>hbase.bucketcache.ioengine</code> and <code>hbase.bucketcache.size</code> > 0 enables <code>CombinedBlockCache</code>. Let us presume that the RegionServer has been set to run with a 5G heap: i.e. <code>HBASE_HEAPSIZE=5g</code>.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>First, edit the RegionServer’s <em>hbase-env.sh</em> and set <code>HBASE_OFFHEAPSIZE</code> to a value greater than the off-heap size wanted, in this case, 4 GB (expressed as 4G). Let’s set it to 5G. That’ll be 4G for our off-heap cache and 1G for any other uses of off-heap memory (there are other users of off-heap memory other than BlockCache; e.g. DFSClient in RegionServer can make use of off-heap memory). See <a href="#direct.memory">Direct Memory Usage In HBase</a>.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HBASE_OFFHEAPSIZE=<span class="integer">5</span>G</code></pre> </div> </div> </li> <li> <p>Next, add the following configuration to the RegionServer’s <em>hbase-site.xml</em>.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.bucketcache.ioengine<span class="tag"></name></span> <span class="tag"><value></span>offheap<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hfile.block.cache.size<span class="tag"></name></span> <span class="tag"><value></span>0.2<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.bucketcache.size<span class="tag"></name></span> <span class="tag"><value></span>4196<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </li> <li> <p>Restart or rolling restart your cluster, and check the logs for any issues.</p> </li> </ol> </div> <div class="paragraph"> <p>In the above, we set the BucketCache to be 4G. We configured the on-heap LruBlockCache have 20% (0.2) of the RegionServer’s heap size (0.2 * 5G = 1G). In other words, you configure the L1 LruBlockCache as you would normally (as if there were no L2 cache present).</p> </div> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-10641">HBASE-10641</a> introduced the ability to configure multiple sizes for the buckets of the BucketCache, in HBase 0.98 and newer. To configurable multiple bucket sizes, configure the new property <code>hfile.block.cache.sizes</code> (instead of <code>hfile.block.cache.size</code>) to a comma-separated list of block sizes, ordered from smallest to largest, with no spaces. The goal is to optimize the bucket sizes based on your data access patterns. The following example configures buckets of size 4096 and 8192.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hfile.block.cache.sizes<span class="tag"></name></span> <span class="tag"><value></span>4096,8192<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div id="direct.memory" class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Direct Memory Usage In HBase</div> <div class="paragraph"> <p>The default maximum direct memory varies by JVM. Traditionally it is 64M or some relation to allocated heap size (-Xmx) or no limit at all (JDK7 apparently). HBase servers use direct memory, in particular short-circuit reading, the hosted DFSClient will allocate direct memory buffers. If you do off-heap block caching, you’ll be making use of direct memory. Starting your JVM, make sure the <code>-XX:MaxDirectMemorySize</code> setting in <em>conf/hbase-env.sh</em> is set to some value that is higher than what you have allocated to your off-heap BlockCache (<code>hbase.bucketcache.size</code>). It should be larger than your off-heap block cache and then some for DFSClient usage (How much the DFSClient uses is not easy to quantify; it is the number of open HFiles * <code>hbase.dfs.client.read.shortcircuit.buffer.size</code> where <code>hbase.dfs.client.read.shortcircuit.buffer.size</code> is set to 128k in HBase — see <em>hbase-default.xml</em> default configurations). Direct memory, which is part of the Java process heap, is separate from the object heap allocated by -Xmx. The value allocated by <code>MaxDirectMemorySize</code> must not exceed physical RAM, and is likely to be less than the total available RAM due to other memory requirements and system constraints.</p> </div> <div class="paragraph"> <p>You can see how much memory — on-heap and off-heap/direct — a RegionServer is configured to use and how much it is using at any one time by looking at the <em>Server Metrics: Memory</em> tab in the UI. It can also be gotten via JMX. In particular the direct memory currently used by the server can be found on the <code>java.nio.type=BufferPool,name=direct</code> bean. Terracotta has a <a href="http://terracotta.org/documentation/4.0/bigmemorygo/configuration/storage-options">good write up</a> on using off-heap memory in Java. It is for their product BigMemory but a lot of the issues noted apply in general to any attempt at going off-heap. Check it out.</p> </div> </td> </tr> </table> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">hbase.bucketcache.percentage.in.combinedcache</div> <div class="paragraph"> <p>This is a pre-HBase 1.0 configuration removed because it was confusing. It was a float that you would set to some value between 0.0 and 1.0. Its default was 0.9. If the deploy was using CombinedBlockCache, then the LruBlockCache L1 size was calculated to be <code>(1 - hbase.bucketcache.percentage.in.combinedcache) * size-of-bucketcache</code> and the BucketCache size was <code>hbase.bucketcache.percentage.in.combinedcache * size-of-bucket-cache</code>. where size-of-bucket-cache itself is EITHER the value of the configuration <code>hbase.bucketcache.size</code> IF it was specified as Megabytes OR <code>hbase.bucketcache.size</code> * <code>-XX:MaxDirectMemorySize</code> if <code>hbase.bucketcache.size</code> is between 0 and 1.0.</p> </div> <div class="paragraph"> <p>In 1.0, it should be more straight-forward. L1 LruBlockCache size is set as a fraction of java heap using <code>hfile.block.cache.size setting</code> (not the best name) and L2 is set as above either in absolute Megabytes or as a fraction of allocated maximum direct memory.</p> </div> </td> </tr> </table> </div> </div> </div> </div> <div class="sect3"> <h4 id="_compressed_blockcache"><a class="anchor" href="#_compressed_blockcache"></a>67.4.6. Compressed BlockCache</h4> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-11331">HBASE-11331</a> introduced lazy BlockCache decompression, more simply referred to as compressed BlockCache. When compressed BlockCache is enabled data and encoded data blocks are cached in the BlockCache in their on-disk format, rather than being decompressed and decrypted before caching.</p> </div> <div class="paragraph"> <p>For a RegionServer hosting more data than can fit into cache, enabling this feature with SNAPPY compression has been shown to result in 50% increase in throughput and 30% improvement in mean latency while, increasing garbage collection by 80% and increasing overall CPU load by 2%. See HBASE-11331 for more details about how performance was measured and achieved. For a RegionServer hosting data that can comfortably fit into cache, or if your workload is sensitive to extra CPU or garbage-collection load, you may receive less benefit.</p> </div> <div class="paragraph"> <p>The compressed BlockCache is disabled by default. To enable it, set <code>hbase.block.data.cachecompressed</code> to <code>true</code> in <em>hbase-site.xml</em> on all RegionServers.</p> </div> </div> </div> <div class="sect2"> <h3 id="regionserver_splitting_implementation"><a class="anchor" href="#regionserver_splitting_implementation"></a>67.5. RegionServer Splitting Implementation</h3> <div class="paragraph"> <p>As write requests are handled by the region server, they accumulate in an in-memory storage system called the <em>memstore</em>. Once the memstore fills, its content are written to disk as additional store files. This event is called a <em>memstore flush</em>. As store files accumulate, the RegionServer will <a href="#compaction">compact</a> them into fewer, larger files. After each flush or compaction finishes, the amount of data stored in the region has changed. The RegionServer consults the region split policy to determine if the region has grown too large or should be split for another policy-specific reason. A region split request is enqueued if the policy recommends it.</p> </div> <div class="paragraph"> <p>Logically, the process of splitting a region is simple. We find a suitable point in the keyspace of the region where we should divide the region in half, then split the region’s data into two new regions at that point. The details of the process however are not simple. When a split happens, the newly created <em>daughter regions</em> do not rewrite all the data into new files immediately. Instead, they create small files similar to symbolic link files, named <a href="http://www.google.com/url?q=http%3A%2F%2Fhbase.apache.org%2Fapidocs%2Forg%2Fapache%2Fhadoop%2Fhbase%2Fio%2FReference.html&sa=D&sntz=1&usg=AFQjCNEkCbADZ3CgKHTtGYI8bJVwp663CA">Reference files</a>, which point to either the top or bottom part of the parent store file according to the split point. The reference file is used just like a regular data file, but only half of the records are considered. The region can only be split if there are no more references to the immutable data files of the parent region. Those reference files are cleaned gradually by compactions, so that the region will stop referring to its parents files, and can be split further.</p> </div> <div class="paragraph"> <p>Although splitting the region is a local decision made by the RegionServer, the split process itself must coordinate with many actors. The RegionServer notifies the Master before and after the split, updates the <code>.META.</code> table so that clients can discover the new daughter regions, and rearranges the directory structure and data files in HDFS. Splitting is a multi-task process. To enable rollback in case of an error, the RegionServer keeps an in-memory journal about the execution state. The steps taken by the RegionServer to execute the split are illustrated in <a href="#regionserver_split_process_image">RegionServer Split Process</a>. Each step is labeled with its step number. Actions from RegionServers or Master are shown in red, while actions from the clients are show in green.</p> </div> <div id="regionserver_split_process_image" class="imageblock"> <div class="content"> <img src="images/region_split_process.png" alt="Region Split Process"> </div> <div class="title">Figure 1. RegionServer Split Process</div> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>The RegionServer decides locally to split the region, and prepares the split. <strong>THE SPLIT TRANSACTION IS STARTED.</strong> As a first step, the RegionServer acquires a shared read lock on the table to prevent schema modifications during the splitting process. Then it creates a znode in zookeeper under <code>/hbase/region-in-transition/region-name</code>, and sets the znode’s state to <code>SPLITTING</code>.</p> </li> <li> <p>The Master learns about this znode, since it has a watcher for the parent <code>region-in-transition</code> znode.</p> </li> <li> <p>The RegionServer creates a sub-directory named <code>.splits</code> under the parent’s <code>region</code> directory in HDFS.</p> </li> <li> <p>The RegionServer closes the parent region and marks the region as offline in its local data structures. <strong>THE SPLITTING REGION IS NOW OFFLINE.</strong> At this point, client requests coming to the parent region will throw <code>NotServingRegionException</code>. The client will retry with some backoff. The closing region is flushed.</p> </li> <li> <p>The RegionServer creates region directories under the <code>.splits</code> directory, for daughter regions A and B, and creates necessary data structures. Then it splits the store files, in the sense that it creates two <a href="http://www.google.com/url?q=http%3A%2F%2Fhbase.apache.org%2Fapidocs%2Forg%2Fapache%2Fhadoop%2Fhbase%2Fio%2FReference.html&sa=D&sntz=1&usg=AFQjCNEkCbADZ3CgKHTtGYI8bJVwp663CA">Reference</a> files per store file in the parent region. Those reference files will point to the parent regions’files.</p> </li> <li> <p>The RegionServer creates the actual region directory in HDFS, and moves the reference files for each daughter.</p> </li> <li> <p>The RegionServer sends a <code>Put</code> request to the <code>.META.</code> table, to set the parent as offline in the <code>.META.</code> table and add information about daughter regions. At this point, there won’t be individual entries in <code>.META.</code> for the daughters. Clients will see that the parent region is split if they scan <code>.META.</code>, but won’t know about the daughters until they appear in <code>.META.</code>. Also, if this <code>Put</code> to <code>.META</code>. succeeds, the parent will be effectively split. If the RegionServer fails before this RPC succeeds, Master and the next Region Server opening the region will clean dirty state about the region split. After the <code>.META.</code> update, though, the region split will be rolled-forward by Master.</p> </li> <li> <p>The RegionServer opens daughters A and B in parallel.</p> </li> <li> <p>The RegionServer adds the daughters A and B to <code>.META.</code>, together with information that it hosts the regions. <strong>THE SPLIT REGIONS (DAUGHTERS WITH REFERENCES TO PARENT) ARE NOW ONLINE.</strong> After this point, clients can discover the new regions and issue requests to them. Clients cache the <code>.META.</code> entries locally, but when they make requests to the RegionServer or <code>.META.</code>, their caches will be invalidated, and they will learn about the new regions from <code>.META.</code>.</p> </li> <li> <p>The RegionServer updates znode <code>/hbase/region-in-transition/region-name</code> in ZooKeeper to state <code>SPLIT</code>, so that the master can learn about it. The balancer can freely re-assign the daughter regions to other region servers if necessary. <strong>THE SPLIT TRANSACTION IS NOW FINISHED.</strong></p> </li> <li> <p>After the split, <code>.META.</code> and HDFS will still contain references to the parent region. Those references will be removed when compactions in daughter regions rewrite the data files. Garbage collection tasks in the master periodically check whether the daughter regions still refer to the parent region’s files. If not, the parent region will be removed.</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="wal"><a class="anchor" href="#wal"></a>67.6. Write Ahead Log (WAL)</h3> <div class="sect3"> <h4 id="purpose.wal"><a class="anchor" href="#purpose.wal"></a>67.6.1. Purpose</h4> <div class="paragraph"> <p>The <em>Write Ahead Log (WAL)</em> records all changes to data in HBase, to file-based storage. Under normal operations, the WAL is not needed because data changes move from the MemStore to StoreFiles. However, if a RegionServer crashes or becomes unavailable before the MemStore is flushed, the WAL ensures that the changes to the data can be replayed. If writing to the WAL fails, the entire operation to modify the data fails.</p> </div> <div class="paragraph"> <p>HBase uses an implementation of the <a href="http://hbase.apache.org/devapidocs/org/apache/hadoop/hbase/wal/WAL.html">WAL</a> interface. Usually, there is only one instance of a WAL per RegionServer. The RegionServer records Puts and Deletes to it, before recording them to the <a href="#store.memstore">MemStore</a> for the affected <a href="#store">[store]</a>.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">The HLog</div> <div class="paragraph"> <p>Prior to 2.0, the interface for WALs in HBase was named <code>HLog</code>. In 0.94, HLog was the name of the implementation of the WAL. You will likely find references to the HLog in documentation tailored to these older versions.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>The WAL resides in HDFS in the <em>/hbase/WALs/</em> directory (prior to HBase 0.94, they were stored in <em>/hbase/.logs/</em>), with subdirectories per region.</p> </div> <div class="paragraph"> <p>For more general information about the concept of write ahead logs, see the Wikipedia <a href="http://en.wikipedia.org/wiki/Write-ahead_logging">Write-Ahead Log</a> article.</p> </div> </div> <div class="sect3"> <h4 id="_multiwal"><a class="anchor" href="#_multiwal"></a>67.6.2. MultiWAL</h4> <div class="paragraph"> <p>With a single WAL per RegionServer, the RegionServer must write to the WAL serially, because HDFS files must be sequential. This causes the WAL to be a performance bottleneck.</p> </div> <div class="paragraph"> <p>HBase 1.0 introduces support MultiWal in <a href="https://issues.apache.org/jira/browse/HBASE-5699">HBASE-5699</a>. MultiWAL allows a RegionServer to write multiple WAL streams in parallel, by using multiple pipelines in the underlying HDFS instance, which increases total throughput during writes. This parallelization is done by partitioning incoming edits by their Region. Thus, the current implementation will not help with increasing the throughput to a single Region.</p> </div> <div class="paragraph"> <p>RegionServers using the original WAL implementation and those using the MultiWAL implementation can each handle recovery of either set of WALs, so a zero-downtime configuration update is possible through a rolling restart.</p> </div> <div class="paragraph"> <div class="title">Configure MultiWAL</div> <p>To configure MultiWAL for a RegionServer, set the value of the property <code>hbase.wal.provider</code> to <code>multiwal</code> by pasting in the following XML:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.wal.provider<span class="tag"></name></span> <span class="tag"><value></span>multiwal<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Restart the RegionServer for the changes to take effect.</p> </div> <div class="paragraph"> <p>To disable MultiWAL for a RegionServer, unset the property and restart the RegionServer.</p> </div> </div> <div class="sect3"> <h4 id="wal_flush"><a class="anchor" href="#wal_flush"></a>67.6.3. WAL Flushing</h4> <div class="paragraph"> <p>TODO (describe).</p> </div> </div> <div class="sect3"> <h4 id="_wal_splitting"><a class="anchor" href="#_wal_splitting"></a>67.6.4. WAL Splitting</h4> <div class="paragraph"> <p>A RegionServer serves many regions. All of the regions in a region server share the same active WAL file. Each edit in the WAL file includes information about which region it belongs to. When a region is opened, the edits in the WAL file which belong to that region need to be replayed. Therefore, edits in the WAL file must be grouped by region so that particular sets can be replayed to regenerate the data in a particular region. The process of grouping the WAL edits by region is called <em>log splitting</em>. It is a critical process for recovering data if a region server fails.</p> </div> <div class="paragraph"> <p>Log splitting is done by the HMaster during cluster start-up or by the ServerShutdownHandler as a region server shuts down. So that consistency is guaranteed, affected regions are unavailable until data is restored. All WAL edits need to be recovered and replayed before a given region can become available again. As a result, regions affected by log splitting are unavailable until the process completes.</p> </div> <div class="olist arabic"> <div class="title">Procedure: Log Splitting, Step by Step</div> <ol class="arabic"> <li> <p>The <em>/hbase/WALs/<host>,<port>,<startcode></em> directory is renamed.</p> <div class="paragraph"> <p>Renaming the directory is important because a RegionServer may still be up and accepting requests even if the HMaster thinks it is down. If the RegionServer does not respond immediately and does not heartbeat its ZooKeeper session, the HMaster may interpret this as a RegionServer failure. Renaming the logs directory ensures that existing, valid WAL files which are still in use by an active but busy RegionServer are not written to by accident.</p> </div> <div class="paragraph"> <p>The new directory is named according to the following pattern:</p> </div> <div class="listingblock"> <div class="content"> <pre>/hbase/WALs/<host>,<port>,<startcode>-splitting</pre> </div> </div> <div class="paragraph"> <p>An example of such a renamed directory might look like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>/hbase/WALs/srv.example.com,60020,1254173957298-splitting</pre> </div> </div> </li> <li> <p>Each log file is split, one at a time.</p> <div class="paragraph"> <p>The log splitter reads the log file one edit entry at a time and puts each edit entry into the buffer corresponding to the edit’s region. At the same time, the splitter starts several writer threads. Writer threads pick up a corresponding buffer and write the edit entries in the buffer to a temporary recovered edit file. The temporary edit file is stored to disk with the following naming pattern:</p> </div> <div class="listingblock"> <div class="content"> <pre>/hbase/<table_name>/<region_id>/recovered.edits/.temp</pre> </div> </div> <div class="paragraph"> <p>This file is used to store all the edits in the WAL log for this region. After log splitting completes, the <em>.temp</em> file is renamed to the sequence ID of the first log written to the file.</p> </div> <div class="paragraph"> <p>To determine whether all edits have been written, the sequence ID is compared to the sequence of the last edit that was written to the HFile. If the sequence of the last edit is greater than or equal to the sequence ID included in the file name, it is clear that all writes from the edit file have been completed.</p> </div> </li> <li> <p>After log splitting is complete, each affected region is assigned to a RegionServer.</p> <div class="paragraph"> <p>When the region is opened, the <em>recovered.edits</em> folder is checked for recovered edits files. If any such files are present, they are replayed by reading the edits and saving them to the MemStore. After all edit files are replayed, the contents of the MemStore are written to disk (HFile) and the edit files are deleted.</p> </div> </li> </ol> </div> <div class="sect4"> <h5 id="_handling_of_errors_during_log_splitting"><a class="anchor" href="#_handling_of_errors_during_log_splitting"></a>Handling of Errors During Log Splitting</h5> <div class="paragraph"> <p>If you set the <code>hbase.hlog.split.skip.errors</code> option to <code>true</code>, errors are treated as follows:</p> </div> <div class="ulist"> <ul> <li> <p>Any error encountered during splitting will be logged.</p> </li> <li> <p>The problematic WAL log will be moved into the <em>.corrupt</em> directory under the hbase <code>rootdir</code>,</p> </li> <li> <p>Processing of the WAL will continue</p> </li> </ul> </div> <div class="paragraph"> <p>If the <code>hbase.hlog.split.skip.errors</code> option is set to <code>false</code>, the default, the exception will be propagated and the split will be logged as failed. See <a href="https://issues.apache.org/jira/browse/HBASE-2958">HBASE-2958 When hbase.hlog.split.skip.errors is set to false, we fail the split but thats it</a>. We need to do more than just fail split if this flag is set.</p> </div> <div class="sect5"> <h6 id="_how_eofexceptions_are_treated_when_splitting_a_crashed_regionserver_s_wals"><a class="anchor" href="#_how_eofexceptions_are_treated_when_splitting_a_crashed_regionserver_s_wals"></a>How EOFExceptions are treated when splitting a crashed RegionServer’s WALs</h6> <div class="paragraph"> <p>If an EOFException occurs while splitting logs, the split proceeds even when <code>hbase.hlog.split.skip.errors</code> is set to <code>false</code>. An EOFException while reading the last log in the set of files to split is likely, because the RegionServer was likely in the process of writing a record at the time of a crash. For background, see <a href="https://issues.apache.org/jira/browse/HBASE-2643">HBASE-2643 Figure how to deal with eof splitting logs</a></p> </div> </div> </div> <div class="sect4"> <h5 id="_performance_improvements_during_log_splitting"><a class="anchor" href="#_performance_improvements_during_log_splitting"></a>Performance Improvements during Log Splitting</h5> <div class="paragraph"> <p>WAL log splitting and recovery can be resource intensive and take a long time, depending on the number of RegionServers involved in the crash and the size of the regions. <a href="#distributed.log.splitting">Distributed Log Splitting</a> and <a href="#distributed.log.replay">Distributed Log Replay</a> were developed to improve performance during log splitting.</p> </div> <div class="sect5"> <h6 id="distributed.log.splitting"><a class="anchor" href="#distributed.log.splitting"></a>Distributed Log Splitting</h6> <div class="paragraph"> <p><em>Distributed Log Splitting</em> was added in HBase version 0.92 (<a href="https://issues.apache.org/jira/browse/HBASE-1364">HBASE-1364</a>) by Prakash Khemani from Facebook. It reduces the time to complete log splitting dramatically, improving the availability of regions and tables. For example, recovering a crashed cluster took around 9 hours with single-threaded log splitting, but only about six minutes with distributed log splitting.</p> </div> <div class="paragraph"> <p>The information in this section is sourced from Jimmy Xiang’s blog post at <a href="http://blog.cloudera.com/blog/2012/07/hbase-log-splitting/" class="bare">http://blog.cloudera.com/blog/2012/07/hbase-log-splitting/</a>.</p> </div> <div class="paragraph"> <div class="title">Enabling or Disabling Distributed Log Splitting</div> <p>Distributed log processing is enabled by default since HBase 0.92. The setting is controlled by the <code>hbase.master.distributed.log.splitting</code> property, which can be set to <code>true</code> or <code>false</code>, but defaults to <code>true</code>.</p> </div> <div id="log.splitting.step.by.step" class="paragraph"> <div class="title">Distributed Log Splitting, Step by Step</div> <p>After configuring distributed log splitting, the HMaster controls the process. The HMaster enrolls each RegionServer in the log splitting process, and the actual work of splitting the logs is done by the RegionServers. The general process for log splitting, as described in <a href="#log.splitting.step.by.step">Distributed Log Splitting, Step by Step</a> still applies here.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>If distributed log processing is enabled, the HMaster creates a <em>split log manager</em> instance when the cluster is started.</p> <div class="olist loweralpha"> <ol class="loweralpha" type="a"> <li> <p>The split log manager manages all log files which need to be scanned and split.</p> </li> <li> <p>The split log manager places all the logs into the ZooKeeper splitlog node (<em>/hbase/splitlog</em>) as tasks.</p> </li> <li> <p>You can view the contents of the splitlog by issuing the following <code>zkCli</code> command. Example output is shown.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">ls /hbase/splitlog [hdfs%3A%2F%2Fhost2.sample.com%3A56020%2Fhbase%2F.logs%2Fhost8.sample.com%2C57020%2C1340474893275-splitting%2Fhost8.sample.com%253A57020.1340474893900, hdfs%3A%2F%2Fhost2.sample.com%3A56020%2Fhbase%2F.logs%2Fhost3.sample.com%2C57020%2C1340474893299-splitting%2Fhost3.sample.com%253A57020.1340474893931, hdfs%3A%2F%2Fhost2.sample.com%3A56020%2Fhbase%2F.logs%2Fhost4.sample.com%2C57020%2C1340474893287-splitting%2Fhost4.sample.com%253A57020.1340474893946]</code></pre> </div> </div> <div class="paragraph"> <p>The output contains some non-ASCII characters. When decoded, it looks much more simple:</p> </div> <div class="listingblock"> <div class="content"> <pre>[hdfs://host2.sample.com:56020/hbase/.logs /host8.sample.com,57020,1340474893275-splitting /host8.sample.com%3A57020.1340474893900, hdfs://host2.sample.com:56020/hbase/.logs /host3.sample.com,57020,1340474893299-splitting /host3.sample.com%3A57020.1340474893931, hdfs://host2.sample.com:56020/hbase/.logs /host4.sample.com,57020,1340474893287-splitting /host4.sample.com%3A57020.1340474893946]</pre> </div> </div> <div class="paragraph"> <p>The listing represents WAL file names to be scanned and split, which is a list of log splitting tasks.</p> </div> </li> </ol> </div> </li> <li> <p>The split log manager monitors the log-splitting tasks and workers.</p> <div class="paragraph"> <p>The split log manager is responsible for the following ongoing tasks:</p> </div> <div class="ulist"> <ul> <li> <p>Once the split log manager publishes all the tasks to the splitlog znode, it monitors these task nodes and waits for them to be processed.</p> </li> <li> <p>Checks to see if there are any dead split log workers queued up. If it finds tasks claimed by unresponsive workers, it will resubmit those tasks. If the resubmit fails due to some ZooKeeper exception, the dead worker is queued up again for retry.</p> </li> <li> <p>Checks to see if there are any unassigned tasks. If it finds any, it create an ephemeral rescan node so that each split log worker is notified to re-scan unassigned tasks via the <code>nodeChildrenChanged</code> ZooKeeper event.</p> </li> <li> <p>Checks for tasks which are assigned but expired. If any are found, they are moved back to <code>TASK_UNASSIGNED</code> state again so that they can be retried. It is possible that these tasks are assigned to slow workers, or they may already be finished. This is not a problem, because log splitting tasks have the property of idempotence. In other words, the same log splitting task can be processed many times without causing any problem.</p> </li> <li> <p>The split log manager watches the HBase split log znodes constantly. If any split log task node data is changed, the split log manager retrieves the node data. The node data contains the current state of the task. You can use the <code>zkCli</code> <code>get</code> command to retrieve the current state of a task. In the example output below, the first line of the output shows that the task is currently unassigned.</p> <div class="listingblock"> <div class="content"> <pre>get /hbase/splitlog/hdfs%3A%2F%2Fhost2.sample.com%3A56020%2Fhbase%2F.logs%2Fhost6.sample.com%2C57020%2C1340474893287-splitting%2Fhost6.sample.com%253A57020.1340474893945 unassigned host2.sample.com:57000 cZxid = 0×7115 ctime = Sat Jun 23 11:13:40 PDT 2012 ...</pre> </div> </div> <div class="paragraph"> <p>Based on the state of the task whose data is changed, the split log manager does one of the following:</p> </div> </li> <li> <p>Resubmit the task if it is unassigned</p> </li> <li> <p>Heartbeat the task if it is assigned</p> </li> <li> <p>Resubmit or fail the task if it is resigned (see <a href="#distributed.log.replay.failure.reasons">Reasons a Task Will Fail</a>)</p> </li> <li> <p>Resubmit or fail the task if it is completed with errors (see <a href="#distributed.log.replay.failure.reasons">Reasons a Task Will Fail</a>)</p> </li> <li> <p>Resubmit or fail the task if it could not complete due to errors (see <a href="#distributed.log.replay.failure.reasons">Reasons a Task Will Fail</a>)</p> </li> <li> <p>Delete the task if it is successfully completed or failed</p> <div id="distributed.log.replay.failure.reasons" class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Reasons a Task Will Fail</div> <div class="ulist"> <ul> <li> <p>The task has been deleted.</p> </li> <li> <p>The node no longer exists.</p> </li> <li> <p>The log status manager failed to move the state of the task to <code>TASK_UNASSIGNED</code>.</p> </li> <li> <p>The number of resubmits is over the resubmit threshold.</p> </li> </ul> </div> </td> </tr> </table> </div> </li> </ul> </div> </li> <li> <p>Each RegionServer’s split log worker performs the log-splitting tasks.</p> <div class="paragraph"> <p>Each RegionServer runs a daemon thread called the <em>split log worker</em>, which does the work to split the logs. The daemon thread starts when the RegionServer starts, and registers itself to watch HBase znodes. If any splitlog znode children change, it notifies a sleeping worker thread to wake up and grab more tasks. If if a worker’s current task’s node data is changed, the worker checks to see if the task has been taken by another worker. If so, the worker thread stops work on the current task.</p> </div> <div class="paragraph"> <p>The worker monitors the splitlog znode constantly. When a new task appears, the split log worker retrieves the task paths and checks each one until it finds an unclaimed task, which it attempts to claim. If the claim was successful, it attempts to perform the task and updates the task’s <code>state</code> property based on the splitting outcome. At this point, the split log worker scans for another unclaimed task.</p> </div> <div class="ulist"> <div class="title">How the Split Log Worker Approaches a Task</div> <ul> <li> <p>It queries the task state and only takes action if the task is in `TASK_UNASSIGNED `state.</p> </li> <li> <p>If the task is is in <code>TASK_UNASSIGNED</code> state, the worker attempts to set the state to <code>TASK_OWNED</code> by itself. If it fails to set the state, another worker will try to grab it. The split log manager will also ask all workers to rescan later if the task remains unassigned.</p> </li> <li> <p>If the worker succeeds in taking ownership of the task, it tries to get the task state again to make sure it really gets it asynchronously. In the meantime, it starts a split task executor to do the actual work:</p> <div class="ulist"> <ul> <li> <p>Get the HBase root folder, create a temp folder under the root, and split the log file to the temp folder.</p> </li> <li> <p>If the split was successful, the task executor sets the task to state <code>TASK_DONE</code>.</p> </li> <li> <p>If the worker catches an unexpected IOException, the task is set to state <code>TASK_ERR</code>.</p> </li> <li> <p>If the worker is shutting down, set the the task to state <code>TASK_RESIGNED</code>.</p> </li> <li> <p>If the task is taken by another worker, just log it.</p> </li> </ul> </div> </li> </ul> </div> </li> <li> <p>The split log manager monitors for uncompleted tasks.</p> <div class="paragraph"> <p>The split log manager returns when all tasks are completed successfully. If all tasks are completed with some failures, the split log manager throws an exception so that the log splitting can be retried. Due to an asynchronous implementation, in very rare cases, the split log manager loses track of some completed tasks. For that reason, it periodically checks for remaining uncompleted task in its task map or ZooKeeper. If none are found, it throws an exception so that the log splitting can be retried right away instead of hanging there waiting for something that won’t happen.</p> </div> </li> </ol> </div> </div> <div class="sect5"> <h6 id="distributed.log.replay"><a class="anchor" href="#distributed.log.replay"></a>Distributed Log Replay</h6> <div class="paragraph"> <p>After a RegionServer fails, its failed regions are assigned to another RegionServer, which are marked as "recovering" in ZooKeeper. A split log worker directly replays edits from the WAL of the failed RegionServer to the regions at its new location. When a region is in "recovering" state, it can accept writes but no reads (including Append and Increment), region splits or merges.</p> </div> <div class="paragraph"> <p>Distributed Log Replay extends the <a href="#distributed.log.splitting">Distributed Log Splitting</a> framework. It works by directly replaying WAL edits to another RegionServer instead of creating <em>recovered.edits</em> files. It provides the following advantages over distributed log splitting alone:</p> </div> <div class="ulist"> <ul> <li> <p>It eliminates the overhead of writing and reading a large number of <em>recovered.edits</em> files. It is not unusual for thousands of <em>recovered.edits</em> files to be created and written concurrently during a RegionServer recovery. Many small random writes can degrade overall system performance.</p> </li> <li> <p>It allows writes even when a region is in recovering state. It only takes seconds for a recovering region to accept writes again.</p> </li> </ul> </div> <div class="paragraph"> <div class="title">Enabling Distributed Log Replay</div> <p>To enable distributed log replay, set <code>hbase.master.distributed.log.replay</code> to <code>true</code>. This will be the default for HBase 0.99 (<a href="https://issues.apache.org/jira/browse/HBASE-10888">HBASE-10888</a>).</p> </div> <div class="paragraph"> <p>You must also enable HFile version 3 (which is the default HFile format starting in HBase 0.99. See <a href="https://issues.apache.org/jira/browse/HBASE-10855">HBASE-10855</a>). Distributed log replay is unsafe for rolling upgrades.</p> </div> </div> </div> </div> <div class="sect3"> <h4 id="wal.disable"><a class="anchor" href="#wal.disable"></a>67.6.5. Disabling the WAL</h4> <div class="paragraph"> <p>It is possible to disable the WAL, to improve performance in certain specific situations. However, disabling the WAL puts your data at risk. The only situation where this is recommended is during a bulk load. This is because, in the event of a problem, the bulk load can be re-run with no risk of data loss.</p> </div> <div class="paragraph"> <p>The WAL is disabled by calling the HBase client field <code>Mutation.writeToWAL(false)</code>. Use the <code>Mutation.setDurability(Durability.SKIP_WAL)</code> and Mutation.getDurability() methods to set and get the field’s value. There is no way to disable the WAL for only a specific table.</p> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> If you disable the WAL for anything other than bulk loads, your data is at risk. </td> </tr> </table> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="regions.arch"><a class="anchor" href="#regions.arch"></a>68. Regions</h2> <div class="sectionbody"> <div class="paragraph"> <p>Regions are the basic element of availability and distribution for tables, and are comprised of a Store per Column Family. The hierarchy of objects is as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre>Table (HBase table) Region (Regions for the table) Store (Store per ColumnFamily for each Region for the table) MemStore (MemStore for each Store for each Region for the table) StoreFile (StoreFiles for each Store for each Region for the table) Block (Blocks within a StoreFile within a Store for each Region for the table)</pre> </div> </div> <div class="paragraph"> <p>For a description of what HBase files look like when written to HDFS, see <a href="#trouble.namenode.hbase.objects">Browsing HDFS for HBase Objects</a>.</p> </div> <div class="sect2"> <h3 id="arch.regions.size"><a class="anchor" href="#arch.regions.size"></a>68.1. Considerations for Number of Regions</h3> <div class="paragraph"> <p>In general, HBase is designed to run with a small (20-200) number of relatively large (5-20Gb) regions per server. The considerations for this are as follows:</p> </div> <div class="sect3"> <h4 id="too_many_regions"><a class="anchor" href="#too_many_regions"></a>68.1.1. Why should I keep my Region count low?</h4> <div class="paragraph"> <p>Typically you want to keep your region count low on HBase for numerous reasons. Usually right around 100 regions per RegionServer has yielded the best results. Here are some of the reasons below for keeping region count low:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>MSLAB (MemStore-local allocation buffer) requires 2MB per MemStore (that’s 2MB per family per region). 1000 regions that have 2 families each is 3.9GB of heap used, and it’s not even storing data yet. NB: the 2MB value is configurable.</p> </li> <li> <p>If you fill all the regions at somewhat the same rate, the global memory usage makes it that it forces tiny flushes when you have too many regions which in turn generates compactions. Rewriting the same data tens of times is the last thing you want. An example is filling 1000 regions (with one family) equally and let’s consider a lower bound for global MemStore usage of 5GB (the region server would have a big heap). Once it reaches 5GB it will force flush the biggest region, at that point they should almost all have about 5MB of data so it would flush that amount. 5MB inserted later, it would flush another region that will now have a bit over 5MB of data, and so on. This is currently the main limiting factor for the number of regions; see <a href="#ops.capacity.regions.count">Number of regions per RS - upper bound</a> for detailed formula.</p> </li> <li> <p>The master as is is allergic to tons of regions, and will take a lot of time assigning them and moving them around in batches. The reason is that it’s heavy on ZK usage, and it’s not very async at the moment (could really be improved — and has been improved a bunch in 0.96 HBase).</p> </li> <li> <p>In older versions of HBase (pre-HFile v2, 0.90 and previous), tons of regions on a few RS can cause the store file index to rise, increasing heap usage and potentially creating memory pressure or OOME on the RSs</p> </li> </ol> </div> <div class="paragraph"> <p>Another issue is the effect of the number of regions on MapReduce jobs; it is typical to have one mapper per HBase region. Thus, hosting only 5 regions per RS may not be enough to get sufficient number of tasks for a MapReduce job, while 1000 regions will generate far too many tasks.</p> </div> <div class="paragraph"> <p>See <a href="#ops.capacity.regions">Determining region count and size</a> for configuration guidelines.</p> </div> </div> </div> <div class="sect2"> <h3 id="regions.arch.assignment"><a class="anchor" href="#regions.arch.assignment"></a>68.2. Region-RegionServer Assignment</h3> <div class="paragraph"> <p>This section describes how Regions are assigned to RegionServers.</p> </div> <div class="sect3"> <h4 id="regions.arch.assignment.startup"><a class="anchor" href="#regions.arch.assignment.startup"></a>68.2.1. Startup</h4> <div class="paragraph"> <p>When HBase starts regions are assigned as follows (short version):</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>The Master invokes the <code>AssignmentManager</code> upon startup.</p> </li> <li> <p>The <code>AssignmentManager</code> looks at the existing region assignments in <code>hbase:meta</code>.</p> </li> <li> <p>If the region assignment is still valid (i.e., if the RegionServer is still online) then the assignment is kept.</p> </li> <li> <p>If the assignment is invalid, then the <code>LoadBalancerFactory</code> is invoked to assign the region. The load balancer (<code>StochasticLoadBalancer</code> by default in HBase 1.0) assign the region to a RegionServer.</p> </li> <li> <p><code>hbase:meta</code> is updated with the RegionServer assignment (if needed) and the RegionServer start codes (start time of the RegionServer process) upon region opening by the RegionServer.</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="regions.arch.assignment.failover"><a class="anchor" href="#regions.arch.assignment.failover"></a>68.2.2. Failover</h4> <div class="paragraph"> <p>When a RegionServer fails:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>The regions immediately become unavailable because the RegionServer is down.</p> </li> <li> <p>The Master will detect that the RegionServer has failed.</p> </li> <li> <p>The region assignments will be considered invalid and will be re-assigned just like the startup sequence.</p> </li> <li> <p>In-flight queries are re-tried, and not lost.</p> </li> <li> <p>Operations are switched to a new RegionServer within the following amount of time:</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">ZooKeeper session timeout + split time + assignment/replay time</code></pre> </div> </div> </li> </ol> </div> </div> <div class="sect3"> <h4 id="regions.arch.balancer"><a class="anchor" href="#regions.arch.balancer"></a>68.2.3. Region Load Balancing</h4> <div class="paragraph"> <p>Regions can be periodically moved by the <a href="#master.processes.loadbalancer">LoadBalancer</a>.</p> </div> </div> <div class="sect3"> <h4 id="regions.arch.states"><a class="anchor" href="#regions.arch.states"></a>68.2.4. Region State Transition</h4> <div class="paragraph"> <p>HBase maintains a state for each region and persists the state in <code>hbase:meta</code>. The state of the <code>hbase:meta</code> region itself is persisted in ZooKeeper. You can see the states of regions in transition in the Master web UI. Following is the list of possible region states.</p> </div> <div class="ulist"> <div class="title">Possible Region States</div> <ul> <li> <p><code>OFFLINE</code>: the region is offline and not opening</p> </li> <li> <p><code>OPENING</code>: the region is in the process of being opened</p> </li> <li> <p><code>OPEN</code>: the region is open and the RegionServer has notified the master</p> </li> <li> <p><code>FAILED_OPEN</code>: the RegionServer failed to open the region</p> </li> <li> <p><code>CLOSING</code>: the region is in the process of being closed</p> </li> <li> <p><code>CLOSED</code>: the RegionServer has closed the region and notified the master</p> </li> <li> <p><code>FAILED_CLOSE</code>: the RegionServer failed to close the region</p> </li> <li> <p><code>SPLITTING</code>: the RegionServer notified the master that the region is splitting</p> </li> <li> <p><code>SPLIT</code>: the RegionServer notified the master that the region has finished splitting</p> </li> <li> <p><code>SPLITTING_NEW</code>: this region is being created by a split which is in progress</p> </li> <li> <p><code>MERGING</code>: the RegionServer notified the master that this region is being merged with another region</p> </li> <li> <p><code>MERGED</code>: the RegionServer notified the master that this region has been merged</p> </li> <li> <p><code>MERGING_NEW</code>: this region is being created by a merge of two regions</p> </li> </ul> </div> <div class="imageblock"> <div class="content"> <img src="images/region_states.png" alt="region states"> </div> <div class="title">Figure 2. Region State Transitions</div> </div> <div class="ulist"> <div class="title">Graph Legend</div> <ul> <li> <p>Brown: Offline state, a special state that can be transient (after closed before opening), terminal (regions of disabled tables), or initial (regions of newly created tables)</p> </li> <li> <p>Palegreen: Online state that regions can serve requests</p> </li> <li> <p>Lightblue: Transient states</p> </li> <li> <p>Red: Failure states that need OPS attention</p> </li> <li> <p>Gold: Terminal states of regions split/merged</p> </li> <li> <p>Grey: Initial states of regions created through split/merge</p> </li> </ul> </div> <div class="olist arabic"> <div class="title">Transition State Descriptions</div> <ol class="arabic"> <li> <p>The master moves a region from <code>OFFLINE</code> to <code>OPENING</code> state and tries to assign the region to a RegionServer. The RegionServer may or may not have received the open region request. The master retries sending the open region request to the RegionServer until the RPC goes through or the master runs out of retries. After the RegionServer receives the open region request, the RegionServer begins opening the region.</p> </li> <li> <p>If the master is running out of retries, the master prevents the RegionServer from opening the region by moving the region to <code>CLOSING</code> state and trying to close it, even if the RegionServer is starting to open the region.</p> </li> <li> <p>After the RegionServer opens the region, it continues to try to notify the master until the master moves the region to <code>OPEN</code> state and notifies the RegionServer. The region is now open.</p> </li> <li> <p>If the RegionServer cannot open the region, it notifies the master. The master moves the region to <code>CLOSED</code> state and tries to open the region on a different RegionServer.</p> </li> <li> <p>If the master cannot open the region on any of a certain number of regions, it moves the region to <code>FAILED_OPEN</code> state, and takes no further action until an operator intervenes from the HBase shell, or the server is dead.</p> </li> <li> <p>The master moves a region from <code>OPEN</code> to <code>CLOSING</code> state. The RegionServer holding the region may or may not have received the close region request. The master retries sending the close request to the server until the RPC goes through or the master runs out of retries.</p> </li> <li> <p>If the RegionServer is not online, or throws <code>NotServingRegionException</code>, the master moves the region to <code>OFFLINE</code> state and re-assigns it to a different RegionServer.</p> </li> <li> <p>If the RegionServer is online, but not reachable after the master runs out of retries, the master moves the region to <code>FAILED_CLOSE</code> state and takes no further action until an operator intervenes from the HBase shell, or the server is dead.</p> </li> <li> <p>If the RegionServer gets the close region request, it closes the region and notifies the master. The master moves the region to <code>CLOSED</code> state and re-assigns it to a different RegionServer.</p> </li> <li> <p>Before assigning a region, the master moves the region to <code>OFFLINE</code> state automatically if it is in <code>CLOSED</code> state.</p> </li> <li> <p>When a RegionServer is about to split a region, it notifies the master. The master moves the region to be split from <code>OPEN</code> to <code>SPLITTING</code> state and add the two new regions to be created to the RegionServer. These two regions are in <code>SPLITING_NEW</code> state initially.</p> </li> <li> <p>After notifying the master, the RegionServer starts to split the region. Once past the point of no return, the RegionServer notifies the master again so the master can update the <code>hbase:meta</code> table. However, the master does not update the region states until it is notified by the server that the split is done. If the split is successful, the splitting region is moved from <code>SPLITTING</code> to <code>SPLIT</code> state and the two new regions are moved from <code>SPLITTING_NEW</code> to <code>OPEN</code> state.</p> </li> <li> <p>If the split fails, the splitting region is moved from <code>SPLITTING</code> back to <code>OPEN</code> state, and the two new regions which were created are moved from <code>SPLITTING_NEW</code> to <code>OFFLINE</code> state.</p> </li> <li> <p>When a RegionServer is about to merge two regions, it notifies the master first. The master moves the two regions to be merged from <code>OPEN</code> to <code>MERGING</code> state, and adds the new region which will hold the contents of the merged regions region to the RegionServer. The new region is in <code>MERGING_NEW</code> state initially.</p> </li> <li> <p>After notifying the master, the RegionServer starts to merge the two regions. Once past the point of no return, the RegionServer notifies the master again so the master can update the META. However, the master does not update the region states until it is notified by the RegionServer that the merge has completed. If the merge is successful, the two merging regions are moved from <code>MERGING</code> to <code>MERGED</code> state and the new region is moved from <code>MERGING_NEW</code> to <code>OPEN</code> state.</p> </li> <li> <p>If the merge fails, the two merging regions are moved from <code>MERGING</code> back to <code>OPEN</code> state, and the new region which was created to hold the contents of the merged regions is moved from <code>MERGING_NEW</code> to <code>OFFLINE</code> state.</p> </li> <li> <p>For regions in <code>FAILED_OPEN</code> or <code>FAILED_CLOSE</code> states, the master tries to close them again when they are reassigned by an operator via HBase Shell.</p> </li> </ol> </div> </div> </div> <div class="sect2"> <h3 id="regions.arch.locality"><a class="anchor" href="#regions.arch.locality"></a>68.3. Region-RegionServer Locality</h3> <div class="paragraph"> <p>Over time, Region-RegionServer locality is achieved via HDFS block replication. The HDFS client does the following by default when choosing locations to write replicas:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>First replica is written to local node</p> </li> <li> <p>Second replica is written to a random node on another rack</p> </li> <li> <p>Third replica is written on the same rack as the second, but on a different node chosen randomly</p> </li> <li> <p>Subsequent replicas are written on random nodes on the cluster. See <em>Replica Placement: The First Baby Steps</em> on this page: <a href="http://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html">HDFS Architecture</a></p> </li> </ol> </div> <div class="paragraph"> <p>Thus, HBase eventually achieves locality for a region after a flush or a compaction. In a RegionServer failover situation a RegionServer may be assigned regions with non-local StoreFiles (because none of the replicas are local), however as new data is written in the region, or the table is compacted and StoreFiles are re-written, they will become "local" to the RegionServer.</p> </div> <div class="paragraph"> <p>For more information, see <em>Replica Placement: The First Baby Steps</em> on this page: <a href="http://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html">HDFS Architecture</a> and also Lars George’s blog on <a href="http://www.larsgeorge.com/2010/05/hbase-file-locality-in-hdfs.html">HBase and HDFS locality</a>.</p> </div> </div> <div class="sect2"> <h3 id="arch.region.splits"><a class="anchor" href="#arch.region.splits"></a>68.4. Region Splits</h3> <div class="paragraph"> <p>Regions split when they reach a configured threshold. Below we treat the topic in short. For a longer exposition, see <a href="http://hortonworks.com/blog/apache-hbase-region-splitting-and-merging/">Apache HBase Region Splitting and Merging</a> by our Enis Soztutar.</p> </div> <div class="paragraph"> <p>Splits run unaided on the RegionServer; i.e. the Master does not participate. The RegionServer splits a region, offlines the split region and then adds the daughter regions to <code>hbase:meta</code>, opens daughters on the parent’s hosting RegionServer and then reports the split to the Master. See <a href="#disable.splitting">Managed Splitting</a> for how to manually manage splits (and for why you might do this).</p> </div> <div class="sect3"> <h4 id="_custom_split_policies"><a class="anchor" href="#_custom_split_policies"></a>68.4.1. Custom Split Policies</h4> <div class="paragraph"> <p>ou can override the default split policy using a custom <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/RegionSplitPolicy.html">RegionSplitPolicy</a>(HBase 0.94+). Typically a custom split policy should extend HBase’s default split policy: <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/IncreasingToUpperBoundRegionSplitPolicy.html">IncreasingToUpperBoundRegionSplitPolicy</a>.</p> </div> <div class="paragraph"> <p>The policy can set globally through the HBase configuration or on a per-table basis.</p> </div> <div class="listingblock"> <div class="title">Configuring the Split Policy Globally in <em>hbase-site.xml</em></div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.region.split.policy<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.regionserver.IncreasingToUpperBoundRegionSplitPolicy<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="listingblock"> <div class="title">Configuring a Split Policy On a Table Using the Java API</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HTableDescriptor tableDesc = <span class="keyword">new</span> HTableDescriptor(<span class="string"><span class="delimiter">"</span><span class="content">test</span><span class="delimiter">"</span></span>); tableDesc.setValue(HTableDescriptor.SPLIT_POLICY, ConstantSizeRegionSplitPolicy.class.getName()); tableDesc.addFamily(<span class="keyword">new</span> HColumnDescriptor(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">cf1</span><span class="delimiter">"</span></span>))); admin.createTable(tableDesc); ----</code></pre> </div> </div> <div class="listingblock"> <div class="title">Configuring the Split Policy On a Table Using HBase Shell</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase> create <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, {METHOD => <span class="string"><span class="delimiter">'</span><span class="content">table_att</span><span class="delimiter">'</span></span>, CONFIG => {<span class="string"><span class="delimiter">'</span><span class="content">SPLIT_POLICY</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">org.apache.hadoop.hbase.regionserver.ConstantSizeRegionSplitPolicy</span><span class="delimiter">'</span></span>}}, {NAME => <span class="string"><span class="delimiter">'</span><span class="content">cf1</span><span class="delimiter">'</span></span>}</code></pre> </div> </div> <div class="paragraph"> <p>The default split policy can be overwritten using a custom <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/RegionSplitPolicy.html">RegionSplitPolicy(HBase 0.94+)</a>. Typically a custom split policy should extend HBase’s default split policy: <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/ConstantSizeRegionSplitPolicy.html">ConstantSizeRegionSplitPolicy</a>.</p> </div> <div class="paragraph"> <p>The policy can be set globally through the HBaseConfiguration used or on a per table basis:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HTableDescriptor myHtd = ...; myHtd.setValue(HTableDescriptor.SPLIT_POLICY, MyCustomSplitPolicy.class.getName());</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="manual_region_splitting_decisions"><a class="anchor" href="#manual_region_splitting_decisions"></a>68.5. Manual Region Splitting</h3> <div class="paragraph"> <p>It is possible to manually split your table, either at table creation (pre-splitting), or at a later time as an administrative action. You might choose to split your region for one or more of the following reasons. There may be other valid reasons, but the need to manually split your table might also point to problems with your schema design.</p> </div> <div class="ulist"> <div class="title">Reasons to Manually Split Your Table</div> <ul> <li> <p>Your data is sorted by timeseries or another similar algorithm that sorts new data at the end of the table. This means that the Region Server holding the last region is always under load, and the other Region Servers are idle, or mostly idle. See also <a href="#timeseries">Monotonically Increasing Row Keys/Timeseries Data</a>.</p> </li> <li> <p>You have developed an unexpected hotspot in one region of your table. For instance, an application which tracks web searches might be inundated by a lot of searches for a celebrity in the event of news about that celebrity. See <a href="#perf.one.region">perf.one.region</a> for more discussion about this particular scenario.</p> </li> <li> <p>After a big increase in the number of RegionServers in your cluster, to get the load spread out quickly.</p> </li> <li> <p>Before a bulk-load which is likely to cause unusual and uneven load across regions.</p> </li> </ul> </div> <div class="paragraph"> <p>See <a href="#disable.splitting">Managed Splitting</a> for a discussion about the dangers and possible benefits of managing splitting completely manually.</p> </div> <div class="sect3"> <h4 id="_determining_split_points"><a class="anchor" href="#_determining_split_points"></a>68.5.1. Determining Split Points</h4> <div class="paragraph"> <p>The goal of splitting your table manually is to improve the chances of balancing the load across the cluster in situations where good rowkey design alone won’t get you there. Keeping that in mind, the way you split your regions is very dependent upon the characteristics of your data. It may be that you already know the best way to split your table. If not, the way you split your table depends on what your keys are like.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Alphanumeric Rowkeys</dt> <dd> <p>If your rowkeys start with a letter or number, you can split your table at letter or number boundaries. For instance, the following command creates a table with regions that split at each vowel, so the first region has A-D, the second region has E-H, the third region has I-N, the fourth region has O-V, and the fifth region has U-Z.</p> </dd> <dt class="hdlist1">Using a Custom Algorithm</dt> <dd> <p>The RegionSplitter tool is provided with HBase, and uses a <em>SplitAlgorithm</em> to determine split points for you. As parameters, you give it the algorithm, desired number of regions, and column families. It includes two split algorithms. The first is the <code><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/util/RegionSplitter.HexStringSplit.html">HexStringSplit</a></code> algorithm, which assumes the row keys are hexadecimal strings. The second, <code><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/util/RegionSplitter.UniformSplit.html">UniformSplit</a></code>, assumes the row keys are random byte arrays. You will probably need to develop your own <code><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/util/RegionSplitter.SplitAlgorithm.html">SplitAlgorithm</a></code>, using the provided ones as models.</p> </dd> </dl> </div> </div> </div> <div class="sect2"> <h3 id="_online_region_merges"><a class="anchor" href="#_online_region_merges"></a>68.6. Online Region Merges</h3> <div class="paragraph"> <p>Both Master and RegionServer participate in the event of online region merges. Client sends merge RPC to the master, then the master moves the regions together to the RegionServer where the more heavily loaded region resided. Finally the master sends the merge request to this RegionServer which then runs the merge. Similar to process of region splitting, region merges run as a local transaction on the RegionServer. It offlines the regions and then merges two regions on the file system, atomically delete merging regions from <code>hbase:meta</code> and adds the merged region to <code>hbase:meta</code>, opens the merged region on the RegionServer and reports the merge to the Master.</p> </div> <div class="paragraph"> <p>An example of region merges in the HBase shell</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ hbase> merge_region 'ENCODED_REGIONNAME', 'ENCODED_REGIONNAME' $ hbase> merge_region 'ENCODED_REGIONNAME', 'ENCODED_REGIONNAME', true</code></pre> </div> </div> <div class="paragraph"> <p>It’s an asynchronous operation and call returns immediately without waiting merge completed. Passing <code>true</code> as the optional third parameter will force a merge. Normally only adjacent regions can be merged. The <code>force</code> parameter overrides this behaviour and is for expert use only.</p> </div> </div> <div class="sect2"> <h3 id="_store"><a class="anchor" href="#_store"></a>68.7. Store</h3> <div class="paragraph"> <p>A Store hosts a MemStore and 0 or more StoreFiles (HFiles). A Store corresponds to a column family for a table for a given region.</p> </div> <div class="sect3"> <h4 id="store.memstore"><a class="anchor" href="#store.memstore"></a>68.7.1. MemStore</h4> <div class="paragraph"> <p>The MemStore holds in-memory modifications to the Store. Modifications are Cells/KeyValues. When a flush is requested, the current MemStore is moved to a snapshot and is cleared. HBase continues to serve edits from the new MemStore and backing snapshot until the flusher reports that the flush succeeded. At this point, the snapshot is discarded. Note that when the flush happens, MemStores that belong to the same region will all be flushed.</p> </div> </div> <div class="sect3"> <h4 id="_memstore_flush"><a class="anchor" href="#_memstore_flush"></a>68.7.2. MemStore Flush</h4> <div class="paragraph"> <p>A MemStore flush can be triggered under any of the conditions listed below. The minimum flush unit is per region, not at individual MemStore level.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>When a MemStore reaches the size specified by <code>hbase.hregion.memstore.flush.size</code>, all MemStores that belong to its region will be flushed out to disk.</p> </li> <li> <p>When the overall MemStore usage reaches the value specified by <code>hbase.regionserver.global.memstore.upperLimit</code>, MemStores from various regions will be flushed out to disk to reduce overall MemStore usage in a RegionServer. The flush order is based on the descending order of a region’s MemStore usage. Regions will have their MemStores flushed until the overall MemStore usage drops to or slightly below <code>hbase.regionserver.global.memstore.lowerLimit</code>.</p> </li> <li> <p>When the number of WAL per region server reaches the value specified in <code>hbase.regionserver.max.logs</code>, MemStores from various regions will be flushed out to disk to reduce WAL count. The flush order is based on time. Regions with the oldest MemStores are flushed first until WAL count drops below <code>hbase.regionserver.max.logs</code>.</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="hregion.scans"><a class="anchor" href="#hregion.scans"></a>68.7.3. Scans</h4> <div class="ulist"> <ul> <li> <p>When a client issues a scan against a table, HBase generates <code>RegionScanner</code> objects, one per region, to serve the scan request.</p> </li> <li> <p>The <code>RegionScanner</code> object contains a list of <code>StoreScanner</code> objects, one per column family.</p> </li> <li> <p>Each <code>StoreScanner</code> object further contains a list of <code>StoreFileScanner</code> objects, corresponding to each StoreFile and HFile of the corresponding column family, and a list of <code>KeyValueScanner</code> objects for the MemStore.</p> </li> <li> <p>The two lists are merged into one, which is sorted in ascending order with the scan object for the MemStore at the end of the list.</p> </li> <li> <p>When a <code>StoreFileScanner</code> object is constructed, it is associated with a <code>MultiVersionConcurrencyControl</code> read point, which is the current <code>memstoreTS</code>, filtering out any new updates beyond the read point.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="hfile"><a class="anchor" href="#hfile"></a>68.7.4. StoreFile (HFile)</h4> <div class="paragraph"> <p>StoreFiles are where your data lives.</p> </div> <div class="sect4"> <h5 id="_hfile_format"><a class="anchor" href="#_hfile_format"></a>HFile Format</h5> <div class="paragraph"> <p>The <em>HFile</em> file format is based on the SSTable file described in the <a href="http://research.google.com/archive/bigtable.html">BigTable [2006</a>] paper and on Hadoop’s <a href="http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop/io/file/tfile/TFile.html">TFile</a> (The unit test suite and the compression harness were taken directly from TFile). Schubert Zhang’s blog post on <a href="http://cloudepr.blogspot.com/2009/09/hfile-block-indexed-file-format-to.html">HFile: A Block-Indexed File Format to Store Sorted Key-Value Pairs</a> makes for a thorough introduction to HBase’s HFile. Matteo Bertozzi has also put up a helpful description, <a href="http://th30z.blogspot.com/2011/02/hbase-io-hfile.html?spref=tw">HBase I/O: HFile</a>.</p> </div> <div class="paragraph"> <p>For more information, see the <a href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/io/hfile/HFile.html">HFile source code</a>. Also see <a href="#hfilev2">HBase file format with inline blocks (version 2)</a> for information about the HFile v2 format that was included in 0.92.</p> </div> </div> <div class="sect4"> <h5 id="_hfile_tool"><a class="anchor" href="#_hfile_tool"></a>HFile Tool</h5> <div class="paragraph"> <p>To view a textualized version of HFile content, you can use the <code>org.apache.hadoop.hbase.io.hfile.HFile</code> tool. Type the following to see usage:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ ${HBASE_HOME}/bin/hbase org.apache.hadoop.hbase.io.hfile.HFile</code></pre> </div> </div> <div class="paragraph"> <p>For example, to view the content of the file <em>hdfs://10.81.47.41:8020/hbase/TEST/1418428042/DSMP/4759508618286845475</em>, type the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash"> $ ${HBASE_HOME}/bin/hbase org.apache.hadoop.hbase.io.hfile.HFile -v -f hdfs://10.81.47.41:8020/hbase/TEST/1418428042/DSMP/4759508618286845475</code></pre> </div> </div> <div class="paragraph"> <p>If you leave off the option -v to see just a summary on the HFile. See usage for other things to do with the <code>HFile</code> tool.</p> </div> </div> <div class="sect4"> <h5 id="store.file.dir"><a class="anchor" href="#store.file.dir"></a>StoreFile Directory Structure on HDFS</h5> <div class="paragraph"> <p>For more information of what StoreFiles look like on HDFS with respect to the directory structure, see <a href="#trouble.namenode.hbase.objects">Browsing HDFS for HBase Objects</a>.</p> </div> </div> </div> <div class="sect3"> <h4 id="hfile.blocks"><a class="anchor" href="#hfile.blocks"></a>68.7.5. Blocks</h4> <div class="paragraph"> <p>StoreFiles are composed of blocks. The blocksize is configured on a per-ColumnFamily basis.</p> </div> <div class="paragraph"> <p>Compression happens at the block level within StoreFiles. For more information on compression, see <a href="#compression">Compression and Data Block Encoding In HBase</a>.</p> </div> <div class="paragraph"> <p>For more information on blocks, see the <a href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/io/hfile/HFileBlock.html">HFileBlock source code</a>.</p> </div> </div> <div class="sect3"> <h4 id="_keyvalue"><a class="anchor" href="#_keyvalue"></a>68.7.6. KeyValue</h4> <div class="paragraph"> <p>The KeyValue class is the heart of data storage in HBase. KeyValue wraps a byte array and takes offsets and lengths into the passed array which specify where to start interpreting the content as KeyValue.</p> </div> <div class="paragraph"> <p>The KeyValue format inside a byte array is:</p> </div> <div class="ulist"> <ul> <li> <p>keylength</p> </li> <li> <p>valuelength</p> </li> <li> <p>key</p> </li> <li> <p>value</p> </li> </ul> </div> <div class="paragraph"> <p>The Key is further decomposed as:</p> </div> <div class="ulist"> <ul> <li> <p>rowlength</p> </li> <li> <p>row (i.e., the rowkey)</p> </li> <li> <p>columnfamilylength</p> </li> <li> <p>columnfamily</p> </li> <li> <p>columnqualifier</p> </li> <li> <p>timestamp</p> </li> <li> <p>keytype (e.g., Put, Delete, DeleteColumn, DeleteFamily)</p> </li> </ul> </div> <div class="paragraph"> <p>KeyValue instances are <em>not</em> split across blocks. For example, if there is an 8 MB KeyValue, even if the block-size is 64kb this KeyValue will be read in as a coherent block. For more information, see the <a href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/KeyValue.html">KeyValue source code</a>.</p> </div> <div class="sect4"> <h5 id="keyvalue.example"><a class="anchor" href="#keyvalue.example"></a>Example</h5> <div class="paragraph"> <p>To emphasize the points above, examine what happens with two Puts for two different columns for the same row:</p> </div> <div class="ulist"> <ul> <li> <p>Put #1: <code>rowkey=row1, cf:attr1=value1</code></p> </li> <li> <p>Put #2: <code>rowkey=row1, cf:attr2=value2</code></p> </li> </ul> </div> <div class="paragraph"> <p>Even though these are for the same row, a KeyValue is created for each column:</p> </div> <div class="paragraph"> <p>Key portion for Put #1:</p> </div> <div class="ulist"> <ul> <li> <p><code>rowlength -----------→ 4</code></p> </li> <li> <p><code>row -----------------→ row1</code></p> </li> <li> <p><code>columnfamilylength --→ 2</code></p> </li> <li> <p><code>columnfamily --------→ cf</code></p> </li> <li> <p><code>columnqualifier -----→ attr1</code></p> </li> <li> <p><code>timestamp -----------→ server time of Put</code></p> </li> <li> <p><code>keytype -------------→ Put</code></p> </li> </ul> </div> <div class="paragraph"> <p>Key portion for Put #2:</p> </div> <div class="ulist"> <ul> <li> <p><code>rowlength -----------→ 4</code></p> </li> <li> <p><code>row -----------------→ row1</code></p> </li> <li> <p><code>columnfamilylength --→ 2</code></p> </li> <li> <p><code>columnfamily --------→ cf</code></p> </li> <li> <p><code>columnqualifier -----→ attr2</code></p> </li> <li> <p><code>timestamp -----------→ server time of Put</code></p> </li> <li> <p><code>keytype -------------→ Put</code></p> </li> </ul> </div> <div class="paragraph"> <p>It is critical to understand that the rowkey, ColumnFamily, and column (aka columnqualifier) are embedded within the KeyValue instance. The longer these identifiers are, the bigger the KeyValue is.</p> </div> </div> </div> <div class="sect3"> <h4 id="_compaction"><a class="anchor" href="#_compaction"></a>68.7.7. Compaction</h4> <div class="ulist"> <div class="title">Ambiguous Terminology</div> <ul> <li> <p>A <em>StoreFile</em> is a facade of HFile. In terms of compaction, use of StoreFile seems to have prevailed in the past.</p> </li> <li> <p>A <em>Store</em> is the same thing as a ColumnFamily. StoreFiles are related to a Store, or ColumnFamily.</p> </li> <li> <p>If you want to read more about StoreFiles versus HFiles and Stores versus ColumnFamilies, see <a href="https://issues.apache.org/jira/browse/HBASE-11316">HBASE-11316</a>.</p> </li> </ul> </div> <div class="paragraph"> <p>When the MemStore reaches a given size (<code>hbase.hregion.memstore.flush.size</code>), it flushes its contents to a StoreFile. The number of StoreFiles in a Store increases over time. <em>Compaction</em> is an operation which reduces the number of StoreFiles in a Store, by merging them together, in order to increase performance on read operations. Compactions can be resource-intensive to perform, and can either help or hinder performance depending on many factors.</p> </div> <div class="paragraph"> <p>Compactions fall into two categories: minor and major. Minor and major compactions differ in the following ways.</p> </div> <div class="paragraph"> <p><em>Minor compactions</em> usually select a small number of small, adjacent StoreFiles and rewrite them as a single StoreFile. Minor compactions do not drop (filter out) deletes or expired versions, because of potential side effects. See <a href="#compaction.and.deletes">[compaction.and.deletes]</a> and <a href="#compaction.and.versions">[compaction.and.versions]</a> for information on how deletes and versions are handled in relation to compactions. The end result of a minor compaction is fewer, larger StoreFiles for a given Store.</p> </div> <div class="paragraph"> <p>The end result of a <em>major compaction</em> is a single StoreFile per Store. Major compactions also process delete markers and max versions. See <a href="#compaction.and.deletes">[compaction.and.deletes]</a> and <a href="#compaction.and.versions">[compaction.and.versions]</a> for information on how deletes and versions are handled in relation to compactions.</p> </div> <div class="paragraph"> <div class="title">Compaction and Deletions</div> <p>When an explicit deletion occurs in HBase, the data is not actually deleted. Instead, a <em>tombstone</em> marker is written. The tombstone marker prevents the data from being returned with queries. During a major compaction, the data is actually deleted, and the tombstone marker is removed from the StoreFile. If the deletion happens because of an expired TTL, no tombstone is created. Instead, the expired data is filtered out and is not written back to the compacted StoreFile.</p> </div> <div class="paragraph"> <div class="title">Compaction and Versions</div> <p>When you create a Column Family, you can specify the maximum number of versions to keep, by specifying <code>HColumnDescriptor.setMaxVersions(int versions)</code>. The default value is <code>3</code>. If more versions than the specified maximum exist, the excess versions are filtered out and not written back to the compacted StoreFile.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Major Compactions Can Impact Query Results</div> <div class="paragraph"> <p>In some situations, older versions can be inadvertently resurrected if a newer version is explicitly deleted. See <a href="#major.compactions.change.query.results">Major compactions change query results</a> for a more in-depth explanation. This situation is only possible before the compaction finishes.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>In theory, major compactions improve performance. However, on a highly loaded system, major compactions can require an inappropriate number of resources and adversely affect performance. In a default configuration, major compactions are scheduled automatically to run once in a 7-day period. This is sometimes inappropriate for systems in production. You can manage major compactions manually. See <a href="#managed.compactions">Managed Compactions</a>.</p> </div> <div class="paragraph"> <p>Compactions do not perform region merges. See <a href="#ops.regionmgt.merge">Merge</a> for more information on region merging.</p> </div> <div class="sect4"> <h5 id="compaction.file.selection"><a class="anchor" href="#compaction.file.selection"></a>Compaction Policy - HBase 0.96.x and newer</h5> <div class="paragraph"> <p>Compacting large StoreFiles, or too many StoreFiles at once, can cause more IO load than your cluster is able to handle without causing performance problems. The method by which HBase selects which StoreFiles to include in a compaction (and whether the compaction is a minor or major compaction) is called the <em>compaction policy</em>.</p> </div> <div class="paragraph"> <p>Prior to HBase 0.96.x, there was only one compaction policy. That original compaction policy is still available as <code>RatioBasedCompactionPolicy</code>. The new compaction default policy, called <code>ExploringCompactionPolicy</code>, was subsequently backported to HBase 0.94 and HBase 0.95, and is the default in HBase 0.96 and newer. It was implemented in <a href="https://issues.apache.org/jira/browse/HBASE-7842">HBASE-7842</a>. In short, <code>ExploringCompactionPolicy</code> attempts to select the best possible set of StoreFiles to compact with the least amount of work, while the <code>RatioBasedCompactionPolicy</code> selects the first set that meets the criteria.</p> </div> <div class="paragraph"> <p>Regardless of the compaction policy used, file selection is controlled by several configurable parameters and happens in a multi-step approach. These parameters will be explained in context, and then will be given in a table which shows their descriptions, defaults, and implications of changing them.</p> </div> <div class="sect5"> <h6 id="compaction.being.stuck"><a class="anchor" href="#compaction.being.stuck"></a>Being Stuck</h6> <div class="paragraph"> <p>When the MemStore gets too large, it needs to flush its contents to a StoreFile. However, a Store can only have <code>hbase.hstore.blockingStoreFiles</code> files, so the MemStore needs to wait for the number of StoreFiles to be reduced by one or more compactions. However, if the MemStore grows larger than <code>hbase.hregion.memstore.flush.size</code>, it is not able to flush its contents to a StoreFile. If the MemStore is too large and the number of StoreFiles is also too high, the algorithm is said to be "stuck". The compaction algorithm checks for this "stuck" situation and provides mechanisms to alleviate it.</p> </div> </div> <div class="sect5"> <h6 id="exploringcompaction.policy"><a class="anchor" href="#exploringcompaction.policy"></a>The ExploringCompactionPolicy Algorithm</h6> <div class="paragraph"> <p>The ExploringCompactionPolicy algorithm considers each possible set of adjacent StoreFiles before choosing the set where compaction will have the most benefit.</p> </div> <div class="paragraph"> <p>One situation where the ExploringCompactionPolicy works especially well is when you are bulk-loading data and the bulk loads create larger StoreFiles than the StoreFiles which are holding data older than the bulk-loaded data. This can "trick" HBase into choosing to perform a major compaction each time a compaction is needed, and cause a lot of extra overhead. With the ExploringCompactionPolicy, major compactions happen much less frequently because minor compactions are more efficient.</p> </div> <div class="paragraph"> <p>In general, ExploringCompactionPolicy is the right choice for most situations, and thus is the default compaction policy. You can also use ExploringCompactionPolicy along with <a href="#ops.stripe">Experimental: Stripe Compactions</a>.</p> </div> <div class="paragraph"> <p>The logic of this policy can be examined in <em><a href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/regionserver/compactions/ExploringCompactionPolicy.html">hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/compactions/ExploringCompactionPolicy.java</a></em>. The following is a walk-through of the logic of the ExploringCompactionPolicy.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Make a list of all existing StoreFiles in the Store. The rest of the algorithm filters this list to come up with the subset of HFiles which will be chosen for compaction.</p> </li> <li> <p>If this was a user-requested compaction, attempt to perform the requested compaction type, regardless of what would normally be chosen. Note that even if the user requests a major compaction, it may not be possible to perform a major compaction. This may be because not all StoreFiles in the Column Family are available to compact or because there are too many Stores in the Column Family.</p> </li> <li> <p>Some StoreFiles are automatically excluded from consideration. These include:</p> <div class="ulist"> <ul> <li> <p>StoreFiles that are larger than <code>hbase.hstore.compaction.max.size</code></p> </li> <li> <p>StoreFiles that were created by a bulk-load operation which explicitly excluded compaction. You may decide to exclude StoreFiles resulting from bulk loads, from compaction. To do this, specify the <code>hbase.mapreduce.hfileoutputformat.compaction.exclude</code> parameter during the bulk load operation.</p> </li> </ul> </div> </li> <li> <p>Iterate through the list from step 1, and make a list of all potential sets of StoreFiles to compact together. A potential set is a grouping of <code>hbase.hstore.compaction.min</code> contiguous StoreFiles in the list. For each set, perform some sanity-checking and figure out whether this is the best compaction that could be done:</p> <div class="ulist"> <ul> <li> <p>If the number of StoreFiles in this set (not the size of the StoreFiles) is fewer than <code>hbase.hstore.compaction.min</code> or more than <code>hbase.hstore.compaction.max</code>, take it out of consideration.</p> </li> <li> <p>Compare the size of this set of StoreFiles with the size of the smallest possible compaction that has been found in the list so far. If the size of this set of StoreFiles represents the smallest compaction that could be done, store it to be used as a fall-back if the algorithm is "stuck" and no StoreFiles would otherwise be chosen. See <a href="#compaction.being.stuck">Being Stuck</a>.</p> </li> <li> <p>Do size-based sanity checks against each StoreFile in this set of StoreFiles.</p> <div class="ulist"> <ul> <li> <p>If the size of this StoreFile is larger than <code>hbase.hstore.compaction.max.size</code>, take it out of consideration.</p> </li> <li> <p>If the size is greater than or equal to <code>hbase.hstore.compaction.min.size</code>, sanity-check it against the file-based ratio to see whether it is too large to be considered.</p> <div class="paragraph"> <p>The sanity-checking is successful if:</p> </div> </li> <li> <p>There is only one StoreFile in this set, or</p> </li> <li> <p>For each StoreFile, its size multiplied by <code>hbase.hstore.compaction.ratio</code> (or <code>hbase.hstore.compaction.ratio.offpeak</code> if off-peak hours are configured and it is during off-peak hours) is less than the sum of the sizes of the other HFiles in the set.</p> </li> </ul> </div> </li> </ul> </div> </li> <li> <p>If this set of StoreFiles is still in consideration, compare it to the previously-selected best compaction. If it is better, replace the previously-selected best compaction with this one.</p> </li> <li> <p>When the entire list of potential compactions has been processed, perform the best compaction that was found. If no StoreFiles were selected for compaction, but there are multiple StoreFiles, assume the algorithm is stuck (see <a href="#compaction.being.stuck">Being Stuck</a>) and if so, perform the smallest compaction that was found in step 3.</p> </li> </ol> </div> </div> <div class="sect5"> <h6 id="compaction.ratiobasedcompactionpolicy.algorithm"><a class="anchor" href="#compaction.ratiobasedcompactionpolicy.algorithm"></a>RatioBasedCompactionPolicy Algorithm</h6> <div class="paragraph"> <p>The RatioBasedCompactionPolicy was the only compaction policy prior to HBase 0.96, though ExploringCompactionPolicy has now been backported to HBase 0.94 and 0.95. To use the RatioBasedCompactionPolicy rather than the ExploringCompactionPolicy, set <code>hbase.hstore.defaultengine.compactionpolicy.class</code> to <code>RatioBasedCompactionPolicy</code> in the <em>hbase-site.xml</em> file. To switch back to the ExploringCompactionPolicy, remove the setting from the <em>hbase-site.xml</em>.</p> </div> <div class="paragraph"> <p>The following section walks you through the algorithm used to select StoreFiles for compaction in the RatioBasedCompactionPolicy.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>The first phase is to create a list of all candidates for compaction. A list is created of all StoreFiles not already in the compaction queue, and all StoreFiles newer than the newest file that is currently being compacted. This list of StoreFiles is ordered by the sequence ID. The sequence ID is generated when a Put is appended to the write-ahead log (WAL), and is stored in the metadata of the HFile.</p> </li> <li> <p>Check to see if the algorithm is stuck (see <a href="#compaction.being.stuck">Being Stuck</a>, and if so, a major compaction is forced. This is a key area where <a href="#exploringcompaction.policy">The ExploringCompactionPolicy Algorithm</a> is often a better choice than the RatioBasedCompactionPolicy.</p> </li> <li> <p>If the compaction was user-requested, try to perform the type of compaction that was requested. Note that a major compaction may not be possible if all HFiles are not available for compaction or if too many StoreFiles exist (more than <code>hbase.hstore.compaction.max</code>).</p> </li> <li> <p>Some StoreFiles are automatically excluded from consideration. These include:</p> <div class="ulist"> <ul> <li> <p>StoreFiles that are larger than <code>hbase.hstore.compaction.max.size</code></p> </li> <li> <p>StoreFiles that were created by a bulk-load operation which explicitly excluded compaction. You may decide to exclude StoreFiles resulting from bulk loads, from compaction. To do this, specify the <code>hbase.mapreduce.hfileoutputformat.compaction.exclude</code> parameter during the bulk load operation.</p> </li> </ul> </div> </li> <li> <p>The maximum number of StoreFiles allowed in a major compaction is controlled by the <code>hbase.hstore.compaction.max</code> parameter. If the list contains more than this number of StoreFiles, a minor compaction is performed even if a major compaction would otherwise have been done. However, a user-requested major compaction still occurs even if there are more than <code>hbase.hstore.compaction.max</code> StoreFiles to compact.</p> </li> <li> <p>If the list contains fewer than <code>hbase.hstore.compaction.min</code> StoreFiles to compact, a minor compaction is aborted. Note that a major compaction can be performed on a single HFile. Its function is to remove deletes and expired versions, and reset locality on the StoreFile.</p> </li> <li> <p>The value of the <code>hbase.hstore.compaction.ratio</code> parameter is multiplied by the sum of StoreFiles smaller than a given file, to determine whether that StoreFile is selected for compaction during a minor compaction. For instance, if hbase.hstore.compaction.ratio is 1.2, FileX is 5MB, FileY is 2MB, and FileZ is 3MB:</p> <div class="listingblock"> <div class="content"> <pre>5 <= 1.2 x (2 + 3) or 5 <= 6</pre> </div> </div> <div class="paragraph"> <p>In this scenario, FileX is eligible for minor compaction. If FileX were 7MB, it would not be eligible for minor compaction. This ratio favors smaller StoreFile. You can configure a different ratio for use in off-peak hours, using the parameter <code>hbase.hstore.compaction.ratio.offpeak</code>, if you also configure <code>hbase.offpeak.start.hour</code> and <code>hbase.offpeak.end.hour</code>.</p> </div> </li> <li> <p>If the last major compaction was too long ago and there is more than one StoreFile to be compacted, a major compaction is run, even if it would otherwise have been minor. By default, the maximum time between major compactions is 7 days, plus or minus a 4.8 hour period, and determined randomly within those parameters. Prior to HBase 0.96, the major compaction period was 24 hours. See <code>hbase.hregion.majorcompaction</code> in the table below to tune or disable time-based major compactions.</p> </li> </ol> </div> </div> <div class="sect5"> <h6 id="compaction.parameters"><a class="anchor" href="#compaction.parameters"></a>Parameters Used by Compaction Algorithm</h6> <div class="paragraph"> <p>This table contains the main configuration parameters for compaction. This list is not exhaustive. To tune these parameters from the defaults, edit the <em>hbase-default.xml</em> file. For a full list of all configuration parameters available, see <a href="#config.files">config.files</a></p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Parameter</th> <th class="tableblock halign-left valign-top">Description</th> <th class="tableblock halign-left valign-top">Default</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hstore.compaction.min</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The minimum number of StoreFiles which must be eligible for compaction before compaction can run. The goal of tuning <code>hbase.hstore.compaction.min</code> is to avoid ending up with too many tiny StoreFiles to compact. Setting this value to 2 would cause a minor compaction each time you have two StoreFiles in a Store, and this is probably not appropriate. If you set this value too high, all the other values will need to be adjusted accordingly. For most cases, the default value is appropriate. In previous versions of HBase, the parameter hbase.hstore.compaction.min was called <code>hbase.hstore.compactionThreshold</code>.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">3</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hstore.compaction.max</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The maximum number of StoreFiles which will be selected for a single minor compaction, regardless of the number of eligible StoreFiles. Effectively, the value of hbase.hstore.compaction.max controls the length of time it takes a single compaction to complete. Setting it larger means that more StoreFiles are included in a compaction. For most cases, the default value is appropriate.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">10</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hstore.compaction.min.size</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>A StoreFile smaller than this size will always be eligible for minor compaction. StoreFiles this size or larger are evaluated by <code>hbase.hstore.compaction.ratio</code> to determine if they are eligible. Because this limit represents the "automatic include" limit for all StoreFiles smaller than this value, this value may need to be reduced in write-heavy environments where many files in the 1-2 MB range are being flushed, because every StoreFile will be targeted for compaction and the resulting StoreFiles may still be under the minimum size and require further compaction. If this parameter is lowered, the ratio check is triggered more quickly. This addressed some issues seen in earlier versions of HBase but changing this parameter is no longer necessary in most situations.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">128 MB</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hstore.compaction.max.size</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>An StoreFile larger than this size will be excluded from compaction. The effect of raising <code>hbase.hstore.compaction.max.size</code> is fewer, larger StoreFiles that do not get compacted often. If you feel that compaction is happening too often without much benefit, you can try raising this value.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>Long.MAX_VALUE</code></p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hstore.compaction.ratio</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>For minor compaction, this ratio is used to determine whether a given StoreFile which is larger than <code>hbase.hstore.compaction.min.size</code> is eligible for compaction. Its effect is to limit compaction of large StoreFile. The value of <code>hbase.hstore.compaction.ratio</code> is expressed as a floating-point decimal.</p> </div> <div class="ulist"> <ul> <li> <p>A large ratio, such as 10, will produce a single giant StoreFile. Conversely, a value of .25, will produce behavior similar to the BigTable compaction algorithm, producing four StoreFiles.</p> </li> <li> <p>A moderate value of between 1.0 and 1.4 is recommended. When tuning this value, you are balancing write costs with read costs. Raising the value (to something like 1.4) will have more write costs, because you will compact larger StoreFiles. However, during reads, HBase will need to seek through fewer StoreFiles to accomplish the read. Consider this approach if you cannot take advantage of <a href="#bloom">[bloom]</a>.</p> </li> <li> <p>Alternatively, you can lower this value to something like 1.0 to reduce the background cost of writes, and use to limit the number of StoreFiles touched during reads. For most cases, the default value is appropriate.</p> </li> </ul> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>1.2F</code></p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hstore.compaction.ratio.offpeak</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The compaction ratio used during off-peak compactions, if off-peak hours are also configured (see below). Expressed as a floating-point decimal. This allows for more aggressive (or less aggressive, if you set it lower than <code>hbase.hstore.compaction.ratio</code>) compaction during a set time period. Ignored if off-peak is disabled (default). This works the same as hbase.hstore.compaction.ratio.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>5.0F</code></p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.offpeak.start.hour</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The start of off-peak hours, expressed as an integer between 0 and 23, inclusive. Set to -1 to disable off-peak.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>-1</code> (disabled)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.offpeak.end.hour</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The end of off-peak hours, expressed as an integer between 0 and 23, inclusive. Set to -1 to disable off-peak.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>-1</code> (disabled)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.regionserver.thread.compaction.throttle</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>There are two different thread pools for compactions, one for large compactions and the other for small compactions. This helps to keep compaction of lean tables (such as <code>hbase:meta</code>) fast. If a compaction is larger than this threshold, it goes into the large compaction pool. In most cases, the default value is appropriate.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>2 x hbase.hstore.compaction.max x hbase.hregion.memstore.flush.size</code> (which defaults to <code>128</code>)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hregion.majorcompaction</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>Time between major compactions, expressed in milliseconds. Set to 0 to disable time-based automatic major compactions. User-requested and size-based major compactions will still run. This value is multiplied by <code>hbase.hregion.majorcompaction.jitter</code> to cause compaction to start at a somewhat-random time during a given window of time.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock">7 days (<code>604800000</code> milliseconds)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.hregion.majorcompaction.jitter</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>A multiplier applied to hbase.hregion.majorcompaction to cause compaction to occur a given amount of time either side of <code>hbase.hregion.majorcompaction</code>. The smaller the number, the closer the compactions will happen to the <code>hbase.hregion.majorcompaction</code> interval. Expressed as a floating-point decimal.</p> </div></div></td> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>.50F</code></p></td> </tr> </tbody> </table> </div> </div> <div class="sect4"> <h5 id="compaction.file.selection.old"><a class="anchor" href="#compaction.file.selection.old"></a>Compaction File Selection</h5> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Legacy Information</div> <div class="paragraph"> <p>This section has been preserved for historical reasons and refers to the way compaction worked prior to HBase 0.96.x. You can still use this behavior if you enable <a href="#compaction.ratiobasedcompactionpolicy.algorithm">RatioBasedCompactionPolicy Algorithm</a>. For information on the way that compactions work in HBase 0.96.x and later, see <a href="#compaction">[compaction]</a>.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>To understand the core algorithm for StoreFile selection, there is some ASCII-art in the <a href="http://hbase.apache.org/xref/org/apache/hadoop/hbase/regionserver/Store.html#836">Store source code</a> that will serve as useful reference.</p> </div> <div class="paragraph"> <p>It has been copied below:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="comment">/* normal skew: * * older ----> newer * _ * | | _ * | | | | _ * --|-|- |-|- |-|---_-------_------- minCompactSize * | | | | | | | | _ | | * | | | | | | | | | | | | * | | | | | | | | | | | | */</span></code></pre> </div> </div> <div class="ulist"> <div class="title">Important knobs:</div> <ul> <li> <p><code>hbase.hstore.compaction.ratio</code> Ratio used in compaction file selection algorithm (default 1.2f).</p> </li> <li> <p><code>hbase.hstore.compaction.min</code> (in HBase v 0.90 this is called <code>hbase.hstore.compactionThreshold</code>) (files) Minimum number of StoreFiles per Store to be selected for a compaction to occur (default 2).</p> </li> <li> <p><code>hbase.hstore.compaction.max</code> (files) Maximum number of StoreFiles to compact per minor compaction (default 10).</p> </li> <li> <p><code>hbase.hstore.compaction.min.size</code> (bytes) Any StoreFile smaller than this setting with automatically be a candidate for compaction. Defaults to <code>hbase.hregion.memstore.flush.size</code> (128 mb).</p> </li> <li> <p><code>hbase.hstore.compaction.max.size</code> (.92) (bytes) Any StoreFile larger than this setting with automatically be excluded from compaction (default Long.MAX_VALUE).</p> </li> </ul> </div> <div class="paragraph"> <p>The minor compaction StoreFile selection logic is size based, and selects a file for compaction when the <code>file ⇐ sum(smaller_files) * hbase.hstore.compaction.ratio</code>.</p> </div> <div class="sect5"> <h6 id="compaction.file.selection.example1"><a class="anchor" href="#compaction.file.selection.example1"></a>Minor Compaction File Selection - Example #1 (Basic Example)</h6> <div class="paragraph"> <p>This example mirrors an example from the unit test <code>TestCompactSelection</code>.</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.hstore.compaction.ratio</code> = 1.0f</p> </li> <li> <p><code>hbase.hstore.compaction.min</code> = 3 (files)</p> </li> <li> <p><code>hbase.hstore.compaction.max</code> = 5 (files)</p> </li> <li> <p><code>hbase.hstore.compaction.min.size</code> = 10 (bytes)</p> </li> <li> <p><code>hbase.hstore.compaction.max.size</code> = 1000 (bytes)</p> </li> </ul> </div> <div class="paragraph"> <p>The following StoreFiles exist: 100, 50, 23, 12, and 12 bytes apiece (oldest to newest). With the above parameters, the files that would be selected for minor compaction are 23, 12, and 12.</p> </div> <div class="paragraph"> <p>Why?</p> </div> <div class="ulist"> <ul> <li> <p>100 → No, because sum(50, 23, 12, 12) * 1.0 = 97.</p> </li> <li> <p>50 → No, because sum(23, 12, 12) * 1.0 = 47.</p> </li> <li> <p>23 → Yes, because sum(12, 12) * 1.0 = 24.</p> </li> <li> <p>12 → Yes, because the previous file has been included, and because this does not exceed the the max-file limit of 5</p> </li> <li> <p>12 → Yes, because the previous file had been included, and because this does not exceed the the max-file limit of 5.</p> </li> </ul> </div> </div> <div class="sect5"> <h6 id="compaction.file.selection.example2"><a class="anchor" href="#compaction.file.selection.example2"></a>Minor Compaction File Selection - Example #2 (Not Enough Files ToCompact)</h6> <div class="paragraph"> <p>This example mirrors an example from the unit test <code>TestCompactSelection</code>.</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.hstore.compaction.ratio</code> = 1.0f</p> </li> <li> <p><code>hbase.hstore.compaction.min</code> = 3 (files)</p> </li> <li> <p><code>hbase.hstore.compaction.max</code> = 5 (files)</p> </li> <li> <p><code>hbase.hstore.compaction.min.size</code> = 10 (bytes)</p> </li> <li> <p><code>hbase.hstore.compaction.max.size</code> = 1000 (bytes)</p> </li> </ul> </div> <div class="paragraph"> <p>The following StoreFiles exist: 100, 25, 12, and 12 bytes apiece (oldest to newest). With the above parameters, no compaction will be started.</p> </div> <div class="paragraph"> <p>Why?</p> </div> <div class="ulist"> <ul> <li> <p>100 → No, because sum(25, 12, 12) * 1.0 = 47</p> </li> <li> <p>25 → No, because sum(12, 12) * 1.0 = 24</p> </li> <li> <p>12 → No. Candidate because sum(12) * 1.0 = 12, there are only 2 files to compact and that is less than the threshold of 3</p> </li> <li> <p>12 → No. Candidate because the previous StoreFile was, but there are not enough files to compact</p> </li> </ul> </div> </div> <div class="sect5"> <h6 id="compaction.file.selection.example3"><a class="anchor" href="#compaction.file.selection.example3"></a>Minor Compaction File Selection - Example #3 (Limiting Files To Compact)</h6> <div class="paragraph"> <p>This example mirrors an example from the unit test <code>TestCompactSelection</code>.</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.hstore.compaction.ratio</code> = 1.0f</p> </li> <li> <p><code>hbase.hstore.compaction.min</code> = 3 (files)</p> </li> <li> <p><code>hbase.hstore.compaction.max</code> = 5 (files)</p> </li> <li> <p><code>hbase.hstore.compaction.min.size</code> = 10 (bytes)</p> </li> <li> <p><code>hbase.hstore.compaction.max.size</code> = 1000 (bytes)</p> </li> </ul> </div> <div class="paragraph"> <p>The following StoreFiles exist: 7, 6, 5, 4, 3, 2, and 1 bytes apiece (oldest to newest). With the above parameters, the files that would be selected for minor compaction are 7, 6, 5, 4, 3.</p> </div> <div class="paragraph"> <p>Why?</p> </div> <div class="ulist"> <ul> <li> <p>7 → Yes, because sum(6, 5, 4, 3, 2, 1) * 1.0 = 21. Also, 7 is less than the min-size</p> </li> <li> <p>6 → Yes, because sum(5, 4, 3, 2, 1) * 1.0 = 15. Also, 6 is less than the min-size.</p> </li> <li> <p>5 → Yes, because sum(4, 3, 2, 1) * 1.0 = 10. Also, 5 is less than the min-size.</p> </li> <li> <p>4 → Yes, because sum(3, 2, 1) * 1.0 = 6. Also, 4 is less than the min-size.</p> </li> <li> <p>3 → Yes, because sum(2, 1) * 1.0 = 3. Also, 3 is less than the min-size.</p> </li> <li> <p>2 → No. Candidate because previous file was selected and 2 is less than the min-size, but the max-number of files to compact has been reached.</p> </li> <li> <p>1 → No. Candidate because previous file was selected and 1 is less than the min-size, but max-number of files to compact has been reached.</p> </li> </ul> </div> <div id="compaction.config.impact" class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Impact of Key Configuration Options</div> This information is now included in the configuration parameter table in <a href="#compaction.configuration.parameters">[compaction.configuration.parameters]</a>. </td> </tr> </table> </div> </div> </div> <div class="sect4"> <h5 id="ops.stripe"><a class="anchor" href="#ops.stripe"></a>Experimental: Stripe Compactions</h5> <div class="paragraph"> <p>Stripe compactions is an experimental feature added in HBase 0.98 which aims to improve compactions for large regions or non-uniformly distributed row keys. In order to achieve smaller and/or more granular compactions, the StoreFiles within a region are maintained separately for several row-key sub-ranges, or "stripes", of the region. The stripes are transparent to the rest of HBase, so other operations on the HFiles or data work without modification.</p> </div> <div class="paragraph"> <p>Stripe compactions change the HFile layout, creating sub-regions within regions. These sub-regions are easier to compact, and should result in fewer major compactions. This approach alleviates some of the challenges of larger regions.</p> </div> <div class="paragraph"> <p>Stripe compaction is fully compatible with <a href="#compaction">[compaction]</a> and works in conjunction with either the ExploringCompactionPolicy or RatioBasedCompactionPolicy. It can be enabled for existing tables, and the table will continue to operate normally if it is disabled later.</p> </div> </div> <div class="sect4"> <h5 id="ops.stripe.when"><a class="anchor" href="#ops.stripe.when"></a>When To Use Stripe Compactions</h5> <div class="paragraph"> <p>Consider using stripe compaction if you have either of the following:</p> </div> <div class="ulist"> <ul> <li> <p>Large regions. You can get the positive effects of smaller regions without additional overhead for MemStore and region management overhead.</p> </li> <li> <p>Non-uniform keys, such as time dimension in a key. Only the stripes receiving the new keys will need to compact. Old data will not compact as often, if at all</p> </li> </ul> </div> <div class="paragraph"> <div class="title">Performance Improvements</div> <p>Performance testing has shown that the performance of reads improves somewhat, and variability of performance of reads and writes is greatly reduced. An overall long-term performance improvement is seen on large non-uniform-row key regions, such as a hash-prefixed timestamp key. These performance gains are the most dramatic on a table which is already large. It is possible that the performance improvement might extend to region splits.</p> </div> <div class="sect5"> <h6 id="ops.stripe.enable"><a class="anchor" href="#ops.stripe.enable"></a>Enabling Stripe Compaction</h6> <div class="paragraph"> <p>You can enable stripe compaction for a table or a column family, by setting its <code>hbase.hstore.engine.class</code> to <code>org.apache.hadoop.hbase.regionserver.StripeStoreEngine</code>. You also need to set the <code>hbase.hstore.blockingStoreFiles</code> to a high number, such as 100 (rather than the default value of 10).</p> </div> <div class="olist arabic"> <div class="title">Procedure: Enable Stripe Compaction</div> <ol class="arabic"> <li> <p>If the table already exists, disable the table.</p> </li> <li> <p>Run one of following commands in the HBase shell. Replace the table name <code>orders_table</code> with the name of your table.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">alter</span> <span class="string"><span class="delimiter">'</span><span class="content">orders_table</span><span class="delimiter">'</span></span>, CONFIGURATION => {<span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.engine.class</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">org.apache.hadoop.hbase.regionserver.StripeStoreEngine</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.blockingStoreFiles</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">100</span><span class="delimiter">'</span></span>} <span class="class">alter</span> <span class="string"><span class="delimiter">'</span><span class="content">orders_table</span><span class="delimiter">'</span></span>, {NAME => <span class="string"><span class="delimiter">'</span><span class="content">blobs_cf</span><span class="delimiter">'</span></span>, CONFIGURATION => {<span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.engine.class</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">org.apache.hadoop.hbase.regionserver.StripeStoreEngine</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.blockingStoreFiles</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">100</span><span class="delimiter">'</span></span>}} <span class="class">create</span> <span class="string"><span class="delimiter">'</span><span class="content">orders_table</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">blobs_cf</span><span class="delimiter">'</span></span>, CONFIGURATION => {<span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.engine.class</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">org.apache.hadoop.hbase.regionserver.StripeStoreEngine</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.blockingStoreFiles</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">100</span><span class="delimiter">'</span></span>}</code></pre> </div> </div> </li> <li> <p>Configure other options if needed. See <a href="#ops.stripe.config">Configuring Stripe Compaction</a> for more information.</p> </li> <li> <p>Enable the table.</p> </li> </ol> </div> <div class="olist arabic"> <div class="title">Procedure: Disable Stripe Compaction</div> <ol class="arabic"> <li> <p>Disable the table.</p> </li> <li> <p>Set the <code>hbase.hstore.engine.class</code> option to either nil or <code>org.apache.hadoop.hbase.regionserver.DefaultStoreEngine</code>. Either option has the same effect.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">alter</span> <span class="string"><span class="delimiter">'</span><span class="content">orders_table</span><span class="delimiter">'</span></span>, CONFIGURATION => {<span class="string"><span class="delimiter">'</span><span class="content">hbase.hstore.engine.class</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">rg.apache.hadoop.hbase.regionserver.DefaultStoreEngine</span><span class="delimiter">'</span></span>}</code></pre> </div> </div> </li> <li> <p>Enable the table.</p> </li> </ol> </div> <div class="paragraph"> <p>When you enable a large table after changing the store engine either way, a major compaction will likely be performed on most regions. This is not necessary on new tables.</p> </div> </div> <div class="sect5"> <h6 id="ops.stripe.config"><a class="anchor" href="#ops.stripe.config"></a>Configuring Stripe Compaction</h6> <div class="paragraph"> <p>Each of the settings for stripe compaction should be configured at the table or column family, after disabling the table. If you use HBase shell, the general command pattern is as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="sql"><span class="class">alter</span> <span class="string"><span class="delimiter">'</span><span class="content">orders_table</span><span class="delimiter">'</span></span>, CONFIGURATION => {<span class="string"><span class="delimiter">'</span><span class="content">key</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>, ..., <span class="string"><span class="delimiter">'</span><span class="content">key</span><span class="delimiter">'</span></span> => <span class="string"><span class="delimiter">'</span><span class="content">value</span><span class="delimiter">'</span></span>}}</code></pre> </div> </div> <div id="ops.stripe.config.sizing" class="paragraph"> <div class="title">Region and stripe sizing</div> <p>You can configure your stripe sizing based upon your region sizing. By default, your new regions will start with one stripe. On the next compaction after the stripe has grown too large (16 x MemStore flushes size), it is split into two stripes. Stripe splitting continues as the region grows, until the region is large enough to split.</p> </div> <div class="paragraph"> <p>You can improve this pattern for your own data. A good rule is to aim for a stripe size of at least 1 GB, and about 8-12 stripes for uniform row keys. For example, if your regions are 30 GB, 12 x 2.5 GB stripes might be a good starting point.</p> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 9. Stripe Sizing Settings</caption> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Setting</th> <th class="tableblock halign-left valign-top">Notes</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.store.stripe.initialStripeCount</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The number of stripes to create when stripe compaction is enabled. You can use it as follows:</p> </div> <div class="ulist"> <ul> <li> <p>For relatively uniform row keys, if you know the approximate target number of stripes from the above, you can avoid some splitting overhead by starting with several stripes (2, 5, 10…). If the early data is not representative of overall row key distribution, this will not be as efficient.</p> </li> <li> <p>For existing tables with a large amount of data, this setting will effectively pre-split your stripes.</p> </li> <li> <p>For keys such as hash-prefixed sequential keys, with more than one hash prefix per region, pre-splitting may make sense.</p> </li> </ul> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.store.stripe.sizeToSplit</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The maximum size a stripe grows before splitting. Use this in conjunction with <code>hbase.store.stripe.splitPartCount</code> to control the target stripe size (<code>sizeToSplit = splitPartsCount * target stripe size</code>), according to the above sizing considerations.</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock"><code>hbase.store.stripe.splitPartCount</code></p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>The number of new stripes to create when splitting a stripe. The default is 2, which is appropriate for most cases. For non-uniform row keys, you can experiment with increasing the number to 3 or 4, to isolate the arriving updates into narrower slice of the region without additional splits being required.</p> </div></div></td> </tr> </tbody> </table> <div id="ops.stripe.config.memstore" class="paragraph"> <div class="title">MemStore Size Settings</div> <p>By default, the flush creates several files from one MemStore, according to existing stripe boundaries and row keys to flush. This approach minimizes write amplification, but can be undesirable if the MemStore is small and there are many stripes, because the files will be too small.</p> </div> <div class="paragraph"> <p>In this type of situation, you can set <code>hbase.store.stripe.compaction.flushToL0</code> to <code>true</code>. This will cause a MemStore flush to create a single file instead. When at least <code>hbase.store.stripe.compaction.minFilesL0</code> such files (by default, 4) accumulate, they will be compacted into striped files.</p> </div> <div id="ops.stripe.config.compact" class="paragraph"> <div class="title">Normal Compaction Configuration and Stripe Compaction</div> <p>All the settings that apply to normal compactions (see <a href="#compaction.configuration.parameters">[compaction.configuration.parameters]</a>) apply to stripe compactions. The exceptions are the minimum and maximum number of files, which are set to higher values by default because the files in stripes are smaller. To control these for stripe compactions, use <code>hbase.store.stripe.compaction.minFiles</code> and <code>hbase.store.stripe.compaction.maxFiles</code>, rather than <code>hbase.hstore.compaction.min</code> and <code>hbase.hstore.compaction.max</code>.</p> </div> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="arch.bulk.load"><a class="anchor" href="#arch.bulk.load"></a>69. Bulk Loading</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="arch.bulk.load.overview"><a class="anchor" href="#arch.bulk.load.overview"></a>69.1. Overview</h3> <div class="paragraph"> <p>HBase includes several methods of loading data into tables. The most straightforward method is to either use the <code>TableOutputFormat</code> class from a MapReduce job, or use the normal client APIs; however, these are not always the most efficient methods.</p> </div> <div class="paragraph"> <p>The bulk load feature uses a MapReduce job to output table data in HBase’s internal data format, and then directly loads the generated StoreFiles into a running cluster. Using bulk load will use less CPU and network resources than simply using the HBase API.</p> </div> </div> <div class="sect2"> <h3 id="arch.bulk.load.limitations"><a class="anchor" href="#arch.bulk.load.limitations"></a>69.2. Bulk Load Limitations</h3> <div class="paragraph"> <p>As bulk loading bypasses the write path, the WAL doesn’t get written to as part of the process. Replication works by reading the WAL files so it won’t see the bulk loaded data – and the same goes for the edits that use <code>Put.setDurability(SKIP_WAL)</code>. One way to handle that is to ship the raw files or the HFiles to the other cluster and do the other processing there.</p> </div> </div> <div class="sect2"> <h3 id="arch.bulk.load.arch"><a class="anchor" href="#arch.bulk.load.arch"></a>69.3. Bulk Load Architecture</h3> <div class="paragraph"> <p>The HBase bulk load process consists of two main steps.</p> </div> <div class="sect3"> <h4 id="arch.bulk.load.prep"><a class="anchor" href="#arch.bulk.load.prep"></a>69.3.1. Preparing data via a MapReduce job</h4> <div class="paragraph"> <p>The first step of a bulk load is to generate HBase data files (StoreFiles) from a MapReduce job using <code>HFileOutputFormat2</code>. This output format writes out data in HBase’s internal storage format so that they can be later loaded very efficiently into the cluster.</p> </div> <div class="paragraph"> <p>In order to function efficiently, <code>HFileOutputFormat2</code> must be configured such that each output HFile fits within a single region. In order to do this, jobs whose output will be bulk loaded into HBase use Hadoop’s <code>TotalOrderPartitioner</code> class to partition the map output into disjoint ranges of the key space, corresponding to the key ranges of the regions in the table.</p> </div> <div class="paragraph"> <p><code>HFileOutputFormat2</code> includes a convenience function, <code>configureIncrementalLoad()</code>, which automatically sets up a <code>TotalOrderPartitioner</code> based on the current region boundaries of a table.</p> </div> </div> <div class="sect3"> <h4 id="arch.bulk.load.complete"><a class="anchor" href="#arch.bulk.load.complete"></a>69.3.2. Completing the data load</h4> <div class="paragraph"> <p>After a data import has been prepared, either by using the <code>importtsv</code> tool with the “importtsv.bulk.output” option or by some other MapReduce job using the <code>HFileOutputFormat</code>, the <code>completebulkload</code> tool is used to import the data into the running cluster. This command line tool iterates through the prepared data files, and for each one determines the region the file belongs to. It then contacts the appropriate RegionServer which adopts the HFile, moving it into its storage directory and making the data available to clients.</p> </div> <div class="paragraph"> <p>If the region boundaries have changed during the course of bulk load preparation, or between the preparation and completion steps, the <code>completebulkload</code> utility will automatically split the data files into pieces corresponding to the new boundaries. This process is not optimally efficient, so users should take care to minimize the delay between preparing a bulk load and importing it into the cluster, especially if other clients are simultaneously loading data through other means.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ hadoop jar hbase-server-VERSION.jar completebulkload [-c /path/to/hbase/config/hbase-site.xml] /user/todd/myoutput mytable</code></pre> </div> </div> <div class="paragraph"> <p>The <code>-c config-file</code> option can be used to specify a file containing the appropriate hbase parameters (e.g., hbase-site.xml) if not supplied already on the CLASSPATH (In addition, the CLASSPATH must contain the directory that has the zookeeper configuration file if zookeeper is NOT managed by HBase).</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> If the target table does not already exist in HBase, this tool will create the table automatically. </td> </tr> </table> </div> </div> </div> <div class="sect2"> <h3 id="arch.bulk.load.also"><a class="anchor" href="#arch.bulk.load.also"></a>69.4. See Also</h3> <div class="paragraph"> <p>For more information about the referenced utilities, see <a href="#importtsv">[importtsv]</a> and <a href="#completebulkload">[completebulkload]</a>.</p> </div> <div class="paragraph"> <p>See <a href="http://blog.cloudera.com/blog/2013/09/how-to-use-hbase-bulk-loading-and-why/">How-to: Use HBase Bulk Loading, and Why</a> for a recent blog on current state of bulk loading.</p> </div> </div> <div class="sect2"> <h3 id="arch.bulk.load.adv"><a class="anchor" href="#arch.bulk.load.adv"></a>69.5. Advanced Usage</h3> <div class="paragraph"> <p>Although the <code>importtsv</code> tool is useful in many cases, advanced users may want to generate data programatically, or import data from other formats. To get started doing so, dig into <code>ImportTsv.java</code> and check the JavaDoc for HFileOutputFormat.</p> </div> <div class="paragraph"> <p>The import step of the bulk load can also be done programmatically. See the <code>LoadIncrementalHFiles</code> class for more information.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="arch.hdfs"><a class="anchor" href="#arch.hdfs"></a>70. HDFS</h2> <div class="sectionbody"> <div class="paragraph"> <p>As HBase runs on HDFS (and each StoreFile is written as a file on HDFS), it is important to have an understanding of the HDFS Architecture especially in terms of how it stores files, handles failovers, and replicates blocks.</p> </div> <div class="paragraph"> <p>See the Hadoop documentation on <a href="http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html">HDFS Architecture</a> for more information.</p> </div> <div class="sect2"> <h3 id="arch.hdfs.nn"><a class="anchor" href="#arch.hdfs.nn"></a>70.1. NameNode</h3> <div class="paragraph"> <p>The NameNode is responsible for maintaining the filesystem metadata. See the above HDFS Architecture link for more information.</p> </div> </div> <div class="sect2"> <h3 id="arch.hdfs.dn"><a class="anchor" href="#arch.hdfs.dn"></a>70.2. DataNode</h3> <div class="paragraph"> <p>The DataNodes are responsible for storing HDFS blocks. See the above HDFS Architecture link for more information.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="arch.timelineconsistent.reads"><a class="anchor" href="#arch.timelineconsistent.reads"></a>71. Timeline-consistent High Available Reads</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="casestudies.timelineconsistent.intro"><a class="anchor" href="#casestudies.timelineconsistent.intro"></a>71.1. Introduction</h3> <div class="paragraph"> <p>HBase, architecturally, always had the strong consistency guarantee from the start. All reads and writes are routed through a single region server, which guarantees that all writes happen in an order, and all reads are seeing the most recent committed data.</p> </div> <div class="paragraph"> <p>However, because of this single homing of the reads to a single location, if the server becomes unavailable, the regions of the table that were hosted in the region server become unavailable for some time. There are three phases in the region recovery process - detection, assignment, and recovery. Of these, the detection is usually the longest and is presently in the order of 20-30 seconds depending on the ZooKeeper session timeout. During this time and before the recovery is complete, the clients will not be able to read the region data.</p> </div> <div class="paragraph"> <p>However, for some use cases, either the data may be read-only, or doing reads against some stale data is acceptable. With timeline-consistent high available reads, HBase can be used for these kind of latency-sensitive use cases where the application can expect to have a time bound on the read completion.</p> </div> <div class="paragraph"> <p>For achieving high availability for reads, HBase provides a feature called <em>region replication</em>. In this model, for each region of a table, there will be multiple replicas that are opened in different RegionServers. By default, the region replication is set to 1, so only a single region replica is deployed and there will not be any changes from the original model. If region replication is set to 2 or more, then the master will assign replicas of the regions of the table. The Load Balancer ensures that the region replicas are not co-hosted in the same region servers and also in the same rack (if possible).</p> </div> <div class="paragraph"> <p>All of the replicas for a single region will have a unique replica_id, starting from 0. The region replica having replica_id==0 is called the primary region, and the others <em>secondary regions</em> or secondaries. Only the primary can accept writes from the client, and the primary will always contain the latest changes. Since all writes still have to go through the primary region, the writes are not highly-available (meaning they might block for some time if the region becomes unavailable).</p> </div> </div> <div class="sect2"> <h3 id="_timeline_consistency"><a class="anchor" href="#_timeline_consistency"></a>71.2. Timeline Consistency</h3> <div class="paragraph"> <p>With this feature, HBase introduces a Consistency definition, which can be provided per read operation (get or scan).</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">enum</span> Consistency { STRONG, TIMELINE }</code></pre> </div> </div> <div class="paragraph"> <p><code>Consistency.STRONG</code> is the default consistency model provided by HBase. In case the table has region replication = 1, or in a table with region replicas but the reads are done with this consistency, the read is always performed by the primary regions, so that there will not be any change from the previous behaviour, and the client always observes the latest data.</p> </div> <div class="paragraph"> <p>In case a read is performed with <code>Consistency.TIMELINE</code>, then the read RPC will be sent to the primary region server first. After a short interval (<code>hbase.client.primaryCallTimeout.get</code>, 10ms by default), parallel RPC for secondary region replicas will also be sent if the primary does not respond back. After this, the result is returned from whichever RPC is finished first. If the response came back from the primary region replica, we can always know that the data is latest. For this Result.isStale() API has been added to inspect the staleness. If the result is from a secondary region, then Result.isStale() will be set to true. The user can then inspect this field to possibly reason about the data.</p> </div> <div class="paragraph"> <p>In terms of semantics, TIMELINE consistency as implemented by HBase differs from pure eventual consistency in these respects:</p> </div> <div class="ulist"> <ul> <li> <p>Single homed and ordered updates: Region replication or not, on the write side, there is still only 1 defined replica (primary) which can accept writes. This replica is responsible for ordering the edits and preventing conflicts. This guarantees that two different writes are not committed at the same time by different replicas and the data diverges. With this, there is no need to do read-repair or last-timestamp-wins kind of conflict resolution.</p> </li> <li> <p>The secondaries also apply the edits in the order that the primary committed them. This way the secondaries will contain a snapshot of the primaries data at any point in time. This is similar to RDBMS replications and even HBase’s own multi-datacenter replication, however in a single cluster.</p> </li> <li> <p>On the read side, the client can detect whether the read is coming from up-to-date data or is stale data. Also, the client can issue reads with different consistency requirements on a per-operation basis to ensure its own semantic guarantees.</p> </li> <li> <p>The client can still observe edits out-of-order, and can go back in time, if it observes reads from one secondary replica first, then another secondary replica. There is no stickiness to region replicas or a transaction-id based guarantee. If required, this can be implemented later though.</p> </li> </ul> </div> <div class="imageblock"> <div class="content"> <img src="images/timeline_consistency.png" alt="Timeline Consistency"> </div> <div class="title">Figure 3. Timeline Consistency</div> </div> <div class="paragraph"> <p>To better understand the TIMELINE semantics, lets look at the above diagram. Lets say that there are two clients, and the first one writes x=1 at first, then x=2 and x=3 later. As above, all writes are handled by the primary region replica. The writes are saved in the write ahead log (WAL), and replicated to the other replicas asynchronously. In the above diagram, notice that replica_id=1 received 2 updates, and its data shows that x=2, while the replica_id=2 only received a single update, and its data shows that x=1.</p> </div> <div class="paragraph"> <p>If client1 reads with STRONG consistency, it will only talk with the replica_id=0, and thus is guaranteed to observe the latest value of x=3. In case of a client issuing TIMELINE consistency reads, the RPC will go to all replicas (after primary timeout) and the result from the first response will be returned back. Thus the client can see either 1, 2 or 3 as the value of x. Let’s say that the primary region has failed and log replication cannot continue for some time. If the client does multiple reads with TIMELINE consistency, she can observe x=2 first, then x=1, and so on.</p> </div> </div> <div class="sect2"> <h3 id="_tradeoffs"><a class="anchor" href="#_tradeoffs"></a>71.3. Tradeoffs</h3> <div class="paragraph"> <p>Having secondary regions hosted for read availability comes with some tradeoffs which should be carefully evaluated per use case. Following are advantages and disadvantages.</p> </div> <div class="ulist"> <div class="title">Advantages</div> <ul> <li> <p>High availability for read-only tables</p> </li> <li> <p>High availability for stale reads</p> </li> <li> <p>Ability to do very low latency reads with very high percentile (99.9%+) latencies for stale reads</p> </li> </ul> </div> <div class="ulist"> <div class="title">Disadvantages</div> <ul> <li> <p>Double / Triple MemStore usage (depending on region replication count) for tables with region replication > 1</p> </li> <li> <p>Increased block cache usage</p> </li> <li> <p>Extra network traffic for log replication</p> </li> <li> <p>Extra backup RPCs for replicas</p> </li> </ul> </div> <div class="paragraph"> <p>To serve the region data from multiple replicas, HBase opens the regions in secondary mode in the region servers. The regions opened in secondary mode will share the same data files with the primary region replica, however each secondary region replica will have its own MemStore to keep the unflushed data (only primary region can do flushes). Also to serve reads from secondary regions, the blocks of data files may be also cached in the block caches for the secondary regions.</p> </div> </div> <div class="sect2"> <h3 id="_where_is_the_code"><a class="anchor" href="#_where_is_the_code"></a>71.4. Where is the code</h3> <div class="paragraph"> <p>This feature is delivered in two phases, Phase 1 and 2. The first phase is done in time for HBase-1.0.0 release. Meaning that using HBase-1.0.x, you can use all the features that are marked for Phase 1. Phase 2 is committed in HBase-1.1.0, meaning all HBase versions after 1.1.0 should contain Phase 2 items.</p> </div> </div> <div class="sect2"> <h3 id="_propagating_writes_to_region_replicas"><a class="anchor" href="#_propagating_writes_to_region_replicas"></a>71.5. Propagating writes to region replicas</h3> <div class="paragraph"> <p>As discussed above writes only go to the primary region replica. For propagating the writes from the primary region replica to the secondaries, there are two different mechanisms. For read-only tables, you do not need to use any of the following methods. Disabling and enabling the table should make the data available in all region replicas. For mutable tables, you have to use <strong>only</strong> one of the following mechanisms: storefile refresher, or async wal replication. The latter is recommeded.</p> </div> <div class="sect3"> <h4 id="_storefile_refresher"><a class="anchor" href="#_storefile_refresher"></a>71.5.1. StoreFile Refresher</h4> <div class="paragraph"> <p>The first mechanism is store file refresher which is introduced in HBase-1.0+. Store file refresher is a thread per region server, which runs periodically, and does a refresh operation for the store files of the primary region for the secondary region replicas. If enabled, the refresher will ensure that the secondary region replicas see the new flushed, compacted or bulk loaded files from the primary region in a timely manner. However, this means that only flushed data can be read back from the secondary region replicas, and after the refresher is run, making the secondaries lag behind the primary for an a longer time.</p> </div> <div class="paragraph"> <p>For turning this feature on, you should configure <code>hbase.regionserver.storefile.refresh.period</code> to a non-zero value. See Configuration section below.</p> </div> </div> <div class="sect3"> <h4 id="_asnyc_wal_replication"><a class="anchor" href="#_asnyc_wal_replication"></a>71.5.2. Asnyc WAL replication</h4> <div class="paragraph"> <p>The second mechanism for propagation of writes to secondaries is done via “Async WAL Replication” feature and is only available in HBase-1.1+. This works similarly to HBase’s multi-datacenter replication, but instead the data from a region is replicated to the secondary regions. Each secondary replica always receives and observes the writes in the same order that the primary region committed them. In some sense, this design can be thought of as “in-cluster replication”, where instead of replicating to a different datacenter, the data goes to secondary regions to keep secondary region’s in-memory state up to date. The data files are shared between the primary region and the other replicas, so that there is no extra storage overhead. However, the secondary regions will have recent non-flushed data in their memstores, which increases the memory overhead. The primary region writes flush, compaction, and bulk load events to its WAL as well, which are also replicated through wal replication to secondaries. When they observe the flush/compaction or bulk load event, the secondary regions replay the event to pick up the new files and drop the old ones.</p> </div> <div class="paragraph"> <p>Committing writes in the same order as in primary ensures that the secondaries won’t diverge from the primary regions data, but since the log replication is asynchronous, the data might still be stale in secondary regions. Since this feature works as a replication endpoint, the performance and latency characteristics is expected to be similar to inter-cluster replication.</p> </div> <div class="paragraph"> <p>Async WAL Replication is <strong>disabled</strong> by default. You can enable this feature by setting <code>hbase.region.replica.replication.enabled</code> to <code>true</code>. Asyn WAL Replication feature will add a new replication peer named <code>region_replica_replication</code> as a replication peer when you create a table with region replication > 1 for the first time. Once enabled, if you want to disable this feature, you need to do two actions: * Set configuration property <code>hbase.region.replica.replication.enabled</code> to false in <code>hbase-site.xml</code> (see Configuration section below) * Disable the replication peer named <code>region_replica_replication</code> in the cluster using hbase shell or <code>ReplicationAdmin</code> class:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> hbase> disable_peer 'region_replica_replication'</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="_store_file_ttl"><a class="anchor" href="#_store_file_ttl"></a>71.6. Store File TTL</h3> <div class="paragraph"> <p>In both of the write propagation approaches mentioned above, store files of the primary will be opened in secondaries independent of the primary region. So for files that the primary compacted away, the secondaries might still be referring to these files for reading. Both features are using HFileLinks to refer to files, but there is no protection (yet) for guaranteeing that the file will not be deleted prematurely. Thus, as a guard, you should set the configuration property <code>hbase.master.hfilecleaner.ttl</code> to a larger value, such as 1 hour to guarantee that you will not receive IOExceptions for requests going to replicas.</p> </div> </div> <div class="sect2"> <h3 id="_region_replication_for_meta_table_s_region"><a class="anchor" href="#_region_replication_for_meta_table_s_region"></a>71.7. Region replication for META table’s region</h3> <div class="paragraph"> <p>Currently, Async WAL Replication is not done for the META table’s WAL. The meta table’s secondary replicas still refreshes themselves from the persistent store files. Hence the <code>hbase.regionserver.meta.storefile.refresh.period</code> needs to be set to a certain non-zero value for refreshing the meta store files. Note that this configuration is configured differently than <code>hbase.regionserver.storefile.refresh.period</code>.</p> </div> </div> <div class="sect2"> <h3 id="_memory_accounting"><a class="anchor" href="#_memory_accounting"></a>71.8. Memory accounting</h3> <div class="paragraph"> <p>The secondary region replicas refer to the data files of the primary region replica, but they have their own memstores (in HBase-1.1+) and uses block cache as well. However, one distinction is that the secondary region replicas cannot flush the data when there is memory pressure for their memstores. They can only free up memstore memory when the primary region does a flush and this flush is replicated to the secondary. Since in a region server hosting primary replicas for some regions and secondaries for some others, the secondaries might cause extra flushes to the primary regions in the same host. In extreme situations, there can be no memory left for adding new writes coming from the primary via wal replication. For unblocking this situation (and since secondary cannot flush by itself), the secondary is allowed to do a “store file refresh” by doing a file system list operation to pick up new files from primary, and possibly dropping its memstore. This refresh will only be performed if the memstore size of the biggest secondary region replica is at least <code>hbase.region.replica.storefile.refresh.memstore.multiplier</code> (default 4) times bigger than the biggest memstore of a primary replica. One caveat is that if this is performed, the secondary can observe partial row updates across column families (since column families are flushed independently). The default should be good to not do this operation frequently. You can set this value to a large number to disable this feature if desired, but be warned that it might cause the replication to block forever.</p> </div> </div> <div class="sect2"> <h3 id="_secondary_replica_failover"><a class="anchor" href="#_secondary_replica_failover"></a>71.9. Secondary replica failover</h3> <div class="paragraph"> <p>When a secondary region replica first comes online, or fails over, it may have served some edits from it’s memstore. Since the recovery is handled differently for secondary replicas, the secondary has to ensure that it does not go back in time before it starts serving requests after assignment. For doing that, the secondary waits until it observes a full flush cycle (start flush, commit flush) or a “region open event” replicated from the primary. Until this happens, the secondary region replica will reject all read requests by throwing an IOException with message “The region’s reads are disabled”. However, the other replicas will probably still be available to read, thus not causing any impact for the rpc with TIMELINE consistency. To facilitate faster recovery, the secondary region will trigger a flush request from the primary when it is opened. The configuration property <code>hbase.region.replica.wait.for.primary.flush</code> (enabled by default) can be used to disable this feature if needed.</p> </div> </div> <div class="sect2"> <h3 id="_configuration_properties"><a class="anchor" href="#_configuration_properties"></a>71.10. Configuration properties</h3> <div class="paragraph"> <p>To use highly available reads, you should set the following properties in <code>hbase-site.xml</code> file. There is no specific configuration to enable or disable region replicas. Instead you can change the number of region replicas per table to increase or decrease at the table creation or with alter table. The following configuration is for using async wal replication and using meta replicas of 3.</p> </div> <div class="sect3"> <h4 id="_server_side_properties"><a class="anchor" href="#_server_side_properties"></a>71.10.1. Server side properties</h4> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.storefile.refresh.period<span class="tag"></name></span> <span class="tag"><value></span>0<span class="tag"></value></span> <span class="tag"><description></span> The period (in milliseconds) for refreshing the store files for the secondary regions. 0 means this feature is disabled. Secondary regions sees new files (from flushes and compactions) from primary once the secondary region refreshes the list of files in the region (there is no notification mechanism). But too frequent refreshes might cause extra Namenode pressure. If the files cannot be refreshed for longer than HFile TTL (hbase.master.hfilecleaner.ttl) the requests are rejected. Configuring HFile TTL to a larger value is also recommended with this setting. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.regionserver.meta.storefile.refresh.period<span class="tag"></name></span> <span class="tag"><value></span>300000<span class="tag"></value></span> <span class="tag"><description></span> The period (in milliseconds) for refreshing the store files for the hbase:meta tables secondary regions. 0 means this feature is disabled. Secondary regions sees new files (from flushes and compactions) from primary once the secondary region refreshes the list of files in the region (there is no notification mechanism). But too frequent refreshes might cause extra Namenode pressure. If the files cannot be refreshed for longer than HFile TTL (hbase.master.hfilecleaner.ttl) the requests are rejected. Configuring HFile TTL to a larger value is also recommended with this setting. This should be a non-zero number if meta replicas are enabled (via hbase.meta.replica.count set to greater than 1). <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.region.replica.replication.enabled<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span> Whether asynchronous WAL replication to the secondary region replicas is enabled or not. If this is enabled, a replication peer named "region_replica_replication" will be created which will tail the logs and replicate the mutatations to region replicas for tables that have region replication <span class="error">></span> 1. If this is enabled once, disabling this replication also requires disabling the replication peer using shell or ReplicationAdmin java class. Replication to secondary region replicas works over standard inter-cluster replication. So replication, if disabled explicitly, also has to be enabled by setting "hbase.replication"· to true for this feature to work. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.region.replica.replication.memstore.enabled<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span> If you set this to `false`, replicas do not receive memstore updates from the primary RegionServer. If you set this to `true`, you can still disable memstore replication on a per-table basis, by setting the table's `REGION_MEMSTORE_REPLICATION` configuration property to `false`. If memstore replication is disabled, the secondaries will only receive updates for events like flushes and bulkloads, and will not have access to data which the primary has not yet flushed. This preserves the guarantee of row-level consistency, even when the read requests `Consistency.TIMELINE`. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.master.hfilecleaner.ttl<span class="tag"></name></span> <span class="tag"><value></span>3600000<span class="tag"></value></span> <span class="tag"><description></span> The period (in milliseconds) to keep store files in the archive folder before deleting them from the file system.<span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.meta.replica.count<span class="tag"></name></span> <span class="tag"><value></span>3<span class="tag"></value></span> <span class="tag"><description></span> Region replication count for the meta regions. Defaults to 1. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.region.replica.storefile.refresh.memstore.multiplier<span class="tag"></name></span> <span class="tag"><value></span>4<span class="tag"></value></span> <span class="tag"><description></span> The multiplier for a “store file refresh” operation for the secondary region replica. If a region server has memory pressure, the secondary region will refresh it’s store files if the memstore size of the biggest secondary replica is bigger this many times than the memstore size of the biggest primary replica. Set this to a very big value to disable this feature (not recommended). <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.region.replica.wait.for.primary.flush<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span> Whether to wait for observing a full flush cycle from the primary before start serving data in a secondary. Disabling this might cause the secondary region replicas to go back in time for reads between region movements. <span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>One thing to keep in mind also is that, region replica placement policy is only enforced by the <code>StochasticLoadBalancer</code> which is the default balancer. If you are using a custom load balancer property in hbase-site.xml (<code>hbase.master.loadbalancer.class</code>) replicas of regions might end up being hosted in the same server.</p> </div> </div> <div class="sect3"> <h4 id="_client_side_properties"><a class="anchor" href="#_client_side_properties"></a>71.10.2. Client side properties</h4> <div class="paragraph"> <p>Ensure to set the following for all clients (and servers) that will use region replicas.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.ipc.client.specificThreadForWriting<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span> Whether to enable interruption of RPC threads at the client side. This is required for region replicas with fallback RPC’s to secondary regions. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.client.primaryCallTimeout.get<span class="tag"></name></span> <span class="tag"><value></span>10000<span class="tag"></value></span> <span class="tag"><description></span> The timeout (in microseconds), before secondary fallback RPC’s are submitted for get requests with Consistency.TIMELINE to the secondary replicas of the regions. Defaults to 10ms. Setting this lower will increase the number of RPC’s, but will lower the p99 latencies. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.client.primaryCallTimeout.multiget<span class="tag"></name></span> <span class="tag"><value></span>10000<span class="tag"></value></span> <span class="tag"><description></span> The timeout (in microseconds), before secondary fallback RPC’s are submitted for multi-get requests (Table.get(List<span class="tag"><Get></span>)) with Consistency.TIMELINE to the secondary replicas of the regions. Defaults to 10ms. Setting this lower will increase the number of RPC’s, but will lower the p99 latencies. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.client.replicaCallTimeout.scan<span class="tag"></name></span> <span class="tag"><value></span>1000000<span class="tag"></value></span> <span class="tag"><description></span> The timeout (in microseconds), before secondary fallback RPC’s are submitted for scan requests with Consistency.TIMELINE to the secondary replicas of the regions. Defaults to 1 sec. Setting this lower will increase the number of RPC’s, but will lower the p99 latencies. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.meta.replicas.use<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span> Whether to use meta table replicas or not. Default is false. <span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Note HBase-1.0.x users should use <code>hbase.ipc.client.allowsInterrupt</code> rather than <code>hbase.ipc.client.specificThreadForWriting</code>.</p> </div> </div> </div> <div class="sect2"> <h3 id="_user_interface"><a class="anchor" href="#_user_interface"></a>71.11. User Interface</h3> <div class="paragraph"> <p>In the masters user interface, the region replicas of a table are also shown together with the primary regions. You can notice that the replicas of a region will share the same start and end keys and the same region name prefix. The only difference would be the appended replica_id (which is encoded as hex), and the region encoded name will be different. You can also see the replica ids shown explicitly in the UI.</p> </div> </div> <div class="sect2"> <h3 id="_creating_a_table_with_region_replication"><a class="anchor" href="#_creating_a_table_with_region_replication"></a>71.12. Creating a table with region replication</h3> <div class="paragraph"> <p>Region replication is a per-table property. All tables have <code>REGION_REPLICATION = 1</code> by default, which means that there is only one replica per region. You can set and change the number of replicas per region of a table by supplying the <code>REGION_REPLICATION</code> property in the table descriptor.</p> </div> <div class="sect3"> <h4 id="_shell"><a class="anchor" href="#_shell"></a>71.12.1. Shell</h4> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">create <span class="string"><span class="delimiter">'</span><span class="content">t1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">f1</span><span class="delimiter">'</span></span>, {REGION_REPLICATION => <span class="integer">2</span>} describe <span class="string"><span class="delimiter">'</span><span class="content">t1</span><span class="delimiter">'</span></span> <span class="keyword">for</span> i in <span class="integer">1</span>.<span class="float">.100</span> put <span class="string"><span class="delimiter">'</span><span class="content">t1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">"</span><span class="content">r#{i}</span><span class="delimiter">"</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">f1:c1</span><span class="delimiter">'</span></span>, i end flush <span class="string"><span class="delimiter">'</span><span class="content">t1</span><span class="delimiter">'</span></span></code></pre> </div> </div> </div> <div class="sect3"> <h4 id="_java"><a class="anchor" href="#_java"></a>71.12.2. Java</h4> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HTableDescriptor htd = <span class="keyword">new</span> HTableDescriptor(TableName.valueOf(<span class="error">“</span>test_table<span class="error">”</span>)); htd.setRegionReplication(<span class="integer">2</span>); ... admin.createTable(htd);</code></pre> </div> </div> <div class="paragraph"> <p>You can also use <code>setRegionReplication()</code> and alter table to increase, decrease the region replication for a table.</p> </div> </div> </div> <div class="sect2"> <h3 id="_read_api_and_usage"><a class="anchor" href="#_read_api_and_usage"></a>71.13. Read API and Usage</h3> <div class="sect3"> <h4 id="_shell_2"><a class="anchor" href="#_shell_2"></a>71.13.1. Shell</h4> <div class="paragraph"> <p>You can do reads in shell using a the Consistency.TIMELINE semantics as follows</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">001</span>:<span class="integer">0</span>> get <span class="string"><span class="delimiter">'</span><span class="content">t1</span><span class="delimiter">'</span></span>,<span class="string"><span class="delimiter">'</span><span class="content">r6</span><span class="delimiter">'</span></span>, {CONSISTENCY => <span class="string"><span class="delimiter">"</span><span class="content">TIMELINE</span><span class="delimiter">"</span></span>}</code></pre> </div> </div> <div class="paragraph"> <p>You can simulate a region server pausing or becoming unavailable and do a read from the secondary replica:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ kill -STOP <pid or primary region server> hbase(main):001:0> get 't1','r6', {CONSISTENCY => "TIMELINE"}</code></pre> </div> </div> <div class="paragraph"> <p>Using scans is also similar</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase> scan <span class="string"><span class="delimiter">'</span><span class="content">t1</span><span class="delimiter">'</span></span>, {CONSISTENCY => <span class="string"><span class="delimiter">'</span><span class="content">TIMELINE</span><span class="delimiter">'</span></span>}</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="_java_2"><a class="anchor" href="#_java_2"></a>71.13.2. Java</h4> <div class="paragraph"> <p>You can set set the consistency for Gets and Scans and do requests as follows.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Get get = <span class="keyword">new</span> Get(row); get.setConsistency(Consistency.TIMELINE); ... Result result = table.get(get);</code></pre> </div> </div> <div class="paragraph"> <p>You can also pass multiple gets:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Get get1 = <span class="keyword">new</span> Get(row); get1.setConsistency(Consistency.TIMELINE); ... ArrayList<Get> gets = <span class="keyword">new</span> <span class="predefined-type">ArrayList</span><Get>(); gets.add(get1); ... Result<span class="type">[]</span> results = table.get(gets);</code></pre> </div> </div> <div class="paragraph"> <p>And Scans:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Scan scan = <span class="keyword">new</span> Scan(); scan.setConsistency(Consistency.TIMELINE); ... ResultScanner scanner = table.getScanner(scan);</code></pre> </div> </div> <div class="paragraph"> <p>You can inspect whether the results are coming from primary region or not by calling the <code>Result.isStale()</code> method:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Result</span> result = table.get(get); <span class="keyword">if</span> (result.isStale()) { ... }</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="_resources"><a class="anchor" href="#_resources"></a>71.14. Resources</h3> <div class="olist arabic"> <ol class="arabic"> <li> <p>More information about the design and implementation can be found at the jira issue: <a href="https://issues.apache.org/jira/browse/HBASE-10070">HBASE-10070</a></p> </li> <li> <p>HBaseCon 2014 <a href="http://hbasecon.com/sessions/#session15">talk</a> also contains some details and <a href="http://www.slideshare.net/enissoz/hbase-high-availability-for-reads-with-time">slides</a>.</p> </li> </ol> </div> </div> </div> </div> <h1 id="hbase_apis" class="sect0"><a class="anchor" href="#hbase_apis"></a>Apache HBase APIs</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>This chapter provides information about performing operations using HBase native APIs. This information is not exhaustive, and provides a quick reference in addition to the <a href="http://hbase.apache.org/apidocs/index.html">User API Reference</a>. The examples here are not comprehensive or complete, and should be used for purposes of illustration only.</p> </div> <div class="paragraph"> <p>Apache HBase also works with multiple external APIs. See <a href="#external_apis">Apache HBase External APIs</a> for more information.</p> </div> </div> </div> <div class="sect1"> <h2 id="_examples"><a class="anchor" href="#_examples"></a>72. Examples</h2> <div class="sectionbody"> <div class="exampleblock"> <div class="title">Example 38. Create, modify and delete a Table Using Java</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="keyword">package</span> <span class="namespace">com.example.hbase.admin</span>; <span class="keyword">package</span> <span class="namespace">util</span>; <span class="keyword">import</span> <span class="include">java.io.IOException</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.conf.Configuration</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.fs.Path</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.HBaseConfiguration</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.HColumnDescriptor</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.HConstants</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.HTableDescriptor</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.TableName</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.client.Admin</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.client.Connection</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.client.ConnectionFactory</span>; <span class="keyword">import</span> <span class="include">org.apache.hadoop.hbase.io.compress.Compression.Algorithm</span>; <span class="directive">public</span> <span class="type">class</span> <span class="class">Example</span> { <span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> <span class="predefined-type">String</span> TABLE_NAME = <span class="string"><span class="delimiter">"</span><span class="content">MY_TABLE_NAME_TOO</span><span class="delimiter">"</span></span>; <span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> <span class="predefined-type">String</span> CF_DEFAULT = <span class="string"><span class="delimiter">"</span><span class="content">DEFAULT_COLUMN_FAMILY</span><span class="delimiter">"</span></span>; <span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> createOrOverwrite(Admin admin, HTableDescriptor table) <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="keyword">if</span> (admin.tableExists(table.getTableName())) { admin.disableTable(table.getTableName()); admin.deleteTable(table.getTableName()); } admin.createTable(table); } <span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> createSchemaTables(<span class="predefined-type">Configuration</span> config) <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="keyword">try</span> (<span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(config); Admin admin = connection.getAdmin()) { HTableDescriptor table = <span class="keyword">new</span> HTableDescriptor(TableName.valueOf(TABLE_NAME)); table.addFamily(<span class="keyword">new</span> HColumnDescriptor(CF_DEFAULT).setCompressionType(Algorithm.SNAPPY)); <span class="predefined-type">System</span>.out.print(<span class="string"><span class="delimiter">"</span><span class="content">Creating table. </span><span class="delimiter">"</span></span>); createOrOverwrite(admin, table); <span class="predefined-type">System</span>.out.println(<span class="string"><span class="delimiter">"</span><span class="content"> Done.</span><span class="delimiter">"</span></span>); } } <span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> modifySchema (<span class="predefined-type">Configuration</span> config) <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="keyword">try</span> (<span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(config); Admin admin = connection.getAdmin()) { TableName tableName = TableName.valueOf(TABLE_NAME); <span class="keyword">if</span> (admin.tableExists(tableName)) { <span class="predefined-type">System</span>.out.println(<span class="string"><span class="delimiter">"</span><span class="content">Table does not exist.</span><span class="delimiter">"</span></span>); <span class="predefined-type">System</span>.exit(-<span class="integer">1</span>); } HTableDescriptor table = <span class="keyword">new</span> HTableDescriptor(tableName); <span class="comment">// Update existing table</span> HColumnDescriptor newColumn = <span class="keyword">new</span> HColumnDescriptor(<span class="string"><span class="delimiter">"</span><span class="content">NEWCF</span><span class="delimiter">"</span></span>); newColumn.setCompactionCompressionType(Algorithm.GZ); newColumn.setMaxVersions(HConstants.ALL_VERSIONS); admin.addColumn(tableName, newColumn); <span class="comment">// Update existing column family</span> HColumnDescriptor existingColumn = <span class="keyword">new</span> HColumnDescriptor(CF_DEFAULT); existingColumn.setCompactionCompressionType(Algorithm.GZ); existingColumn.setMaxVersions(HConstants.ALL_VERSIONS); table.modifyFamily(existingColumn); admin.modifyTable(tableName, table); <span class="comment">// Disable an existing table</span> admin.disableTable(tableName); <span class="comment">// Delete an existing column family</span> admin.deleteColumn(tableName, CF_DEFAULT.getBytes(<span class="string"><span class="delimiter">"</span><span class="content">UTF-8</span><span class="delimiter">"</span></span>)); <span class="comment">// Delete a table (Need to be disabled first)</span> admin.deleteTable(tableName); } } <span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> main(<span class="predefined-type">String</span>... args) <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); <span class="comment">//Add any necessary configuration files (hbase-site.xml, core-site.xml)</span> config.addResource(<span class="keyword">new</span> Path(<span class="predefined-type">System</span>.getenv(<span class="string"><span class="delimiter">"</span><span class="content">HBASE_CONF_DIR</span><span class="delimiter">"</span></span>), <span class="string"><span class="delimiter">"</span><span class="content">hbase-site.xml</span><span class="delimiter">"</span></span>)); config.addResource(<span class="keyword">new</span> Path(<span class="predefined-type">System</span>.getenv(<span class="string"><span class="delimiter">"</span><span class="content">HADOOP_CONF_DIR</span><span class="delimiter">"</span></span>), <span class="string"><span class="delimiter">"</span><span class="content">core-site.xml</span><span class="delimiter">"</span></span>)); createSchemaTables(config); modifySchema(config); } }</code></pre> </div> </div> </div> </div> </div> </div> <h1 id="external_apis" class="sect0"><a class="anchor" href="#external_apis"></a>Apache HBase External APIs</h1> <div class="openblock partintro"> <div class="content"> This chapter will cover access to Apache HBase either through non-Java languages, or through custom protocols. For information on using the native HBase APIs, refer to <a href="http://hbase.apache.org/apidocs/index.html">User API Reference</a> and the new <a href="#hbase_apis">HBase APIs</a> chapter. </div> </div> <div class="sect1"> <h2 id="nonjava.jvm"><a class="anchor" href="#nonjava.jvm"></a>73. Non-Java Languages Talking to the JVM</h2> <div class="sectionbody"> <div class="paragraph"> <p>Currently the documentation on this topic is in the <a href="http://wiki.apache.org/hadoop/Hbase">Apache HBase Wiki</a>. See also the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/thrift/package-summary.html#package_description">Thrift API Javadoc</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_rest"><a class="anchor" href="#_rest"></a>74. REST</h2> <div class="sectionbody"> <div class="paragraph"> <p>Currently most of the documentation on REST exists in the <a href="http://wiki.apache.org/hadoop/Hbase/Stargate">Apache HBase Wiki on REST</a> (The REST gateway used to be called 'Stargate'). There are also a nice set of blogs on <a href="http://blog.cloudera.com/blog/2013/03/how-to-use-the-apache-hbase-rest-interface-part-1/">How-to: Use the Apache HBase REST Interface</a> by Jesse Anderson.</p> </div> <div class="paragraph"> <p>To run your REST server under SSL, set <code>hbase.rest.ssl.enabled</code> to <code>true</code> and also set the following configs when you launch the REST server: (See example commands in <a href="#jmx_config">JMX config</a>)</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase.rest.ssl.keystore.store hbase.rest.ssl.keystore.password hbase.rest.ssl.keystore.keypassword</code></pre> </div> </div> <div class="paragraph"> <p>HBase ships a simple REST client, see <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/rest/client/package-summary.html">REST client</a> package for details. To enable SSL support for it, please also import your certificate into local java cacerts keystore:</p> </div> <div class="listingblock"> <div class="content"> <pre>keytool -import -trustcacerts -file /home/user/restserver.cert -keystore $JAVA_HOME/jre/lib/security/cacerts</pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="_thrift"><a class="anchor" href="#_thrift"></a>75. Thrift</h2> <div class="sectionbody"> <div class="paragraph"> <p>Documentation about Thrift has moved to <a href="#thrift">Thrift API and Filter Language</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="c"><a class="anchor" href="#c"></a>76. C/C++ Apache HBase Client</h2> <div class="sectionbody"> <div class="paragraph"> <p>FB’s Chip Turner wrote a pure C/C++ client. <a href="https://github.com/facebook/native-cpp-hbase-client">Check it out</a>.</p> </div> </div> </div> <h1 id="thrift" class="sect0"><a class="anchor" href="#thrift"></a>Thrift API and Filter Language</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>Apache <a href="http://thrift.apache.org/">Thrift</a> is a cross-platform, cross-language development framework. HBase includes a Thrift API and filter language. The Thrift API relies on client and server processes. Documentation about the HBase Thrift API is located at <a href="http://wiki.apache.org/hadoop/Hbase/ThriftApi" class="bare">http://wiki.apache.org/hadoop/Hbase/ThriftApi</a>.</p> </div> <div class="paragraph"> <p>You can configure Thrift for secure authentication at the server and client side, by following the procedures in <a href="#security.client.thrift">Client-side Configuration for Secure Operation - Thrift Gateway</a> and <a href="#security.gateway.thrift">Configure the Thrift Gateway to Authenticate on Behalf of the Client</a>.</p> </div> <div class="paragraph"> <p>The rest of this chapter discusses the filter language provided by the Thrift API.</p> </div> </div> </div> <div class="sect1"> <h2 id="thrift.filter_language"><a class="anchor" href="#thrift.filter_language"></a>77. Filter Language</h2> <div class="sectionbody"> <div class="paragraph"> <p>Thrift Filter Language was introduced in HBase 0.92. It allows you to perform server-side filtering when accessing HBase over Thrift or in the HBase shell. You can find out more about shell integration by using the <code>scan help</code> command in the shell.</p> </div> <div class="paragraph"> <p>You specify a filter as a string, which is parsed on the server to construct the filter.</p> </div> <div class="sect2"> <h3 id="general_syntax"><a class="anchor" href="#general_syntax"></a>77.1. General Filter String Syntax</h3> <div class="paragraph"> <p>A simple filter expression is expressed as a string:</p> </div> <div class="listingblock"> <div class="content"> <pre>“FilterName (argument, argument,... , argument)”</pre> </div> </div> <div class="paragraph"> <p>Keep the following syntax guidelines in mind.</p> </div> <div class="ulist"> <ul> <li> <p>Specify the name of the filter followed by the comma-separated argument list in parentheses.</p> </li> <li> <p>If the argument represents a string, it should be enclosed in single quotes (<code>'</code>).</p> </li> <li> <p>Arguments which represent a boolean, an integer, or a comparison operator (such as <, >, or !=), should not be enclosed in quotes</p> </li> <li> <p>The filter name must be a single word. All ASCII characters are allowed except for whitespace, single quotes and parentheses.</p> </li> <li> <p>The filter’s arguments can contain any ASCII character. If single quotes are present in the argument, they must be escaped by an additional preceding single quote.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="_compound_filters_and_operators"><a class="anchor" href="#_compound_filters_and_operators"></a>77.2. Compound Filters and Operators</h3> <div class="dlist"> <div class="title">Binary Operators</div> <dl> <dt class="hdlist1"><code>AND</code></dt> <dd> <p>If the <code>AND</code> operator is used, the key-value must satisfy both filters.</p> </dd> <dt class="hdlist1"><code>OR</code></dt> <dd> <p>If the <code>OR</code> operator is used, the key-value must satisfy at least one of the filters.</p> </dd> </dl> </div> <div class="dlist"> <div class="title">Unary Operators</div> <dl> <dt class="hdlist1"><code>SKIP</code></dt> <dd> <p>For a particular row, if any of the key-values fail the filter condition, the entire row is skipped.</p> </dd> <dt class="hdlist1"><code>WHILE</code></dt> <dd> <p>For a particular row, key-values will be emitted until a key-value is reached that fails the filter condition.</p> </dd> </dl> </div> <div class="exampleblock"> <div class="title">Example 39. Compound Operators</div> <div class="content"> <div class="paragraph"> <p>You can combine multiple operators to create a hierarchy of filters, such as the following example:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">(Filter1 AND Filter2) OR (Filter3 AND Filter4)</code></pre> </div> </div> </div> </div> </div> <div class="sect2"> <h3 id="_order_of_evaluation"><a class="anchor" href="#_order_of_evaluation"></a>77.3. Order of Evaluation</h3> <div class="olist arabic"> <ol class="arabic"> <li> <p>Parentheses have the highest precedence.</p> </li> <li> <p>The unary operators <code>SKIP</code> and <code>WHILE</code> are next, and have the same precedence.</p> </li> <li> <p>The binary operators follow. <code>AND</code> has highest precedence, followed by <code>OR</code>.</p> </li> </ol> </div> <div class="exampleblock"> <div class="title">Example 40. Precedence Example</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Filter1 AND Filter2 OR <span class="predefined-type">Filter</span> is evaluated as (Filter1 AND Filter2) OR Filter3</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Filter1 AND SKIP Filter2 OR Filter3 is evaluated as (Filter1 AND (SKIP Filter2)) OR Filter3</code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>You can use parentheses to explicitly control the order of evaluation.</p> </div> </div> <div class="sect2"> <h3 id="_compare_operator"><a class="anchor" href="#_compare_operator"></a>77.4. Compare Operator</h3> <div class="paragraph"> <p>The following compare operators are provided:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>LESS (<)</p> </li> <li> <p>LESS_OR_EQUAL (⇐)</p> </li> <li> <p>EQUAL (=)</p> </li> <li> <p>NOT_EQUAL (!=)</p> </li> <li> <p>GREATER_OR_EQUAL (>=)</p> </li> <li> <p>GREATER (>)</p> </li> <li> <p>NO_OP (no operation)</p> </li> </ol> </div> <div class="paragraph"> <p>The client should use the symbols (<, ⇐, =, !=, >, >=) to express compare operators.</p> </div> </div> <div class="sect2"> <h3 id="_comparator"><a class="anchor" href="#_comparator"></a>77.5. Comparator</h3> <div class="paragraph"> <p>A comparator can be any of the following:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p><em>BinaryComparator</em> - This lexicographically compares against the specified byte array using Bytes.compareTo(byte[], byte[])</p> </li> <li> <p><em>BinaryPrefixComparator</em> - This lexicographically compares against a specified byte array. It only compares up to the length of this byte array.</p> </li> <li> <p><em>RegexStringComparator</em> - This compares against the specified byte array using the given regular expression. Only EQUAL and NOT_EQUAL comparisons are valid with this comparator</p> </li> <li> <p><em>SubStringComparator</em> - This tests if the given substring appears in a specified byte array. The comparison is case insensitive. Only EQUAL and NOT_EQUAL comparisons are valid with this comparator</p> </li> </ol> </div> <div class="paragraph"> <p>The general syntax of a comparator is: <code>ComparatorType:ComparatorValue</code></p> </div> <div class="paragraph"> <p>The ComparatorType for the various comparators is as follows:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p><em>BinaryComparator</em> - binary</p> </li> <li> <p><em>BinaryPrefixComparator</em> - binaryprefix</p> </li> <li> <p><em>RegexStringComparator</em> - regexstring</p> </li> <li> <p><em>SubStringComparator</em> - substring</p> </li> </ol> </div> <div class="paragraph"> <p>The ComparatorValue can be any value.</p> </div> <div class="olist arabic"> <div class="title">Example ComparatorValues</div> <ol class="arabic"> <li> <p><code>binary:abc</code> will match everything that is lexicographically greater than "abc"</p> </li> <li> <p><code>binaryprefix:abc</code> will match everything whose first 3 characters are lexicographically equal to "abc"</p> </li> <li> <p><code>regexstring:ab*yz</code> will match everything that doesn’t begin with "ab" and ends with "yz"</p> </li> <li> <p><code>substring:abc123</code> will match everything that begins with the substring "abc123"</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="examplephpclientprogram"><a class="anchor" href="#examplephpclientprogram"></a>77.6. Example PHP Client Program that uses the Filter Language</h3> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="php"><span class="inline-delimiter"><?</span> <span class="predefined">$_SERVER</span>[<span class="string"><span class="delimiter">'</span><span class="content">PHP_ROOT</span><span class="delimiter">'</span></span>] = <span class="predefined">realpath</span>(<span class="predefined">dirname</span>(<span class="predefined-constant">__FILE__</span>).<span class="string"><span class="delimiter">'</span><span class="content">/..</span><span class="delimiter">'</span></span>); <span class="predefined">require_once</span> <span class="predefined">$_SERVER</span>[<span class="string"><span class="delimiter">'</span><span class="content">PHP_ROOT</span><span class="delimiter">'</span></span>].<span class="string"><span class="delimiter">'</span><span class="content">/flib/__flib.php</span><span class="delimiter">'</span></span>; flib_init(<span class="constant">FLIB_CONTEXT_SCRIPT</span>); require_module(<span class="string"><span class="delimiter">'</span><span class="content">storage/hbase</span><span class="delimiter">'</span></span>); <span class="local-variable">$hbase</span> = <span class="keyword">new</span> <span class="constant">HBase</span>(<span class="string"><span class="delimiter">'</span><span class="content"><server_name_running_thrift_server></span><span class="delimiter">'</span></span>, <port on which thrift server is running>); <span class="local-variable">$hbase</span>->open(); <span class="local-variable">$client</span> = <span class="local-variable">$hbase</span>->getClient(); <span class="local-variable">$result</span> = <span class="local-variable">$client</span>->scannerOpenWithFilterString(<span class="string"><span class="delimiter">'</span><span class="content">table_name</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">"</span><span class="content">(PrefixFilter ('row2') AND (QualifierFilter (>=, 'binary:xyz'))) AND (TimestampsFilter ( 123, 456))</span><span class="delimiter">"</span></span>); <span class="local-variable">$to_print</span> = <span class="local-variable">$client</span>->scannerGetList(<span class="local-variable">$result</span>,<span class="integer">1</span>); <span class="keyword">while</span> (<span class="local-variable">$to_print</span>) { <span class="predefined">print_r</span>(<span class="local-variable">$to_print</span>); <span class="local-variable">$to_print</span> = <span class="local-variable">$client</span>->scannerGetList(<span class="local-variable">$result</span>,<span class="integer">1</span>); } <span class="local-variable">$client</span>->scannerClose(<span class="local-variable">$result</span>); <span class="inline-delimiter">?></span></code></pre> </div> </div> </div> <div class="sect2"> <h3 id="_example_filter_strings"><a class="anchor" href="#_example_filter_strings"></a>77.7. Example Filter Strings</h3> <div class="ulist"> <ul> <li> <p><code>"PrefixFilter ('Row') AND PageFilter (1) AND FirstKeyOnlyFilter ()"</code> will return all key-value pairs that match the following conditions:</p> <div class="olist arabic"> <ol class="arabic"> <li> <p>The row containing the key-value should have prefix <em>Row</em></p> </li> <li> <p>The key-value must be located in the first row of the table</p> </li> <li> <p>The key-value pair must be the first key-value in the row</p> </li> </ol> </div> </li> <li> <p><code>"(RowFilter (=, 'binary:Row 1') AND TimeStampsFilter (74689, 89734)) OR ColumnRangeFilter ('abc', true, 'xyz', false))"</code> will return all key-value pairs that match both the following conditions:</p> <div class="ulist"> <ul> <li> <p>The key-value is in a row having row key <em>Row 1</em></p> </li> <li> <p>The key-value must have a timestamp of either 74689 or 89734.</p> </li> <li> <p>Or it must match the following condition:</p> <div class="ulist"> <ul> <li> <p>The key-value pair must be in a column that is lexicographically >= abc and < xyz </p> </li> </ul> </div> </li> </ul> </div> </li> <li> <p><code>"SKIP ValueFilter (0)"</code> will skip the entire row if any of the values in the row is not 0</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="individualfiltersyntax"><a class="anchor" href="#individualfiltersyntax"></a>77.8. Individual Filter Syntax</h3> <div class="dlist"> <dl> <dt class="hdlist1">KeyOnlyFilter</dt> <dd> <p>This filter doesn’t take any arguments. It returns only the key component of each key-value.</p> </dd> <dt class="hdlist1">FirstKeyOnlyFilter</dt> <dd> <p>This filter doesn’t take any arguments. It returns only the first key-value from each row.</p> </dd> <dt class="hdlist1">PrefixFilter</dt> <dd> <p>This filter takes one argument – a prefix of a row key. It returns only those key-values present in a row that starts with the specified row prefix</p> </dd> <dt class="hdlist1">ColumnPrefixFilter</dt> <dd> <p>This filter takes one argument – a column prefix. It returns only those key-values present in a column that starts with the specified column prefix. The column prefix must be of the form: <code>“qualifier”</code>.</p> </dd> <dt class="hdlist1">MultipleColumnPrefixFilter</dt> <dd> <p>This filter takes a list of column prefixes. It returns key-values that are present in a column that starts with any of the specified column prefixes. Each of the column prefixes must be of the form: <code>“qualifier”</code>.</p> </dd> <dt class="hdlist1">ColumnCountGetFilter</dt> <dd> <p>This filter takes one argument – a limit. It returns the first limit number of columns in the table.</p> </dd> <dt class="hdlist1">PageFilter</dt> <dd> <p>This filter takes one argument – a page size. It returns page size number of rows from the table.</p> </dd> <dt class="hdlist1">ColumnPaginationFilter</dt> <dd> <p>This filter takes two arguments – a limit and offset. It returns limit number of columns after offset number of columns. It does this for all the rows.</p> </dd> <dt class="hdlist1">InclusiveStopFilter</dt> <dd> <p>This filter takes one argument – a row key on which to stop scanning. It returns all key-values present in rows up to and including the specified row.</p> </dd> <dt class="hdlist1">TimeStampsFilter</dt> <dd> <p>This filter takes a list of timestamps. It returns those key-values whose timestamps matches any of the specified timestamps.</p> </dd> <dt class="hdlist1">RowFilter</dt> <dd> <p>This filter takes a compare operator and a comparator. It compares each row key with the comparator using the compare operator and if the comparison returns true, it returns all the key-values in that row.</p> </dd> <dt class="hdlist1">Family Filter</dt> <dd> <p>This filter takes a compare operator and a comparator. It compares each qualifier name with the comparator using the compare operator and if the comparison returns true, it returns all the key-values in that column.</p> </dd> <dt class="hdlist1">QualifierFilter</dt> <dd> <p>This filter takes a compare operator and a comparator. It compares each qualifier name with the comparator using the compare operator and if the comparison returns true, it returns all the key-values in that column.</p> </dd> <dt class="hdlist1">ValueFilter</dt> <dd> <p>This filter takes a compare operator and a comparator. It compares each value with the comparator using the compare operator and if the comparison returns true, it returns that key-value.</p> </dd> <dt class="hdlist1">DependentColumnFilter</dt> <dd> <p>This filter takes two arguments – a family and a qualifier. It tries to locate this column in each row and returns all key-values in that row that have the same timestamp. If the row doesn’t contain the specified column – none of the key-values in that row will be returned.</p> </dd> <dt class="hdlist1">SingleColumnValueFilter</dt> <dd> <p>This filter takes a column family, a qualifier, a compare operator and a comparator. If the specified column is not found – all the columns of that row will be emitted. If the column is found and the comparison with the comparator returns true, all the columns of the row will be emitted. If the condition fails, the row will not be emitted.</p> </dd> <dt class="hdlist1">SingleColumnValueExcludeFilter</dt> <dd> <p>This filter takes the same arguments and behaves same as SingleColumnValueFilter – however, if the column is found and the condition passes, all the columns of the row will be emitted except for the tested column value.</p> </dd> <dt class="hdlist1">ColumnRangeFilter</dt> <dd> <p>This filter is used for selecting only those keys with columns that are between minColumn and maxColumn. It also takes two boolean variables to indicate whether to include the minColumn and maxColumn or not.</p> </dd> </dl> </div> </div> </div> </div> <h1 id="cp" class="sect0"><a class="anchor" href="#cp"></a>Apache HBase Coprocessors</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>HBase coprocessors are modeled after the coprocessors which are part of Google’s BigTable (<a href="http://www.scribd.com/doc/21631448/Dean-Keynote-Ladis2009" class="bare">http://www.scribd.com/doc/21631448/Dean-Keynote-Ladis2009</a>, pages 66-67.). Coprocessors function in a similar way to Linux kernel modules. They provide a way to run server-level code against locally-stored data. The functionality they provide is very powerful, but also carries great risk and can have adverse effects on the system, at the level of the operating system. The information in this chapter is primarily sourced and heavily reused from Mingjie Lai’s blog post at <a href="https://blogs.apache.org/hbase/entry/coprocessor_introduction" class="bare">https://blogs.apache.org/hbase/entry/coprocessor_introduction</a>.</p> </div> <div class="paragraph"> <p>Coprocessors are not designed to be used by end users of HBase, but by HBase developers who need to add specialized functionality to HBase. One example of the use of coprocessors is pluggable compaction and scan policies, which are provided as coprocessors in <a href="https://issues.apache.org/jira/browse/HBASE-6427">HBASE-6427</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_coprocessor_framework"><a class="anchor" href="#_coprocessor_framework"></a>78. Coprocessor Framework</h2> <div class="sectionbody"> <div class="paragraph"> <p>The implementation of HBase coprocessors diverges from the BigTable implementation. The HBase framework provides a library and runtime environment for executing user code within the HBase region server and master processes.</p> </div> <div class="paragraph"> <p>The framework API is provided in the <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/coprocessor/package-summary.html">coprocessor</a> package.</p> </div> <div class="paragraph"> <p>Two different types of coprocessors are provided by the framework, based on their scope.</p> </div> <div class="dlist"> <div class="title">Types of Coprocessors</div> <dl> <dt class="hdlist1">System Coprocessors</dt> <dd> <p>System coprocessors are loaded globally on all tables and regions hosted by a region server.</p> </dd> <dt class="hdlist1">Table Coprocessors</dt> <dd> <p>You can specify which coprocessors should be loaded on all regions for a table on a per-table basis.</p> </dd> </dl> </div> <div class="paragraph"> <p>The framework provides two different aspects of extensions as well: <em>observers</em> and <em>endpoints</em>.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Observers</dt> <dd> <p>Observers are analogous to triggers in conventional databases. They allow you to insert user code by overriding upcall methods provided by the coprocessor framework. Callback functions are executed from core HBase code when events occur. Callbacks are handled by the framework, and the coprocessor itself only needs to insert the extended or alternate functionality.</p> </dd> <dt class="hdlist1">Endpoints (HBase 0.96.x and later)</dt> <dd> <p>The implementation for endpoints changed significantly in HBase 0.96.x due to the introduction of protocol buffers (protobufs) (<a href="https://issues.apache.org/jira/browse/HBASE-5448">HBASE-5488</a>). If you created endpoints before 0.96.x, you will need to rewrite them. Endpoints are now defined and callable as protobuf services, rather than endpoint invocations passed through as Writable blobs</p> </dd> <dt class="hdlist1">Endpoints (HBase 0.94.x and earlier)</dt> <dd> <p>Dynamic RPC endpoints resemble stored procedures. An endpoint can be invoked at any time from the client. When it is invoked, it is executed remotely at the target region or regions, and results of the executions are returned to the client.</p> </dd> </dl> </div> </div> </div> <div class="sect1"> <h2 id="_examples_2"><a class="anchor" href="#_examples_2"></a>79. Examples</h2> <div class="sectionbody"> <div class="paragraph"> <p>An example of an observer is included in <em>hbase-examples/src/test/java/org/apache/hadoop/hbase/coprocessor/example/TestZooKeeperScanPolicyObserver.java</em>. Several endpoint examples are included in the same directory.</p> </div> </div> </div> <div class="sect1"> <h2 id="_building_a_coprocessor"><a class="anchor" href="#_building_a_coprocessor"></a>80. Building A Coprocessor</h2> <div class="sectionbody"> <div class="paragraph"> <p>Before you can build a processor, it must be developed, compiled, and packaged in a JAR file. The next step is to configure the coprocessor framework to use your coprocessor. You can load the coprocessor from your HBase configuration, so that the coprocessor starts with HBase, or you can configure the coprocessor from the HBase shell, as a table attribute, so that it is loaded dynamically when the table is opened or reopened.</p> </div> <div class="sect2"> <h3 id="_load_from_configuration"><a class="anchor" href="#_load_from_configuration"></a>80.1. Load from Configuration</h3> <div class="paragraph"> <p>To configure a coprocessor to be loaded when HBase starts, modify the RegionServer’s <em>hbase-site.xml</em> and configure one of the following properties, based on the type of observer you are configuring:</p> </div> <div class="ulist"> <ul> <li> <p>`hbase.coprocessor.region.classes`for RegionObservers and Endpoints</p> </li> <li> <p>`hbase.coprocessor.wal.classes`for WALObservers</p> </li> <li> <p>`hbase.coprocessor.master.classes`for MasterObservers</p> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 41. Example RegionObserver Configuration</div> <div class="content"> <div class="paragraph"> <p>In this example, one RegionObserver is configured for all the HBase tables.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.coprocessor.AggregateImplementation<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>If multiple classes are specified for loading, the class names must be comma-separated. The framework attempts to load all the configured classes using the default class loader. Therefore, the jar file must reside on the server-side HBase classpath.</p> </div> <div class="paragraph"> <p>Coprocessors which are loaded in this way will be active on all regions of all tables. These are the system coprocessor introduced earlier. The first listed coprocessors will be assigned the priority <code>Coprocessor.Priority.SYSTEM</code>. Each subsequent coprocessor in the list will have its priority value incremented by one (which reduces its priority, because priorities have the natural sort order of Integers).</p> </div> <div class="paragraph"> <p>When calling out to registered observers, the framework executes their callbacks methods in the sorted order of their priority. Ties are broken arbitrarily.</p> </div> </div> <div class="sect2"> <h3 id="_load_from_the_hbase_shell"><a class="anchor" href="#_load_from_the_hbase_shell"></a>80.2. Load from the HBase Shell</h3> <div class="paragraph"> <p>You can load a coprocessor on a specific table via a table attribute. The following example will load the <code>FooRegionObserver</code> observer when table <code>t1</code> is read or re-read.</p> </div> <div class="exampleblock"> <div class="title">Example 42. Load a Coprocessor On a Table Using HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase(main):005:0> alter 't1', METHOD => 'table_att', 'coprocessor'=>'hdfs:///foo.jar|com.foo.FooRegionObserver|1001|arg1=1,arg2=2' Updating all regions with the new schema... 1/1 regions updated. Done. 0 row(s) in 1.0730 seconds hbase(main):006:0> describe 't1' DESCRIPTION ENABLED {NAME => 't1', coprocessor$1 => 'hdfs:///foo.jar|com.foo.FooRegio false nObserver|1001|arg1=1,arg2=2', FAMILIES => [{NAME => 'c1', DATA_B LOCK_ENCODING => 'NONE', BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSIONS => '3', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => '2147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZ E => '65536', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLO CKCACHE => 'true'}, {NAME => 'f1', DATA_BLOCK_ENCODING => 'NONE', BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSIONS => '3' , COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => '2147483647' , KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE => 'true'}]} 1 row(s) in 0.0190 seconds</pre> </div> </div> </div> </div> <div class="paragraph"> <p>The coprocessor framework will try to read the class information from the coprocessor table attribute value. The value contains four pieces of information which are separated by the <code>|</code> character.</p> </div> <div class="ulist"> <ul> <li> <p>File path: The jar file containing the coprocessor implementation must be in a location where all region servers can read it. You could copy the file onto the local disk on each region server, but it is recommended to store it in HDFS.</p> </li> <li> <p>Class name: The full class name of the coprocessor.</p> </li> <li> <p>Priority: An integer. The framework will determine the execution sequence of all configured observers registered at the same hook using priorities. This field can be left blank. In that case the framework will assign a default priority value.</p> </li> <li> <p>Arguments: This field is passed to the coprocessor implementation.</p> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 43. Unload a Coprocessor From a Table Using HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase(main):007:0> alter 't1', METHOD => 'table_att_unset', hbase(main):008:0* NAME => 'coprocessor$1' Updating all regions with the new schema... 1/1 regions updated. Done. 0 row(s) in 1.1130 seconds hbase(main):009:0> describe 't1' DESCRIPTION ENABLED {NAME => 't1', FAMILIES => [{NAME => 'c1', DATA_BLOCK_ENCODING => false 'NONE', BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSION S => '3', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => '214 7483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN _MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE => 'true '}, {NAME => 'f1', DATA_BLOCK_ENCODING => 'NONE', BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSIONS => '3', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => '2147483647', KEEP_DELETED_C ELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false', ENCO DE_ON_DISK => 'true', BLOCKCACHE => 'true'}]} 1 row(s) in 0.0180 seconds</pre> </div> </div> </div> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> There is no guarantee that the framework will load a given coprocessor successfully. For example, the shell command neither guarantees a jar file exists at a particular location nor verifies whether the given class is actually contained in the jar file. </td> </tr> </table> </div> </div> </div> </div> <div class="sect1"> <h2 id="_check_the_status_of_a_coprocessor"><a class="anchor" href="#_check_the_status_of_a_coprocessor"></a>81. Check the Status of a Coprocessor</h2> <div class="sectionbody"> <div class="paragraph"> <p>To check the status of a coprocessor after it has been configured, use the <code>status</code> HBase Shell command.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):020:0> status 'detailed' version 0.92-tm-6 0 regionsInTransition master coprocessors: [] 1 live servers localhost:52761 1328082515520 requestsPerSecond=3, numberOfOnlineRegions=3, usedHeapMB=32, maxHeapMB=995 -ROOT-,,0 numberOfStores=1, numberOfStorefiles=1, storefileUncompressedSizeMB=0, storefileSizeMB=0, memstoreSizeMB=0, storefileIndexSizeMB=0, readRequestsCount=54, writeRequestsCount=1, rootIndexSizeKB=0, totalStaticIndexSizeKB=0, totalStaticBloomSizeKB=0, totalCompactingKVs=0, currentCompactedKVs=0, compactionProgressPct=NaN, coprocessors=[] .META.,,1 numberOfStores=1, numberOfStorefiles=0, storefileUncompressedSizeMB=0, storefileSizeMB=0, memstoreSizeMB=0, storefileIndexSizeMB=0, readRequestsCount=97, writeRequestsCount=4, rootIndexSizeKB=0, totalStaticIndexSizeKB=0, totalStaticBloomSizeKB=0, totalCompactingKVs=0, currentCompactedKVs=0, compactionProgressPct=NaN, coprocessors=[] t1,,1328082575190.c0491168a27620ffe653ec6c04c9b4d1. numberOfStores=2, numberOfStorefiles=1, storefileUncompressedSizeMB=0, storefileSizeMB=0, memstoreSizeMB=0, storefileIndexSizeMB=0, readRequestsCount=0, writeRequestsCount=0, rootIndexSizeKB=0, totalStaticIndexSizeKB=0, totalStaticBloomSizeKB=0, totalCompactingKVs=0, currentCompactedKVs=0, compactionProgressPct=NaN, coprocessors=[AggregateImplementation] 0 dead servers</pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="_monitor_time_spent_in_coprocessors"><a class="anchor" href="#_monitor_time_spent_in_coprocessors"></a>82. Monitor Time Spent in Coprocessors</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase 0.98.5 introduced the ability to monitor some statistics relating to the amount of time spent executing a given coprocessor. You can see these statistics via the HBase Metrics framework (see <a href="#hbase_metrics">[hbase_metrics]</a> or the Web UI for a given Region Server, via the <em>Coprocessor Metrics</em> tab. These statistics are valuable for debugging and benchmarking the performance impact of a given coprocessor on your cluster. Tracked statistics include min, max, average, and 90th, 95th, and 99th percentile. All times are shown in milliseconds. The statistics are calculated over coprocessor execution samples recorded during the reporting interval, which is 10 seconds by default. The metrics sampling rate as described in <a href="#hbase_metrics">[hbase_metrics]</a>.</p> </div> <div class="imageblock"> <div class="content"> <img src="images/coprocessor_stats.png" alt="coprocessor stats"> </div> <div class="title">Figure 4. Coprocessor Metrics UI</div> </div> </div> </div> <h1 id="performance" class="sect0"><a class="anchor" href="#performance"></a>Apache HBase Performance Tuning</h1> <div class="sect1"> <h2 id="perf.os"><a class="anchor" href="#perf.os"></a>83. Operating System</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="perf.os.ram"><a class="anchor" href="#perf.os.ram"></a>83.1. Memory</h3> <div class="paragraph"> <p>RAM, RAM, RAM. Don’t starve HBase.</p> </div> </div> <div class="sect2"> <h3 id="perf.os.64"><a class="anchor" href="#perf.os.64"></a>83.2. 64-bit</h3> <div class="paragraph"> <p>Use a 64-bit platform (and 64-bit JVM).</p> </div> </div> <div class="sect2"> <h3 id="perf.os.swap"><a class="anchor" href="#perf.os.swap"></a>83.3. Swapping</h3> <div class="paragraph"> <p>Watch out for swapping. Set <code>swappiness</code> to 0.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.network"><a class="anchor" href="#perf.network"></a>84. Network</h2> <div class="sectionbody"> <div class="paragraph"> <p>Perhaps the most important factor in avoiding network issues degrading Hadoop and HBase performance is the switching hardware that is used, decisions made early in the scope of the project can cause major problems when you double or triple the size of your cluster (or more).</p> </div> <div class="paragraph"> <p>Important items to consider:</p> </div> <div class="ulist"> <ul> <li> <p>Switching capacity of the device</p> </li> <li> <p>Number of systems connected</p> </li> <li> <p>Uplink capacity</p> </li> </ul> </div> <div class="sect2"> <h3 id="perf.network.1switch"><a class="anchor" href="#perf.network.1switch"></a>84.1. Single Switch</h3> <div class="paragraph"> <p>The single most important factor in this configuration is that the switching capacity of the hardware is capable of handling the traffic which can be generated by all systems connected to the switch. Some lower priced commodity hardware can have a slower switching capacity than could be utilized by a full switch.</p> </div> </div> <div class="sect2"> <h3 id="perf.network.2switch"><a class="anchor" href="#perf.network.2switch"></a>84.2. Multiple Switches</h3> <div class="paragraph"> <p>Multiple switches are a potential pitfall in the architecture. The most common configuration of lower priced hardware is a simple 1Gbps uplink from one switch to another. This often overlooked pinch point can easily become a bottleneck for cluster communication. Especially with MapReduce jobs that are both reading and writing a lot of data the communication across this uplink could be saturated.</p> </div> <div class="paragraph"> <p>Mitigation of this issue is fairly simple and can be accomplished in multiple ways:</p> </div> <div class="ulist"> <ul> <li> <p>Use appropriate hardware for the scale of the cluster which you’re attempting to build.</p> </li> <li> <p>Use larger single switch configurations i.e. single 48 port as opposed to 2x 24 port</p> </li> <li> <p>Configure port trunking for uplinks to utilize multiple interfaces to increase cross switch bandwidth.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="perf.network.multirack"><a class="anchor" href="#perf.network.multirack"></a>84.3. Multiple Racks</h3> <div class="paragraph"> <p>Multiple rack configurations carry the same potential issues as multiple switches, and can suffer performance degradation from two main areas:</p> </div> <div class="ulist"> <ul> <li> <p>Poor switch capacity performance</p> </li> <li> <p>Insufficient uplink to another rack</p> </li> </ul> </div> <div class="paragraph"> <p>If the the switches in your rack have appropriate switching capacity to handle all the hosts at full speed, the next most likely issue will be caused by homing more of your cluster across racks. The easiest way to avoid issues when spanning multiple racks is to use port trunking to create a bonded uplink to other racks. The downside of this method however, is in the overhead of ports that could potentially be used. An example of this is, creating an 8Gbps port channel from rack A to rack B, using 8 of your 24 ports to communicate between racks gives you a poor ROI, using too few however can mean you’re not getting the most out of your cluster.</p> </div> <div class="paragraph"> <p>Using 10Gbe links between racks will greatly increase performance, and assuming your switches support a 10Gbe uplink or allow for an expansion card will allow you to save your ports for machines as opposed to uplinks.</p> </div> </div> <div class="sect2"> <h3 id="perf.network.ints"><a class="anchor" href="#perf.network.ints"></a>84.4. Network Interfaces</h3> <div class="paragraph"> <p>Are all the network interfaces functioning correctly? Are you sure? See the Troubleshooting Case Study in <a href="#casestudies.slownode">Case Study #1 (Performance Issue On A Single Node)</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.network.call_me_maybe"><a class="anchor" href="#perf.network.call_me_maybe"></a>84.5. Network Consistency and Partition Tolerance</h3> <div class="paragraph"> <p>The <a href="http://en.wikipedia.org/wiki/CAP_theorem">CAP Theorem</a> states that a distributed system can maintain two out of the following three charateristics: - *C*onsistency — all nodes see the same data. - *A*vailability — every request receives a response about whether it succeeded or failed. - *P*artition tolerance — the system continues to operate even if some of its components become unavailable to the others.</p> </div> <div class="paragraph"> <p>HBase favors consistency and partition tolerance, where a decision has to be made. Coda Hale explains why partition tolerance is so important, in <a href="http://codahale.com/you-cant-sacrifice-partition-tolerance/" class="bare">http://codahale.com/you-cant-sacrifice-partition-tolerance/</a>.</p> </div> <div class="paragraph"> <p>Robert Yokota used an automated testing framework called <a href="https://aphyr.com/tags/jepsen">Jepson</a> to test HBase’s partition tolerance in the face of network partitions, using techniques modeled after Aphyr’s <a href="https://aphyr.com/posts/281-call-me-maybe-carly-rae-jepsen-and-the-perils-of-network-partitions">Call Me Maybe</a> series. The results, available as a <a href="http://old.eng.yammer.com/call-me-maybe-hbase/">blog post</a> and an <a href="http://old.eng.yammer.com/call-me-maybe-hbase-addendum/">addendum</a>, show that HBase performs correctly.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="jvm"><a class="anchor" href="#jvm"></a>85. Java</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="gc"><a class="anchor" href="#gc"></a>85.1. The Garbage Collector and Apache HBase</h3> <div class="sect3"> <h4 id="gcpause"><a class="anchor" href="#gcpause"></a>85.1.1. Long GC pauses</h4> <div class="paragraph"> <p>In his presentation, <a href="http://www.slideshare.net/cloudera/hbase-hug-presentation">Avoiding Full GCs with MemStore-Local Allocation Buffers</a>, Todd Lipcon describes two cases of stop-the-world garbage collections common in HBase, especially during loading; CMS failure modes and old generation heap fragmentation brought.</p> </div> <div class="paragraph"> <p>To address the first, start the CMS earlier than default by adding <code>-XX:CMSInitiatingOccupancyFraction</code> and setting it down from defaults. Start at 60 or 70 percent (The lower you bring down the threshold, the more GCing is done, the more CPU used). To address the second fragmentation issue, Todd added an experimental facility, (MSLAB), that must be explicitly enabled in Apache HBase 0.90.x (It’s defaulted to be <em>on</em> in Apache 0.92.x HBase). Set <code>hbase.hregion.memstore.mslab.enabled</code> to true in your <code>Configuration</code>. See the cited slides for background and detail. The latest JVMs do better regards fragmentation so make sure you are running a recent release. Read down in the message, <a href="http://osdir.com/ml/hotspot-gc-use/2011-11/msg00002.html">Identifying concurrent mode failures caused by fragmentation</a>. Be aware that when enabled, each MemStore instance will occupy at least an MSLAB instance of memory. If you have thousands of regions or lots of regions each with many column families, this allocation of MSLAB may be responsible for a good portion of your heap allocation and in an extreme case cause you to OOME. Disable MSLAB in this case, or lower the amount of memory it uses or float less regions per server.</p> </div> <div class="paragraph"> <p>If you have a write-heavy workload, check out <a href="https://issues.apache.org/jira/browse/HBASE-8163">HBASE-8163 MemStoreChunkPool: An improvement for JAVA GC when using MSLAB</a>. It describes configurations to lower the amount of young GC during write-heavy loadings. If you do not have HBASE-8163 installed, and you are trying to improve your young GC times, one trick to consider — courtesy of our Liang Xie — is to set the GC config <code>-XX:PretenureSizeThreshold</code> in <em>hbase-env.sh</em> to be just smaller than the size of <code>hbase.hregion.memstore.mslab.chunksize</code> so MSLAB allocations happen in the tenured space directly rather than first in the young gen. You’d do this because these MSLAB allocations are going to likely make it to the old gen anyways and rather than pay the price of a copies between s0 and s1 in eden space followed by the copy up from young to old gen after the MSLABs have achieved sufficient tenure, save a bit of YGC churn and allocate in the old gen directly.</p> </div> <div class="paragraph"> <p>For more information about GC logs, see <a href="#trouble.log.gc">JVM Garbage Collection Logs</a>.</p> </div> <div class="paragraph"> <p>Consider also enabling the off-heap Block Cache. This has been shown to mitigate GC pause times. See <a href="#block.cache">Block Cache</a></p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.configurations"><a class="anchor" href="#perf.configurations"></a>86. HBase Configurations</h2> <div class="sectionbody"> <div class="paragraph"> <p>See <a href="#recommended_configurations">[recommended_configurations]</a>.</p> </div> <div class="sect2"> <h3 id="perf.compactions.and.splits"><a class="anchor" href="#perf.compactions.and.splits"></a>86.1. Managing Compactions</h3> <div class="paragraph"> <p>For larger systems, managing link:[compactions and splits] may be something you want to consider.</p> </div> </div> <div class="sect2"> <h3 id="perf.handlers"><a class="anchor" href="#perf.handlers"></a>86.2. <code>hbase.regionserver.handler.count</code></h3> <div class="paragraph"> <p>See <a href="#hbase.regionserver.handler.count">[hbase.regionserver.handler.count]</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.hfile.block.cache.size"><a class="anchor" href="#perf.hfile.block.cache.size"></a>86.3. <code>hfile.block.cache.size</code></h3> <div class="paragraph"> <p>See <a href="#hfile.block.cache.size">[hfile.block.cache.size]</a>. A memory setting for the RegionServer process.</p> </div> </div> <div class="sect2"> <h3 id="blockcache.prefetch"><a class="anchor" href="#blockcache.prefetch"></a>86.4. Prefetch Option for Blockcache</h3> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-9857">HBASE-9857</a> adds a new option to prefetch HFile contents when opening the BlockCache, if a Column family or RegionServer property is set. This option is available for HBase 0.98.3 and later. The purpose is to warm the BlockCache as rapidly as possible after the cache is opened, using in-memory table data, and not counting the prefetching as cache misses. This is great for fast reads, but is not a good idea if the data to be preloaded will not fit into the BlockCache. It is useful for tuning the IO impact of prefetching versus the time before all data blocks are in cache.</p> </div> <div class="paragraph"> <p>To enable prefetching on a given column family, you can use HBase Shell or use the API.</p> </div> <div class="exampleblock"> <div class="title">Example 44. Enable Prefetch Using HBase Shell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> create 'MyTable', { NAME => 'myCF', PREFETCH_BLOCKS_ON_OPEN => 'true' }</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 45. Enable Prefetch Using the API</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="comment">// ...</span> HTableDescriptor tableDesc = <span class="keyword">new</span> HTableDescriptor(<span class="string"><span class="delimiter">"</span><span class="content">myTable</span><span class="delimiter">"</span></span>); HColumnDescriptor cfDesc = <span class="keyword">new</span> HColumnDescriptor(<span class="string"><span class="delimiter">"</span><span class="content">myCF</span><span class="delimiter">"</span></span>); cfDesc.setPrefetchBlocksOnOpen(<span class="predefined-constant">true</span>); tableDesc.addFamily(cfDesc); <span class="comment">// ...</span></code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>See the API documentation for <a href="https://hbase.apache.org/apidocs/org/apache/hadoop/hbase/io/hfile/CacheConfig.html">CacheConfig</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.rs.memstore.size"><a class="anchor" href="#perf.rs.memstore.size"></a>86.5. <code>hbase.regionserver.global.memstore.size</code></h3> <div class="paragraph"> <p>See <a href="#hbase.regionserver.global.memstore.size">[hbase.regionserver.global.memstore.size]</a>. This memory setting is often adjusted for the RegionServer process depending on needs.</p> </div> </div> <div class="sect2"> <h3 id="perf.rs.memstore.size.lower.limit"><a class="anchor" href="#perf.rs.memstore.size.lower.limit"></a>86.6. <code>hbase.regionserver.global.memstore.size.lower.limit</code></h3> <div class="paragraph"> <p>See <a href="#hbase.regionserver.global.memstore.size.lower.limit">[hbase.regionserver.global.memstore.size.lower.limit]</a>. This memory setting is often adjusted for the RegionServer process depending on needs.</p> </div> </div> <div class="sect2"> <h3 id="perf.hstore.blockingstorefiles"><a class="anchor" href="#perf.hstore.blockingstorefiles"></a>86.7. <code>hbase.hstore.blockingStoreFiles</code></h3> <div class="paragraph"> <p>See <a href="#hbase.hstore.blockingstorefiles">[hbase.hstore.blockingstorefiles]</a>. If there is blocking in the RegionServer logs, increasing this can help.</p> </div> </div> <div class="sect2"> <h3 id="perf.hregion.memstore.block.multiplier"><a class="anchor" href="#perf.hregion.memstore.block.multiplier"></a>86.8. <code>hbase.hregion.memstore.block.multiplier</code></h3> <div class="paragraph"> <p>See <a href="#hbase.hregion.memstore.block.multiplier">[hbase.hregion.memstore.block.multiplier]</a>. If there is enough RAM, increasing this can help.</p> </div> </div> <div class="sect2"> <h3 id="hbase.regionserver.checksum.verify.performance"><a class="anchor" href="#hbase.regionserver.checksum.verify.performance"></a>86.9. <code>hbase.regionserver.checksum.verify</code></h3> <div class="paragraph"> <p>Have HBase write the checksum into the datablock and save having to do the checksum seek whenever you read.</p> </div> <div class="paragraph"> <p>See <a href="#hbase.regionserver.checksum.verify">[hbase.regionserver.checksum.verify]</a>, <a href="#hbase.hstore.bytes.per.checksum">[hbase.hstore.bytes.per.checksum]</a> and <a href="#hbase.hstore.checksum.algorithm">[hbase.hstore.checksum.algorithm]</a>. For more information see the release note on <a href="https://issues.apache.org/jira/browse/HBASE-5074">HBASE-5074 support checksums in HBase block cache</a>.</p> </div> </div> <div class="sect2"> <h3 id="_tuning_code_callqueue_code_options"><a class="anchor" href="#_tuning_code_callqueue_code_options"></a>86.10. Tuning <code>callQueue</code> Options</h3> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-11355">HBASE-11355</a> introduces several callQueue tuning mechanisms which can increase performance. See the JIRA for some benchmarking information.</p> </div> <div class="paragraph"> <p>To increase the number of callqueues, set <code>hbase.ipc.server.num.callqueue</code> to a value greater than <code>1</code>. To split the callqueue into separate read and write queues, set <code>hbase.ipc.server.callqueue.read.ratio</code> to a value between <code>0</code> and <code>1</code>. This factor weights the queues toward writes (if below .5) or reads (if above .5). Another way to say this is that the factor determines what percentage of the split queues are used for reads. The following examples illustrate some of the possibilities. Note that you always have at least one write queue, no matter what setting you use.</p> </div> <div class="ulist"> <ul> <li> <p>The default value of <code>0</code> does not split the queue.</p> </li> <li> <p>A value of <code>.3</code> uses 30% of the queues for reading and 60% for writing. Given a value of <code>10</code> for <code>hbase.ipc.server.num.callqueue</code>, 3 queues would be used for reads and 7 for writes.</p> </li> <li> <p>A value of <code>.5</code> uses the same number of read queues and write queues. Given a value of <code>10</code> for <code>hbase.ipc.server.num.callqueue</code>, 5 queues would be used for reads and 5 for writes.</p> </li> <li> <p>A value of <code>.6</code> uses 60% of the queues for reading and 30% for reading. Given a value of <code>10</code> for <code>hbase.ipc.server.num.callqueue</code>, 7 queues would be used for reads and 3 for writes.</p> </li> <li> <p>A value of <code>1.0</code> uses one queue to process write requests, and all other queues process read requests. A value higher than <code>1.0</code> has the same effect as a value of <code>1.0</code>. Given a value of <code>10</code> for <code>hbase.ipc.server.num.callqueue</code>, 9 queues would be used for reads and 1 for writes.</p> </li> </ul> </div> <div class="paragraph"> <p>You can also split the read queues so that separate queues are used for short reads (from Get operations) and long reads (from Scan operations), by setting the <code>hbase.ipc.server.callqueue.scan.ratio</code> option. This option is a factor between 0 and 1, which determine the ratio of read queues used for Gets and Scans. More queues are used for Gets if the value is below <code>.5</code> and more are used for scans if the value is above <code>.5</code>. No matter what setting you use, at least one read queue is used for Get operations.</p> </div> <div class="ulist"> <ul> <li> <p>A value of <code>0</code> does not split the read queue.</p> </li> <li> <p>A value of <code>.3</code> uses 60% of the read queues for Gets and 30% for Scans. Given a value of <code>20</code> for <code>hbase.ipc.server.num.callqueue</code> and a value of <code>.5</code> for <code>hbase.ipc.server.callqueue.read.ratio</code>, 10 queues would be used for reads, out of those 10, 7 would be used for Gets and 3 for Scans.</p> </li> <li> <p>A value of <code>.5</code> uses half the read queues for Gets and half for Scans. Given a value of <code>20</code> for <code>hbase.ipc.server.num.callqueue</code> and a value of <code>.5</code> for <code>hbase.ipc.server.callqueue.read.ratio</code>, 10 queues would be used for reads, out of those 10, 5 would be used for Gets and 5 for Scans.</p> </li> <li> <p>A value of <code>.6</code> uses 30% of the read queues for Gets and 60% for Scans. Given a value of <code>20</code> for <code>hbase.ipc.server.num.callqueue</code> and a value of <code>.5</code> for <code>hbase.ipc.server.callqueue.read.ratio</code>, 10 queues would be used for reads, out of those 10, 3 would be used for Gets and 7 for Scans.</p> </li> <li> <p>A value of <code>1.0</code> uses all but one of the read queues for Scans. Given a value of <code>20</code> for <code>hbase.ipc.server.num.callqueue</code> and a value of`.5` for <code>hbase.ipc.server.callqueue.read.ratio</code>, 10 queues would be used for reads, out of those 10, 1 would be used for Gets and 9 for Scans.</p> </li> </ul> </div> <div class="paragraph"> <p>You can use the new option <code>hbase.ipc.server.callqueue.handler.factor</code> to programmatically tune the number of queues:</p> </div> <div class="ulist"> <ul> <li> <p>A value of <code>0</code> uses a single shared queue between all the handlers.</p> </li> <li> <p>A value of <code>1</code> uses a separate queue for each handler.</p> </li> <li> <p>A value between <code>0</code> and <code>1</code> tunes the number of queues against the number of handlers. For instance, a value of <code>.5</code> shares one queue between each two handlers.</p> <div class="paragraph"> <p>Having more queues, such as in a situation where you have one queue per handler, reduces contention when adding a task to a queue or selecting it from a queue. The trade-off is that if you have some queues with long-running tasks, a handler may end up waiting to execute from that queue rather than processing another queue which has waiting tasks.</p> </div> </li> </ul> </div> <div class="paragraph"> <p>For these values to take effect on a given RegionServer, the RegionServer must be restarted. These parameters are intended for testing purposes and should be used carefully.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.zookeeper"><a class="anchor" href="#perf.zookeeper"></a>87. ZooKeeper</h2> <div class="sectionbody"> <div class="paragraph"> <p>See <a href="#zookeeper">ZooKeeper</a> for information on configuring ZooKeeper, and see the part about having a dedicated disk.</p> </div> </div> </div> <div class="sect1"> <h2 id="perf.schema"><a class="anchor" href="#perf.schema"></a>88. Schema Design</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="perf.number.of.cfs"><a class="anchor" href="#perf.number.of.cfs"></a>88.1. Number of Column Families</h3> <div class="paragraph"> <p>See <a href="#number.of.cfs">On the number of column families</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.schema.keys"><a class="anchor" href="#perf.schema.keys"></a>88.2. Key and Attribute Lengths</h3> <div class="paragraph"> <p>See <a href="#keysize">Try to minimize row and column sizes</a>. See also <a href="#perf.compression.however">However…</a> for compression caveats.</p> </div> </div> <div class="sect2"> <h3 id="schema.regionsize"><a class="anchor" href="#schema.regionsize"></a>88.3. Table RegionSize</h3> <div class="paragraph"> <p>The regionsize can be set on a per-table basis via <code>setFileSize</code> on <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HTableDescriptor.html">HTableDescriptor</a> in the event where certain tables require different regionsizes than the configured default regionsize.</p> </div> <div class="paragraph"> <p>See <a href="#ops.capacity.regions">Determining region count and size</a> for more information.</p> </div> </div> <div class="sect2"> <h3 id="schema.bloom"><a class="anchor" href="#schema.bloom"></a>88.4. Bloom Filters</h3> <div class="paragraph"> <p>A Bloom filter, named for its creator, Burton Howard Bloom, is a data structure which is designed to predict whether a given element is a member of a set of data. A positive result from a Bloom filter is not always accurate, but a negative result is guaranteed to be accurate. Bloom filters are designed to be "accurate enough" for sets of data which are so large that conventional hashing mechanisms would be impractical. For more information about Bloom filters in general, refer to <a href="http://en.wikipedia.org/wiki/Bloom_filter" class="bare">http://en.wikipedia.org/wiki/Bloom_filter</a>.</p> </div> <div class="paragraph"> <p>In terms of HBase, Bloom filters provide a lightweight in-memory structure to reduce the number of disk reads for a given Get operation (Bloom filters do not work with Scans) to only the StoreFiles likely to contain the desired Row. The potential performance gain increases with the number of parallel reads.</p> </div> <div class="paragraph"> <p>The Bloom filters themselves are stored in the metadata of each HFile and never need to be updated. When an HFile is opened because a region is deployed to a RegionServer, the Bloom filter is loaded into memory.</p> </div> <div class="paragraph"> <p>HBase includes some tuning mechanisms for folding the Bloom filter to reduce the size and keep the false positive rate within a desired range.</p> </div> <div class="paragraph"> <p>Bloom filters were introduced in <a href="https://issues.apache.org/jira/browse/HBASE-1200">HBASE-1200</a>. Since HBase 0.96, row-based Bloom filters are enabled by default. (<a href="https://issues.apache.org/jira/browse/HBASE-8450">HBASE-</a>)</p> </div> <div class="paragraph"> <p>For more information on Bloom filters in relation to HBase, see <a href="#blooms">Bloom Filters</a> for more information, or the following Quora discussion: <a href="http://www.quora.com/How-are-bloom-filters-used-in-HBase">How are bloom filters used in HBase?</a>.</p> </div> <div class="sect3"> <h4 id="bloom.filters.when"><a class="anchor" href="#bloom.filters.when"></a>88.4.1. When To Use Bloom Filters</h4> <div class="paragraph"> <p>Since HBase 0.96, row-based Bloom filters are enabled by default. You may choose to disable them or to change some tables to use row+column Bloom filters, depending on the characteristics of your data and how it is loaded into HBase.</p> </div> <div class="paragraph"> <p>To determine whether Bloom filters could have a positive impact, check the value of <code>blockCacheHitRatio</code> in the RegionServer metrics. If Bloom filters are enabled, the value of <code>blockCacheHitRatio</code> should increase, because the Bloom filter is filtering out blocks that are definitely not needed.</p> </div> <div class="paragraph"> <p>You can choose to enable Bloom filters for a row or for a row+column combination. If you generally scan entire rows, the row+column combination will not provide any benefit. A row-based Bloom filter can operate on a row+column Get, but not the other way around. However, if you have a large number of column-level Puts, such that a row may be present in every StoreFile, a row-based filter will always return a positive result and provide no benefit. Unless you have one column per row, row+column Bloom filters require more space, in order to store more keys. Bloom filters work best when the size of each data entry is at least a few kilobytes in size.</p> </div> <div class="paragraph"> <p>Overhead will be reduced when your data is stored in a few larger StoreFiles, to avoid extra disk IO during low-level scans to find a specific row.</p> </div> <div class="paragraph"> <p>Bloom filters need to be rebuilt upon deletion, so may not be appropriate in environments with a large number of deletions.</p> </div> </div> <div class="sect3"> <h4 id="_enabling_bloom_filters"><a class="anchor" href="#_enabling_bloom_filters"></a>88.4.2. Enabling Bloom Filters</h4> <div class="paragraph"> <p>Bloom filters are enabled on a Column Family. You can do this by using the setBloomFilterType method of HColumnDescriptor or using the HBase API. Valid values are <code>NONE</code> (the default), <code>ROW</code>, or <code>ROWCOL</code>. See <a href="#bloom.filters.when">When To Use Bloom Filters</a> for more information on <code>ROW</code> versus <code>ROWCOL</code>. See also the API documentation for <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a>.</p> </div> <div class="paragraph"> <p>The following example creates a table and enables a ROWCOL Bloom filter on the <code>colfam1</code> column family.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> create 'mytable',{NAME => 'colfam1', BLOOMFILTER => 'ROWCOL'}</pre> </div> </div> </div> <div class="sect3"> <h4 id="_configuring_server_wide_behavior_of_bloom_filters"><a class="anchor" href="#_configuring_server_wide_behavior_of_bloom_filters"></a>88.4.3. Configuring Server-Wide Behavior of Bloom Filters</h4> <div class="paragraph"> <p>You can configure the following settings in the <em>hbase-site.xml</em>.</p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Parameter</th> <th class="tableblock halign-left valign-top">Default</th> <th class="tableblock halign-left valign-top">Description</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">io.hfile.bloom.enabled</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">yes</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Set to no to kill bloom filters server-wide if something goes wrong</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">io.hfile.bloom.error.rate</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">.01</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The average false positive rate for bloom filters. Folding is used to maintain the false positive rate. Expressed as a decimal representation of a percentage.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">io.hfile.bloom.max.fold</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">7</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The guaranteed maximum fold rate. Changing this setting should not be necessary and is not recommended.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">io.storefile.bloom.max.keys</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">128000000</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">For default (single-block) Bloom filters, this specifies the maximum number of keys.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">io.storefile.delete.family.bloom.enabled</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">true</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Master switch to enable Delete Family Bloom filters and store them in the StoreFile.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">io.storefile.bloom.block.size</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">65536</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Target Bloom block size. Bloom filter blocks of approximately this size are interleaved with data blocks.</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hfile.block.bloom.cacheonwrite</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">false</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Enables cache-on-write for inline blocks of a compound Bloom filter.</p></td> </tr> </tbody> </table> </div> </div> <div class="sect2"> <h3 id="schema.cf.blocksize"><a class="anchor" href="#schema.cf.blocksize"></a>88.5. ColumnFamily BlockSize</h3> <div class="paragraph"> <p>The blocksize can be configured for each ColumnFamily in a table, and defaults to 64k. Larger cell values require larger blocksizes. There is an inverse relationship between blocksize and the resulting StoreFile indexes (i.e., if the blocksize is doubled then the resulting indexes should be roughly halved).</p> </div> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a> and <a href="#store">[store]</a>for more information.</p> </div> </div> <div class="sect2"> <h3 id="cf.in.memory"><a class="anchor" href="#cf.in.memory"></a>88.6. In-Memory ColumnFamilies</h3> <div class="paragraph"> <p>ColumnFamilies can optionally be defined as in-memory. Data is still persisted to disk, just like any other ColumnFamily. In-memory blocks have the highest priority in the <a href="#block.cache">Block Cache</a>, but it is not a guarantee that the entire table will be in memory.</p> </div> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HColumnDescriptor.html">HColumnDescriptor</a> for more information.</p> </div> </div> <div class="sect2"> <h3 id="perf.compression"><a class="anchor" href="#perf.compression"></a>88.7. Compression</h3> <div class="paragraph"> <p>Production systems should use compression with their ColumnFamily definitions. See <a href="#compression">Compression and Data Block Encoding In HBase</a> for more information.</p> </div> <div class="sect3"> <h4 id="perf.compression.however"><a class="anchor" href="#perf.compression.however"></a>88.7.1. However…</h4> <div class="paragraph"> <p>Compression deflates data <em>on disk</em>. When it’s in-memory (e.g., in the MemStore) or on the wire (e.g., transferring between RegionServer and Client) it’s inflated. So while using ColumnFamily compression is a best practice, but it’s not going to completely eliminate the impact of over-sized Keys, over-sized ColumnFamily names, or over-sized Column names.</p> </div> <div class="paragraph"> <p>See <a href="#keysize">Try to minimize row and column sizes</a> on for schema design tips, and <a href="#keyvalue">[keyvalue]</a> for more information on HBase stores data internally.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.general"><a class="anchor" href="#perf.general"></a>89. HBase General Patterns</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="perf.general.constants"><a class="anchor" href="#perf.general.constants"></a>89.1. Constants</h3> <div class="paragraph"> <p>When people get started with HBase they have a tendency to write code that looks like this:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Get get = <span class="keyword">new</span> Get(rowkey); <span class="predefined-type">Result</span> r = table.get(get); <span class="type">byte</span><span class="type">[]</span> b = r.getValue(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>)); <span class="comment">// returns current version of value</span></code></pre> </div> </div> <div class="paragraph"> <p>But especially when inside loops (and MapReduce jobs), converting the columnFamily and column-names to byte-arrays repeatedly is surprisingly expensive. It’s better to use constants for the byte-arrays, like this:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">cf</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> ATTR = <span class="string"><span class="delimiter">"</span><span class="content">attr</span><span class="delimiter">"</span></span>.getBytes(); ... Get get = <span class="keyword">new</span> Get(rowkey); <span class="predefined-type">Result</span> r = table.get(get); <span class="type">byte</span><span class="type">[]</span> b = r.getValue(CF, ATTR); <span class="comment">// returns current version of value</span></code></pre> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.writing"><a class="anchor" href="#perf.writing"></a>90. Writing to HBase</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="perf.batch.loading"><a class="anchor" href="#perf.batch.loading"></a>90.1. Batch Loading</h3> <div class="paragraph"> <p>Use the bulk load tool if you can. See <a href="#arch.bulk.load">Bulk Loading</a>. Otherwise, pay attention to the below.</p> </div> </div> <div class="sect2"> <h3 id="precreate.regions"><a class="anchor" href="#precreate.regions"></a>90.2. Table Creation: Pre-Creating Regions</h3> <div class="paragraph"> <p>Tables in HBase are initially created with one region by default. For bulk imports, this means that all clients will write to the same region until it is large enough to split and become distributed across the cluster. A useful pattern to speed up the bulk import process is to pre-create empty regions. Be somewhat conservative in this, because too-many regions can actually degrade performance.</p> </div> <div class="paragraph"> <p>There are two different approaches to pre-creating splits. The first approach is to rely on the default <code>Admin</code> strategy (which is implemented in <code>Bytes.split</code>)…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="type">byte</span><span class="type">[]</span> startKey = ...; <span class="comment">// your lowest key</span> <span class="type">byte</span><span class="type">[]</span> endKey = ...; <span class="comment">// your highest key</span> <span class="type">int</span> numberOfRegions = ...; <span class="comment">// # of regions to create</span> admin.createTable(table, startKey, endKey, numberOfRegions);</code></pre> </div> </div> <div class="paragraph"> <p>And the other approach is to define the splits yourself…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="type">byte</span><span class="type">[]</span><span class="type">[]</span> splits = ...; <span class="comment">// create your own splits</span> admin.createTable(table, splits);</code></pre> </div> </div> <div class="paragraph"> <p>See <a href="#rowkey.regionsplits">Relationship Between RowKeys and Region Splits</a> for issues related to understanding your keyspace and pre-creating regions. See <a href="#manual_region_splitting_decisions">manual region splitting decisions</a> for discussion on manually pre-splitting regions.</p> </div> </div> <div class="sect2"> <h3 id="def.log.flush"><a class="anchor" href="#def.log.flush"></a>90.3. Table Creation: Deferred Log Flush</h3> <div class="paragraph"> <p>The default behavior for Puts using the Write Ahead Log (WAL) is that <code>WAL</code> edits will be written immediately. If deferred log flush is used, WAL edits are kept in memory until the flush period. The benefit is aggregated and asynchronous <code>WAL</code>- writes, but the potential downside is that if the RegionServer goes down the yet-to-be-flushed edits are lost. This is safer, however, than not using WAL at all with Puts.</p> </div> <div class="paragraph"> <p>Deferred log flush can be configured on tables via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HTableDescriptor.html">HTableDescriptor</a>. The default value of <code>hbase.regionserver.optionallogflushinterval</code> is 1000ms.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.autoflush"><a class="anchor" href="#perf.hbase.client.autoflush"></a>90.4. HBase Client: AutoFlush</h3> <div class="paragraph"> <p>When performing a lot of Puts, make sure that setAutoFlush is set to false on your <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html">Table</a> instance. Otherwise, the Puts will be sent one at a time to the RegionServer. Puts added via <code>table.add(Put)</code> and <code>table.add( <List> Put)</code> wind up in the same write buffer. If <code>autoFlush = false</code>, these messages are not sent until the write-buffer is filled. To explicitly flush the messages, call <code>flushCommits</code>. Calling <code>close</code> on the <code>Table</code> instance will invoke <code>flushCommits</code>.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.putwal"><a class="anchor" href="#perf.hbase.client.putwal"></a>90.5. HBase Client: Turn off WAL on Puts</h3> <div class="paragraph"> <p>A frequent request is to disable the WAL to increase performance of Puts. This is only appropriate for bulk loads, as it puts your data at risk by removing the protection of the WAL in the event of a region server crash. Bulk loads can be re-run in the event of a crash, with little risk of data loss.</p> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> If you disable the WAL for anything other than bulk loads, your data is at risk. </td> </tr> </table> </div> <div class="paragraph"> <p>In general, it is best to use WAL for Puts, and where loading throughput is a concern to use bulk loading techniques instead. For normal Puts, you are not likely to see a performance improvement which would outweigh the risk. To disable the WAL, see <a href="#wal.disable">Disabling the WAL</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.regiongroup"><a class="anchor" href="#perf.hbase.client.regiongroup"></a>90.6. HBase Client: Group Puts by RegionServer</h3> <div class="paragraph"> <p>In addition to using the writeBuffer, grouping <code>Put`s by RegionServer can reduce the number of client RPC calls per writeBuffer flush. There is a utility `HTableUtil</code> currently on TRUNK that does this, but you can either copy that or implement your own version for those still on 0.90.x or earlier.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.write.mr.reducer"><a class="anchor" href="#perf.hbase.write.mr.reducer"></a>90.7. MapReduce: Skip The Reducer</h3> <div class="paragraph"> <p>When writing a lot of data to an HBase table from a MR job (e.g., with <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.html">TableOutputFormat</a>), and specifically where Puts are being emitted from the Mapper, skip the Reducer step. When a Reducer step is used, all of the output (Puts) from the Mapper will get spooled to disk, then sorted/shuffled to other Reducers that will most likely be off-node. It’s far more efficient to just write directly to HBase.</p> </div> <div class="paragraph"> <p>For summary jobs where HBase is used as a source and a sink, then writes will be coming from the Reducer step (e.g., summarize values then write out result). This is a different processing problem than from the the above case.</p> </div> </div> <div class="sect2"> <h3 id="perf.one.region"><a class="anchor" href="#perf.one.region"></a>90.8. Anti-Pattern: One Hot Region</h3> <div class="paragraph"> <p>If all your data is being written to one region at a time, then re-read the section on processing timeseries data.</p> </div> <div class="paragraph"> <p>Also, if you are pre-splitting regions and all your data is <em>still</em> winding up in a single region even though your keys aren’t monotonically increasing, confirm that your keyspace actually works with the split strategy. There are a variety of reasons that regions may appear "well split" but won’t work with your data. As the HBase client communicates directly with the RegionServers, this can be obtained via <a href="hhttp://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#getRegionLocation(byte" class="bare">hhttp://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#getRegionLocation(byte</a>)[Table.getRegionLocation].</p> </div> <div class="paragraph"> <p>See <a href="#precreate.regions">Table Creation: Pre-Creating Regions</a>, as well as <a href="#perf.configurations">HBase Configurations</a></p> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.reading"><a class="anchor" href="#perf.reading"></a>91. Reading from HBase</h2> <div class="sectionbody"> <div class="paragraph"> <p>The mailing list can help if you are having performance issues. For example, here is a good general thread on what to look at addressing read-time issues: <a href="http://search-hadoop.com/m/qOo2yyHtCC1">HBase Random Read latency > 100ms</a></p> </div> <div class="sect2"> <h3 id="perf.hbase.client.caching"><a class="anchor" href="#perf.hbase.client.caching"></a>91.1. Scan Caching</h3> <div class="paragraph"> <p>If HBase is used as an input source for a MapReduce job, for example, make sure that the input <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</a> instance to the MapReduce job has <code>setCaching</code> set to something greater than the default (which is 1). Using the default value means that the map-task will make call back to the region-server for every record processed. Setting this value to 500, for example, will transfer 500 rows at a time to the client to be processed. There is a cost/benefit to have the cache value be large because it costs more in memory for both client and RegionServer, so bigger isn’t always better.</p> </div> <div class="sect3"> <h4 id="perf.hbase.client.caching.mr"><a class="anchor" href="#perf.hbase.client.caching.mr"></a>91.1.1. Scan Caching in MapReduce Jobs</h4> <div class="paragraph"> <p>Scan settings in MapReduce jobs deserve special attention. Timeouts can result (e.g., UnknownScannerException) in Map tasks if it takes longer to process a batch of records before the client goes back to the RegionServer for the next set of data. This problem can occur because there is non-trivial processing occurring per row. If you process rows quickly, set caching higher. If you process rows more slowly (e.g., lots of transformations per row, writes), then set caching lower.</p> </div> <div class="paragraph"> <p>Timeouts can also happen in a non-MapReduce use case (i.e., single threaded HBase client doing a Scan), but the processing that is often performed in MapReduce jobs tends to exacerbate this issue.</p> </div> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.selection"><a class="anchor" href="#perf.hbase.client.selection"></a>91.2. Scan Attribute Selection</h3> <div class="paragraph"> <p>Whenever a Scan is used to process large numbers of rows (and especially when used as a MapReduce source), be aware of which attributes are selected. If <code>scan.addFamily</code> is called then <em>all</em> of the attributes in the specified ColumnFamily will be returned to the client. If only a small number of the available attributes are to be processed, then only those attributes should be specified in the input scan because attribute over-selection is a non-trivial performance penalty over large datasets.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.seek"><a class="anchor" href="#perf.hbase.client.seek"></a>91.3. Avoid scan seeks</h3> <div class="paragraph"> <p>When columns are selected explicitly with <code>scan.addColumn</code>, HBase will schedule seek operations to seek between the selected columns. When rows have few columns and each column has only a few versions this can be inefficient. A seek operation is generally slower if does not seek at least past 5-10 columns/versions or 512-1024 bytes.</p> </div> <div class="paragraph"> <p>In order to opportunistically look ahead a few columns/versions to see if the next column/version can be found that way before a seek operation is scheduled, a new attribute <code>Scan.HINT_LOOKAHEAD</code> can be set the on Scan object. The following code instructs the RegionServer to attempt two iterations of next before a seek is scheduled:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Scan scan = <span class="keyword">new</span> Scan(); scan.addColumn(...); scan.setAttribute(Scan.HINT_LOOKAHEAD, Bytes.toBytes(<span class="integer">2</span>)); table.getScanner(scan);</code></pre> </div> </div> </div> <div class="sect2"> <h3 id="perf.hbase.mr.input"><a class="anchor" href="#perf.hbase.mr.input"></a>91.4. MapReduce - Input Splits</h3> <div class="paragraph"> <p>For MapReduce jobs that use HBase tables as a source, if there a pattern where the "slow" map tasks seem to have the same Input Split (i.e., the RegionServer serving the data), see the Troubleshooting Case Study in <a href="#casestudies.slownode">Case Study #1 (Performance Issue On A Single Node)</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.scannerclose"><a class="anchor" href="#perf.hbase.client.scannerclose"></a>91.5. Close ResultScanners</h3> <div class="paragraph"> <p>This isn’t so much about improving performance but rather <em>avoiding</em> performance problems. If you forget to close <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/ResultScanner.html">ResultScanners</a> you can cause problems on the RegionServers. Always have ResultScanner processing enclosed in try/catch blocks.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Scan scan = <span class="keyword">new</span> Scan(); <span class="comment">// set attrs...</span> ResultScanner rs = table.getScanner(scan); <span class="keyword">try</span> { <span class="keyword">for</span> (<span class="predefined-type">Result</span> r = rs.next(); r != <span class="predefined-constant">null</span>; r = rs.next()) { <span class="comment">// process result...</span> } <span class="keyword">finally</span> { rs.close(); <span class="comment">// always close the ResultScanner!</span> } table.close();</code></pre> </div> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.blockcache"><a class="anchor" href="#perf.hbase.client.blockcache"></a>91.6. Block Cache</h3> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">Scan</a> instances can be set to use the block cache in the RegionServer via the <code>setCacheBlocks</code> method. For input Scans to MapReduce jobs, this should be <code>false</code>. For frequently accessed rows, it is advisable to use the block cache.</p> </div> <div class="paragraph"> <p>Cache more data by moving your Block Cache off-heap. See <a href="#offheap.blockcache">Off-heap Block Cache</a></p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.client.rowkeyonly"><a class="anchor" href="#perf.hbase.client.rowkeyonly"></a>91.7. Optimal Loading of Row Keys</h3> <div class="paragraph"> <p>When performing a table <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Scan.html">scan</a> where only the row keys are needed (no families, qualifiers, values or timestamps), add a FilterList with a <code>MUST_PASS_ALL</code> operator to the scanner using <code>setFilter</code>. The filter list should include both a <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FirstKeyOnlyFilter.html">FirstKeyOnlyFilter</a> and a <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/KeyOnlyFilter.html">KeyOnlyFilter</a>. Using this filter combination will result in a worst case scenario of a RegionServer reading a single value from disk and minimal network traffic to the client for a single row.</p> </div> </div> <div class="sect2"> <h3 id="perf.hbase.read.dist"><a class="anchor" href="#perf.hbase.read.dist"></a>91.8. Concurrency: Monitor Data Spread</h3> <div class="paragraph"> <p>When performing a high number of concurrent reads, monitor the data spread of the target tables. If the target table(s) have too few regions then the reads could likely be served from too few nodes.</p> </div> <div class="paragraph"> <p>See <a href="#precreate.regions">Table Creation: Pre-Creating Regions</a>, as well as <a href="#perf.configurations">HBase Configurations</a></p> </div> </div> <div class="sect2"> <h3 id="blooms"><a class="anchor" href="#blooms"></a>91.9. Bloom Filters</h3> <div class="paragraph"> <p>Enabling Bloom Filters can save your having to go to disk and can help improve read latencies.</p> </div> <div class="paragraph"> <p><a href="http://en.wikipedia.org/wiki/Bloom_filter">Bloom filters</a> were developed over in <a href="https://issues.apache.org/jira/browse/HBASE-1200">HBase-1200 Add bloomfilters</a>. For description of the development process — why static blooms rather than dynamic — and for an overview of the unique properties that pertain to blooms in HBase, as well as possible future directions, see the <em>Development Process</em> section of the document <a href="https://issues.apache.org/jira/secure/attachment/12444007/Bloom_Filters_in_HBase.pdf">BloomFilters in HBase</a> attached to <a href="https://issues.apache.org/jira/browse/HBASE-1200">HBASE-1200</a>. The bloom filters described here are actually version two of blooms in HBase. In versions up to 0.19.x, HBase had a dynamic bloom option based on work done by the <a href="http://www.one-lab.org">European Commission One-Lab Project 034819</a>. The core of the HBase bloom work was later pulled up into Hadoop to implement org.apache.hadoop.io.BloomMapFile. Version 1 of HBase blooms never worked that well. Version 2 is a rewrite from scratch though again it starts with the one-lab work.</p> </div> <div class="paragraph"> <p>See also <a href="#schema.bloom">Bloom Filters</a>.</p> </div> <div class="sect3"> <h4 id="bloom_footprint"><a class="anchor" href="#bloom_footprint"></a>91.9.1. Bloom StoreFile footprint</h4> <div class="paragraph"> <p>Bloom filters add an entry to the <code>StoreFile</code> general <code>FileInfo</code> data structure and then two extra entries to the <code>StoreFile</code> metadata section.</p> </div> <div class="sect4"> <h5 id="_bloomfilter_in_the_code_storefile_fileinfo_code_data_structure"><a class="anchor" href="#_bloomfilter_in_the_code_storefile_fileinfo_code_data_structure"></a>BloomFilter in the <code>StoreFile``FileInfo</code> data structure</h5> <div class="paragraph"> <p><code>FileInfo</code> has a <code>BLOOM_FILTER_TYPE</code> entry which is set to <code>NONE</code>, <code>ROW</code> or <code>ROWCOL.</code></p> </div> </div> <div class="sect4"> <h5 id="_bloomfilter_entries_in_code_storefile_code_metadata"><a class="anchor" href="#_bloomfilter_entries_in_code_storefile_code_metadata"></a>BloomFilter entries in <code>StoreFile</code> metadata</h5> <div class="paragraph"> <p><code>BLOOM_FILTER_META</code> holds Bloom Size, Hash Function used, etc. It’s small in size and is cached on <code>StoreFile.Reader</code> load</p> </div> <div class="paragraph"> <p><code>BLOOM_FILTER_DATA</code> is the actual bloomfilter data. Obtained on-demand. Stored in the LRU cache, if it is enabled (It’s enabled by default).</p> </div> </div> </div> <div class="sect3"> <h4 id="config.bloom"><a class="anchor" href="#config.bloom"></a>91.9.2. Bloom Filter Configuration</h4> <div class="sect4"> <h5 id="__code_io_hfile_bloom_enabled_code_global_kill_switch"><a class="anchor" href="#__code_io_hfile_bloom_enabled_code_global_kill_switch"></a><code>io.hfile.bloom.enabled</code> global kill switch</h5> <div class="paragraph"> <p><code>io.hfile.bloom.enabled</code> in <code>Configuration</code> serves as the kill switch in case something goes wrong. Default = <code>true</code>.</p> </div> </div> <div class="sect4"> <h5 id="__code_io_hfile_bloom_error_rate_code"><a class="anchor" href="#__code_io_hfile_bloom_error_rate_code"></a><code>io.hfile.bloom.error.rate</code></h5> <div class="paragraph"> <p><code>io.hfile.bloom.error.rate</code> = average false positive rate. Default = 1%. Decrease rate by ½ (e.g. to .5%) == +1 bit per bloom entry.</p> </div> </div> <div class="sect4"> <h5 id="__code_io_hfile_bloom_max_fold_code"><a class="anchor" href="#__code_io_hfile_bloom_max_fold_code"></a><code>io.hfile.bloom.max.fold</code></h5> <div class="paragraph"> <p><code>io.hfile.bloom.max.fold</code> = guaranteed minimum fold rate. Most people should leave this alone. Default = 7, or can collapse to at least 1/128th of original size. See the <em>Development Process</em> section of the document <a href="https://issues.apache.org/jira/secure/attachment/12444007/Bloom_Filters_in_HBase.pdf">BloomFilters in HBase</a> for more on what this option means.</p> </div> </div> </div> </div> <div class="sect2"> <h3 id="_hedged_reads"><a class="anchor" href="#_hedged_reads"></a>91.10. Hedged Reads</h3> <div class="paragraph"> <p>Hedged reads are a feature of HDFS, introduced in <a href="https://issues.apache.org/jira/browse/HDFS-5776">HDFS-5776</a>. Normally, a single thread is spawned for each read request. However, if hedged reads are enabled, the client waits some configurable amount of time, and if the read does not return, the client spawns a second read request, against a different block replica of the same data. Whichever read returns first is used, and the other read request is discarded. Hedged reads can be helpful for times where a rare slow read is caused by a transient error such as a failing disk or flaky network connection.</p> </div> <div class="paragraph"> <p>Because a HBase RegionServer is a HDFS client, you can enable hedged reads in HBase, by adding the following properties to the RegionServer’s hbase-site.xml and tuning the values to suit your environment.</p> </div> <div class="ulist"> <div class="title">Configuration for Hedged Reads</div> <ul> <li> <p><code>dfs.client.hedged.read.threadpool.size</code> - the number of threads dedicated to servicing hedged reads. If this is set to 0 (the default), hedged reads are disabled.</p> </li> <li> <p><code>dfs.client.hedged.read.threshold.millis</code> - the number of milliseconds to wait before spawning a second read thread.</p> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 46. Hedged Reads Configuration Example</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>dfs.client.hedged.read.threadpool.size<span class="tag"></name></span> <span class="tag"><value></span>20<span class="tag"></value></span> <span class="comment"></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.client.hedged.read.threshold.millis<span class="tag"></name></span> <span class="tag"><value></span>10<span class="tag"></value></span> <span class="comment"></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> <div class="paragraph"> <p>Use the following metrics to tune the settings for hedged reads on your cluster. See <a href="#hbase_metrics">[hbase_metrics]</a> for more information.</p> </div> <div class="ulist"> <div class="title">Metrics for Hedged Reads</div> <ul> <li> <p>hedgedReadOps - the number of times hedged read threads have been triggered. This could indicate that read requests are often slow, or that hedged reads are triggered too quickly.</p> </li> <li> <p>hedgeReadOpsWin - the number of times the hedged read thread was faster than the original thread. This could indicate that a given RegionServer is having trouble servicing requests.</p> </li> </ul> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.deleting"><a class="anchor" href="#perf.deleting"></a>92. Deleting from HBase</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="perf.deleting.queue"><a class="anchor" href="#perf.deleting.queue"></a>92.1. Using HBase Tables as Queues</h3> <div class="paragraph"> <p>HBase tables are sometimes used as queues. In this case, special care must be taken to regularly perform major compactions on tables used in this manner. As is documented in <a href="#datamodel">Data Model</a>, marking rows as deleted creates additional StoreFiles which then need to be processed on reads. Tombstones only get cleaned up with major compactions.</p> </div> <div class="paragraph"> <p>See also <a href="#compaction">[compaction]</a> and <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Admin.html#majorCompact%28java.lang.String%29">Admin.majorCompact</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.deleting.rpc"><a class="anchor" href="#perf.deleting.rpc"></a>92.2. Delete RPC Behavior</h3> <div class="paragraph"> <p>Be aware that <code>Table.delete(Delete)</code> doesn’t use the writeBuffer. It will execute an RegionServer RPC with each invocation. For a large number of deletes, consider <code>Table.delete(List)</code>.</p> </div> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#delete%28org.apache.hadoop.hbase.client.Delete%29" class="bare">http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html#delete%28org.apache.hadoop.hbase.client.Delete%29</a></p> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.hdfs"><a class="anchor" href="#perf.hdfs"></a>93. HDFS</h2> <div class="sectionbody"> <div class="paragraph"> <p>Because HBase runs on <a href="#arch.hdfs">HDFS</a> it is important to understand how it works and how it affects HBase.</p> </div> <div class="sect2"> <h3 id="perf.hdfs.curr"><a class="anchor" href="#perf.hdfs.curr"></a>93.1. Current Issues With Low-Latency Reads</h3> <div class="paragraph"> <p>The original use-case for HDFS was batch processing. As such, there low-latency reads were historically not a priority. With the increased adoption of Apache HBase this is changing, and several improvements are already in development. See the <a href="https://issues.apache.org/jira/browse/HDFS-1599">Umbrella Jira Ticket for HDFS Improvements for HBase</a>.</p> </div> </div> <div class="sect2"> <h3 id="perf.hdfs.configs.localread"><a class="anchor" href="#perf.hdfs.configs.localread"></a>93.2. Leveraging local data</h3> <div class="paragraph"> <p>Since Hadoop 1.0.0 (also 0.22.1, 0.23.1, CDH3u3 and HDP 1.0) via <a href="https://issues.apache.org/jira/browse/HDFS-2246">HDFS-2246</a>, it is possible for the DFSClient to take a "short circuit" and read directly from the disk instead of going through the DataNode when the data is local. What this means for HBase is that the RegionServers can read directly off their machine’s disks instead of having to open a socket to talk to the DataNode, the former being generally much faster. See JD’s <a href="http://files.meetup.com/1350427/hug_ebay_jdcryans.pdf">Performance Talk</a>. Also see <a href="http://search-hadoop.com/m/zV6dKrLCVh1">HBase, mail # dev - read short circuit</a> thread for more discussion around short circuit reads.</p> </div> <div class="paragraph"> <p>To enable "short circuit" reads, it will depend on your version of Hadoop. The original shortcircuit read patch was much improved upon in Hadoop 2 in <a href="https://issues.apache.org/jira/browse/HDFS-347">HDFS-347</a>. See <a href="http://blog.cloudera.com/blog/2013/08/how-improved-short-circuit-local-reads-bring-better-performance-and-security-to-hadoop/" class="bare">http://blog.cloudera.com/blog/2013/08/how-improved-short-circuit-local-reads-bring-better-performance-and-security-to-hadoop/</a> for details on the difference between the old and new implementations. See <a href="http://archive.cloudera.com/cdh4/cdh/4/hadoop/hadoop-project-dist/hadoop-hdfs/ShortCircuitLocalReads.html">Hadoop shortcircuit reads configuration page</a> for how to enable the latter, better version of shortcircuit. For example, here is a minimal config. enabling short-circuit reads added to <em>hbase-site.xml</em>:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>dfs.client.read.shortcircuit<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"><description></span> This configuration parameter turns on short-circuit local reads. <span class="tag"></description></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>dfs.domain.socket.path<span class="tag"></name></span> <span class="tag"><value></span>/home/stack/sockets/short_circuit_read_socket_PORT<span class="tag"></value></span> <span class="tag"><description></span> Optional. This is a path to a UNIX domain socket that will be used for communication between the DataNode and local HDFS clients. If the string "_PORT" is present in this path, it will be replaced by the TCP port of the DataNode. <span class="tag"></description></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Be careful about permissions for the directory that hosts the shared domain socket; dfsclient will complain if open to other than the hbase user.</p> </div> <div class="paragraph"> <p>If you are running on an old Hadoop, one that is without <a href="https://issues.apache.org/jira/browse/HDFS-347">HDFS-347</a> but that has <a href="https://issues.apache.org/jira/browse/HDFS-2246">HDFS-2246</a>, you must set two configurations. First, the hdfs-site.xml needs to be amended. Set the property <code>dfs.block.local-path-access.user</code> to be the <em>only</em> user that can use the shortcut. This has to be the user that started HBase. Then in hbase-site.xml, set <code>dfs.client.read.shortcircuit</code> to be <code>true</code></p> </div> <div class="paragraph"> <p>Services — at least the HBase RegionServers — will need to be restarted in order to pick up the new configurations.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">dfs.client.read.shortcircuit.buffer.size</div> <div class="paragraph"> <p>The default for this value is too high when running on a highly trafficked HBase. In HBase, if this value has not been set, we set it down from the default of 1M to 128k (Since HBase 0.98.0 and 0.96.1). See <a href="https://issues.apache.org/jira/browse/HBASE-8143">HBASE-8143 HBase on Hadoop 2 with local short circuit reads (ssr) causes OOM</a>). The Hadoop DFSClient in HBase will allocate a direct byte buffer of this size for <em>each</em> block it has open; given HBase keeps its HDFS files open all the time, this can add up quickly.</p> </div> </td> </tr> </table> </div> </div> <div class="sect2"> <h3 id="perf.hdfs.comp"><a class="anchor" href="#perf.hdfs.comp"></a>93.3. Performance Comparisons of HBase vs. HDFS</h3> <div class="paragraph"> <p>A fairly common question on the dist-list is why HBase isn’t as performant as HDFS files in a batch context (e.g., as a MapReduce source or sink). The short answer is that HBase is doing a lot more than HDFS (e.g., reading the KeyValues, returning the most current row or specified timestamps, etc.), and as such HBase is 4-5 times slower than HDFS in this processing context. There is room for improvement and this gap will, over time, be reduced, but HDFS will always be faster in this use-case.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="perf.ec2"><a class="anchor" href="#perf.ec2"></a>94. Amazon EC2</h2> <div class="sectionbody"> <div class="paragraph"> <p>Performance questions are common on Amazon EC2 environments because it is a shared environment. You will not see the same throughput as a dedicated server. In terms of running tests on EC2, run them several times for the same reason (i.e., it’s a shared environment and you don’t know what else is happening on the server).</p> </div> <div class="paragraph"> <p>If you are running on EC2 and post performance questions on the dist-list, please state this fact up-front that because EC2 issues are practically a separate class of performance issues.</p> </div> </div> </div> <div class="sect1"> <h2 id="perf.hbase.mr.cluster"><a class="anchor" href="#perf.hbase.mr.cluster"></a>95. Collocating HBase and MapReduce</h2> <div class="sectionbody"> <div class="paragraph"> <p>It is often recommended to have different clusters for HBase and MapReduce. A better qualification of this is: don’t collocate a HBase that serves live requests with a heavy MR workload. OLTP and OLAP-optimized systems have conflicting requirements and one will lose to the other, usually the former. For example, short latency-sensitive disk reads will have to wait in line behind longer reads that are trying to squeeze out as much throughput as possible. MR jobs that write to HBase will also generate flushes and compactions, which will in turn invalidate blocks in the <a href="#block.cache">Block Cache</a>.</p> </div> <div class="paragraph"> <p>If you need to process the data from your live HBase cluster in MR, you can ship the deltas with <a href="#copy.table">[copy.table]</a> or use replication to get the new data in real time on the OLAP cluster. In the worst case, if you really need to collocate both, set MR to use less Map and Reduce slots than you’d normally configure, possibly just one.</p> </div> <div class="paragraph"> <p>When HBase is used for OLAP operations, it’s preferable to set it up in a hardened way like configuring the ZooKeeper session timeout higher and giving more memory to the MemStores (the argument being that the Block Cache won’t be used much since the workloads are usually long scans).</p> </div> </div> </div> <div class="sect1"> <h2 id="perf.casestudy"><a class="anchor" href="#perf.casestudy"></a>96. Case Studies</h2> <div class="sectionbody"> <div class="paragraph"> <p>For Performance and Troubleshooting Case Studies, see <a href="#casestudies">Apache HBase Case Studies</a>.</p> </div> </div> </div> <h1 id="profiler" class="sect0"><a class="anchor" href="#profiler"></a>Profiler Servlet</h1> <div class="sect1"> <h2 id="_background"><a class="anchor" href="#_background"></a>97. Background</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBASE-21926 introduced a new servlet that supports integrated profiling via async-profiler.</p> </div> </div> </div> <div class="sect1"> <h2 id="_prerequisites_2"><a class="anchor" href="#_prerequisites_2"></a>98. Prerequisites</h2> <div class="sectionbody"> <div class="paragraph"> <p>Go to <a href="https://github.com/jvm-profiling-tools/async-profiler" class="bare">https://github.com/jvm-profiling-tools/async-profiler</a>, download a release appropriate for your platform, and install on every cluster host.</p> </div> <div class="paragraph"> <p>Set <code>ASYNC_PROFILER_HOME</code> in the environment (put it in hbase-env.sh) to the root directory of the async-profiler install location, or pass it on the HBase daemon’s command line as a system property as <code>-Dasync.profiler.home=/path/to/async-profiler</code>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_usage"><a class="anchor" href="#_usage"></a>99. Usage</h2> <div class="sectionbody"> <div class="paragraph"> <p>Once the prerequisites are satisfied, access to async-profiler is available by way of the HBase UI or direct interaction with the infoserver.</p> </div> <div class="paragraph"> <p>Examples:</p> </div> <div class="ulist"> <ul> <li> <p>To collect 30 second CPU profile of current process (returns FlameGraph svg) <code>curl <a href="http://localhost:16030/prof" class="bare">http://localhost:16030/prof</a></code></p> </li> <li> <p>To collect 1 minute CPU profile of current process and output in tree format (html) <code>curl <a href="http://localhost:16030/prof?output=tree&duration=60" class="bare">http://localhost:16030/prof?output=tree&duration=60</a></code></p> </li> <li> <p>To collect 30 second heap allocation profile of current process (returns FlameGraph svg) <code>curl <a href="http://localhost:16030/prof?event=alloc" class="bare">http://localhost:16030/prof?event=alloc</a></code></p> </li> <li> <p>To collect lock contention profile of current process (returns FlameGraph svg) <code>curl <a href="http://localhost:16030/prof?event=lock" class="bare">http://localhost:16030/prof?event=lock</a></code></p> </li> </ul> </div> <div class="paragraph"> <p>The following event types are supported by async-profiler. Use the 'event' parameter to specify. Default is 'cpu'. Not all operating systems will support all types.</p> </div> <div class="paragraph"> <p>Perf events:</p> </div> <div class="ulist"> <ul> <li> <p>cpu</p> </li> <li> <p>page-faults</p> </li> <li> <p>context-switches</p> </li> <li> <p>cycles</p> </li> <li> <p>instructions</p> </li> <li> <p>cache-references</p> </li> <li> <p>cache-misses</p> </li> <li> <p>branches</p> </li> <li> <p>branch-misses</p> </li> <li> <p>bus-cycles</p> </li> <li> <p>L1-dcache-load-misses</p> </li> <li> <p>LLC-load-misses</p> </li> <li> <p>dTLB-load-misses</p> </li> </ul> </div> <div class="paragraph"> <p>Java events:</p> </div> <div class="ulist"> <ul> <li> <p>alloc</p> </li> <li> <p>lock</p> </li> </ul> </div> <div class="paragraph"> <p>The following output formats are supported. Use the 'output' parameter to specify. Default is 'flamegraph'.</p> </div> <div class="paragraph"> <p>Output formats:</p> </div> <div class="ulist"> <ul> <li> <p>summary: A dump of basic profiling statistics.</p> </li> <li> <p>traces: Call traces.</p> </li> <li> <p>flat: Flat profile (top N hot methods).</p> </li> <li> <p>collapsed: Collapsed call traces in the format used by FlameGraph script. This is a collection of call stacks, where each line is a semicolon separated list of frames followed by a counter.</p> </li> <li> <p>svg: FlameGraph in SVG format.</p> </li> <li> <p>tree: Call tree in HTML format.</p> </li> <li> <p>jfr: Call traces in Java Flight Recorder format.</p> </li> </ul> </div> <div class="paragraph"> <p>The 'duration' parameter specifies how long to collect trace data before generating output, specified in seconds. The default is 10 seconds.</p> </div> </div> </div> <div class="sect1"> <h2 id="_ui"><a class="anchor" href="#_ui"></a>100. UI</h2> <div class="sectionbody"> <div class="paragraph"> <p>In the UI, there is a new entry 'Profiler' in the top menu that will run the default action, which is to profile the CPU usage of the local process for thirty seconds and then produce FlameGraph SVG output.</p> </div> </div> </div> <div class="sect1"> <h2 id="_notes"><a class="anchor" href="#_notes"></a>101. Notes</h2> <div class="sectionbody"> <div class="paragraph"> <p>The query parameter <code>pid</code> can be used to specify the process id of a specific process to be profiled. If this parameter is missing the local process in which the infoserver is embedded will be profiled. Profile targets that are not JVMs might work but is not specifically supported. There are security implications. Access to the infoserver should be appropriately restricted.</p> </div> </div> </div> <h1 id="trouble" class="sect0"><a class="anchor" href="#trouble"></a>Troubleshooting and Debugging Apache HBase</h1> <div class="sect1"> <h2 id="trouble.general"><a class="anchor" href="#trouble.general"></a>102. General Guidelines</h2> <div class="sectionbody"> <div class="paragraph"> <p>Always start with the master log (TODO: Which lines?). Normally it’s just printing the same lines over and over again. If not, then there’s an issue. Google or <a href="http://search-hadoop.com">search-hadoop.com</a> should return some hits for those exceptions you’re seeing.</p> </div> <div class="paragraph"> <p>An error rarely comes alone in Apache HBase, usually when something gets screwed up what will follow may be hundreds of exceptions and stack traces coming from all over the place. The best way to approach this type of problem is to walk the log up to where it all began, for example one trick with RegionServers is that they will print some metrics when aborting so grepping for <em>Dump</em> should get you around the start of the problem.</p> </div> <div class="paragraph"> <p>RegionServer suicides are 'normal', as this is what they do when something goes wrong. For example, if ulimit and max transfer threads (the two most important initial settings, see <a href="#ulimit">[ulimit]</a> and <a href="#dfs.datanode.max.transfer.threads"><code>dfs.datanode.max.transfer.threads</code> </a>) aren’t changed, it will make it impossible at some point for DataNodes to create new threads that from the HBase point of view is seen as if HDFS was gone. Think about what would happen if your MySQL database was suddenly unable to access files on your local file system, well it’s the same with HBase and HDFS. Another very common reason to see RegionServers committing seppuku is when they enter prolonged garbage collection pauses that last longer than the default ZooKeeper session timeout. For more information on GC pauses, see the <a href="http://www.cloudera.com/blog/2011/02/avoiding-full-gcs-in-hbase-with-memstore-local-allocation-buffers-part-1/">3 part blog post</a> by Todd Lipcon and <a href="#gcpause">Long GC pauses</a> above.</p> </div> </div> </div> <div class="sect1"> <h2 id="trouble.log"><a class="anchor" href="#trouble.log"></a>103. Logs</h2> <div class="sectionbody"> <div class="paragraph"> <p>The key process logs are as follows… (replace <user> with the user that started the service, and <hostname> for the machine name)</p> </div> <div class="paragraph"> <p>NameNode: <em>$HADOOP_HOME/logs/hadoop-<user>-namenode-<hostname>.log</em></p> </div> <div class="paragraph"> <p>DataNode: <em>$HADOOP_HOME/logs/hadoop-<user>-datanode-<hostname>.log</em></p> </div> <div class="paragraph"> <p>JobTracker: <em>$HADOOP_HOME/logs/hadoop-<user>-jobtracker-<hostname>.log</em></p> </div> <div class="paragraph"> <p>TaskTracker: <em>$HADOOP_HOME/logs/hadoop-<user>-tasktracker-<hostname>.log</em></p> </div> <div class="paragraph"> <p>HMaster: <em>$HBASE_HOME/logs/hbase-<user>-master-<hostname>.log</em></p> </div> <div class="paragraph"> <p>RegionServer: <em>$HBASE_HOME/logs/hbase-<user>-regionserver-<hostname>.log</em></p> </div> <div class="paragraph"> <p>ZooKeeper: <em>TODO</em></p> </div> <div class="sect2"> <h3 id="trouble.log.locations"><a class="anchor" href="#trouble.log.locations"></a>103.1. Log Locations</h3> <div class="paragraph"> <p>For stand-alone deployments the logs are obviously going to be on a single machine, however this is a development configuration only. Production deployments need to run on a cluster.</p> </div> <div class="sect3"> <h4 id="trouble.log.locations.namenode"><a class="anchor" href="#trouble.log.locations.namenode"></a>103.1.1. NameNode</h4> <div class="paragraph"> <p>The NameNode log is on the NameNode server. The HBase Master is typically run on the NameNode server, and well as ZooKeeper.</p> </div> <div class="paragraph"> <p>For smaller clusters the JobTracker/ResourceManager is typically run on the NameNode server as well.</p> </div> </div> <div class="sect3"> <h4 id="trouble.log.locations.datanode"><a class="anchor" href="#trouble.log.locations.datanode"></a>103.1.2. DataNode</h4> <div class="paragraph"> <p>Each DataNode server will have a DataNode log for HDFS, as well as a RegionServer log for HBase.</p> </div> <div class="paragraph"> <p>Additionally, each DataNode server will also have a TaskTracker/NodeManager log for MapReduce task execution.</p> </div> </div> </div> <div class="sect2"> <h3 id="trouble.log.levels"><a class="anchor" href="#trouble.log.levels"></a>103.2. Log Levels</h3> <div class="sect3"> <h4 id="rpc.logging"><a class="anchor" href="#rpc.logging"></a>103.2.1. Enabling RPC-level logging</h4> <div class="paragraph"> <p>Enabling the RPC-level logging on a RegionServer can often given insight on timings at the server. Once enabled, the amount of log spewed is voluminous. It is not recommended that you leave this logging on for more than short bursts of time. To enable RPC-level logging, browse to the RegionServer UI and click on <em>Log Level</em>. Set the log level to <code>DEBUG</code> for the package <code>org.apache.hadoop.ipc</code> (Thats right, for <code>hadoop.ipc</code>, NOT, <code>hbase.ipc</code>). Then tail the RegionServers log. Analyze.</p> </div> <div class="paragraph"> <p>To disable, set the logging level back to <code>INFO</code> level.</p> </div> </div> </div> <div class="sect2"> <h3 id="trouble.log.gc"><a class="anchor" href="#trouble.log.gc"></a>103.3. JVM Garbage Collection Logs</h3> <div class="paragraph"> <p>HBase is memory intensive, and using the default GC you can see long pauses in all threads including the <em>Juliet Pause</em> aka "GC of Death". To help debug this or confirm this is happening GC logging can be turned on in the Java virtual machine.</p> </div> <div class="paragraph"> <p>To enable, in <em>hbase-env.sh</em>, uncomment one of the below lines :</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"># This enables basic gc logging to the .out file. # export SERVER_GC_OPTS="-verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps" # This enables basic gc logging to its own file. # export SERVER_GC_OPTS="-verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps -Xloggc:<FILE-PATH>" # This enables basic GC logging to its own file with automatic log rolling. Only applies to jdk 1.6.0_34+ and 1.7.0_2+. # export SERVER_GC_OPTS="-verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps -Xloggc:<FILE-PATH> -XX:+UseGCLogFileRotation -XX:NumberOfGCLogFiles=1 -XX:GCLogFileSize=512M" # If <FILE-PATH> is not replaced, the log file(.gc) would be generated in the HBASE_LOG_DIR.</code></pre> </div> </div> <div class="paragraph"> <p>At this point you should see logs like so:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="float">64898.952</span>: [GC [<span class="integer">1</span> CMS-initial-mark: <span class="integer">2811538</span>K(<span class="integer">3055704</span>K)] <span class="integer">2812179</span>K(<span class="integer">3061272</span>K), <span class="float">0.0007360</span> secs] [Times: user=<span class="float">0.00</span> sys=<span class="float">0.00</span>, real=<span class="float">0.00</span> secs] <span class="float">64898.953</span>: [CMS-concurrent-mark-start] <span class="float">64898.971</span>: [GC <span class="float">64898.971</span>: [ParNew: <span class="integer">5567</span>K-><span class="integer">576</span>K(<span class="integer">5568</span>K), <span class="float">0.0101110</span> secs] <span class="integer">2817105</span>K-><span class="integer">2812715</span>K(<span class="integer">3061272</span>K), <span class="float">0.0102200</span> secs] [Times: user=<span class="float">0.07</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs]</code></pre> </div> </div> <div class="paragraph"> <p>In this section, the first line indicates a 0.0007360 second pause for the CMS to initially mark. This pauses the entire VM, all threads for that period of time.</p> </div> <div class="paragraph"> <p>The third line indicates a "minor GC", which pauses the VM for 0.0101110 seconds - aka 10 milliseconds. It has reduced the "ParNew" from about 5.5m to 576k. Later on in this cycle we see:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="float">64901.445</span>: [CMS-concurrent-mark: <span class="float">1.542</span>/<span class="float">2.492</span> secs] [Times: user=<span class="float">10.49</span> sys=<span class="float">0.33</span>, real=<span class="float">2.49</span> secs] <span class="float">64901.445</span>: [CMS-concurrent-preclean-start] <span class="float">64901.453</span>: [GC <span class="float">64901.453</span>: [ParNew: <span class="integer">5505</span>K-><span class="integer">573</span>K(<span class="integer">5568</span>K), <span class="float">0.0062440</span> secs] <span class="integer">2868746</span>K-><span class="integer">2864292</span>K(<span class="integer">3061272</span>K), <span class="float">0.0063360</span> secs] [Times: user=<span class="float">0.05</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.476</span>: [GC <span class="float">64901.476</span>: [ParNew: <span class="integer">5563</span>K-><span class="integer">575</span>K(<span class="integer">5568</span>K), <span class="float">0.0072510</span> secs] <span class="integer">2869283</span>K-><span class="integer">2864837</span>K(<span class="integer">3061272</span>K), <span class="float">0.0073320</span> secs] [Times: user=<span class="float">0.05</span> sys=<span class="float">0.01</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.500</span>: [GC <span class="float">64901.500</span>: [ParNew: <span class="integer">5517</span>K-><span class="integer">573</span>K(<span class="integer">5568</span>K), <span class="float">0.0120390</span> secs] <span class="integer">2869780</span>K-><span class="integer">2865267</span>K(<span class="integer">3061272</span>K), <span class="float">0.0121150</span> secs] [Times: user=<span class="float">0.09</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.529</span>: [GC <span class="float">64901.529</span>: [ParNew: <span class="integer">5507</span>K-><span class="integer">569</span>K(<span class="integer">5568</span>K), <span class="float">0.0086240</span> secs] <span class="integer">2870200</span>K-><span class="integer">2865742</span>K(<span class="integer">3061272</span>K), <span class="float">0.0087180</span> secs] [Times: user=<span class="float">0.05</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.554</span>: [GC <span class="float">64901.555</span>: [ParNew: <span class="integer">5516</span>K-><span class="integer">575</span>K(<span class="integer">5568</span>K), <span class="float">0.0107130</span> secs] <span class="integer">2870689</span>K-><span class="integer">2866291</span>K(<span class="integer">3061272</span>K), <span class="float">0.0107820</span> secs] [Times: user=<span class="float">0.06</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.578</span>: [CMS-concurrent-preclean: <span class="float">0.070</span>/<span class="float">0.133</span> secs] [Times: user=<span class="float">0.48</span> sys=<span class="float">0.01</span>, real=<span class="float">0.14</span> secs] <span class="float">64901.578</span>: [CMS-concurrent-abortable-preclean-start] <span class="float">64901.584</span>: [GC <span class="float">64901.584</span>: [ParNew: <span class="integer">5504</span>K-><span class="integer">571</span>K(<span class="integer">5568</span>K), <span class="float">0.0087270</span> secs] <span class="integer">2871220</span>K-><span class="integer">2866830</span>K(<span class="integer">3061272</span>K), <span class="float">0.0088220</span> secs] [Times: user=<span class="float">0.05</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.609</span>: [GC <span class="float">64901.609</span>: [ParNew: <span class="integer">5512</span>K-><span class="integer">569</span>K(<span class="integer">5568</span>K), <span class="float">0.0063370</span> secs] <span class="integer">2871771</span>K-><span class="integer">2867322</span>K(<span class="integer">3061272</span>K), <span class="float">0.0064230</span> secs] [Times: user=<span class="float">0.06</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.615</span>: [CMS-concurrent-abortable-preclean: <span class="float">0.007</span>/<span class="float">0.037</span> secs] [Times: user=<span class="float">0.13</span> sys=<span class="float">0.00</span>, real=<span class="float">0.03</span> secs] <span class="float">64901.616</span>: [GC[YG occupancy: <span class="integer">645</span> K (<span class="integer">5568</span> K)]<span class="float">64901.616</span>: [Rescan (parallel) , <span class="float">0.0020210</span> secs]<span class="float">64901.618</span>: [weak refs processing, <span class="float">0.0027950</span> secs] [<span class="integer">1</span> CMS-remark: <span class="integer">2866753</span>K(<span class="integer">3055704</span>K)] <span class="integer">2867399</span>K(<span class="integer">3061272</span>K), <span class="float">0.0049380</span> secs] [Times: user=<span class="float">0.00</span> sys=<span class="float">0.01</span>, real=<span class="float">0.01</span> secs] <span class="float">64901.621</span>: [CMS-concurrent-sweep-start]</code></pre> </div> </div> <div class="paragraph"> <p>The first line indicates that the CMS concurrent mark (finding garbage) has taken 2.4 seconds. But this is a <em>concurrent</em> 2.4 seconds, Java has not been paused at any point in time.</p> </div> <div class="paragraph"> <p>There are a few more minor GCs, then there is a pause at the 2nd last line:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="float">64901.616</span>: [GC[YG occupancy: <span class="integer">645</span> K (<span class="integer">5568</span> K)]<span class="float">64901.616</span>: [Rescan (parallel) , <span class="float">0.0020210</span> secs]<span class="float">64901.618</span>: [weak refs processing, <span class="float">0.0027950</span> secs] [<span class="integer">1</span> CMS-remark: <span class="integer">2866753</span>K(<span class="integer">3055704</span>K)] <span class="integer">2867399</span>K(<span class="integer">3061272</span>K), <span class="float">0.0049380</span> secs] [Times: user=<span class="float">0.00</span> sys=<span class="float">0.01</span>, real=<span class="float">0.01</span> secs]</code></pre> </div> </div> <div class="paragraph"> <p>The pause here is 0.0049380 seconds (aka 4.9 milliseconds) to 'remark' the heap.</p> </div> <div class="paragraph"> <p>At this point the sweep starts, and you can watch the heap size go down:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="float">64901.637</span>: [GC <span class="float">64901.637</span>: [ParNew: <span class="integer">5501</span>K-><span class="integer">569</span>K(<span class="integer">5568</span>K), <span class="float">0.0097350</span> secs] <span class="integer">2871958</span>K-><span class="integer">2867441</span>K(<span class="integer">3061272</span>K), <span class="float">0.0098370</span> secs] [Times: user=<span class="float">0.05</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] ... lines removed ... <span class="float">64904.936</span>: [GC <span class="float">64904.936</span>: [ParNew: <span class="integer">5532</span>K-><span class="integer">568</span>K(<span class="integer">5568</span>K), <span class="float">0.0070720</span> secs] <span class="integer">1365024</span>K-><span class="integer">1360689</span>K(<span class="integer">3061272</span>K), <span class="float">0.0071930</span> secs] [Times: user=<span class="float">0.05</span> sys=<span class="float">0.00</span>, real=<span class="float">0.01</span> secs] <span class="float">64904.953</span>: [CMS-concurrent-sweep: <span class="float">2.030</span>/<span class="float">3.332</span> secs] [Times: user=<span class="float">9.57</span> sys=<span class="float">0.26</span>, real=<span class="float">3.33</span> secs]</code></pre> </div> </div> <div class="paragraph"> <p>At this point, the CMS sweep took 3.332 seconds, and heap went from about ~ 2.8 GB to 1.3 GB (approximate).</p> </div> <div class="paragraph"> <p>The key points here is to keep all these pauses low. CMS pauses are always low, but if your ParNew starts growing, you can see minor GC pauses approach 100ms, exceed 100ms and hit as high at 400ms.</p> </div> <div class="paragraph"> <p>This can be due to the size of the ParNew, which should be relatively small. If your ParNew is very large after running HBase for a while, in one example a ParNew was about 150MB, then you might have to constrain the size of ParNew (The larger it is, the longer the collections take but if its too small, objects are promoted to old gen too quickly). In the below we constrain new gen size to 64m.</p> </div> <div class="paragraph"> <p>Add the below line in <em>hbase-env.sh</em>:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">export SERVER_GC_OPTS="$SERVER_GC_OPTS -XX:NewSize=64m -XX:MaxNewSize=64m"</code></pre> </div> </div> <div class="paragraph"> <p>Similarly, to enable GC logging for client processes, uncomment one of the below lines in <em>hbase-env.sh</em>:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"># This enables basic gc logging to the .out file. # export CLIENT_GC_OPTS="-verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps" # This enables basic gc logging to its own file. # export CLIENT_GC_OPTS="-verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps -Xloggc:<FILE-PATH>" # This enables basic GC logging to its own file with automatic log rolling. Only applies to jdk 1.6.0_34+ and 1.7.0_2+. # export CLIENT_GC_OPTS="-verbose:gc -XX:+PrintGCDetails -XX:+PrintGCDateStamps -Xloggc:<FILE-PATH> -XX:+UseGCLogFileRotation -XX:NumberOfGCLogFiles=1 -XX:GCLogFileSize=512M" # If <FILE-PATH> is not replaced, the log file(.gc) would be generated in the HBASE_LOG_DIR .</code></pre> </div> </div> <div class="paragraph"> <p>For more information on GC pauses, see the <a href="http://www.cloudera.com/blog/2011/02/avoiding-full-gcs-in-hbase-with-memstore-local-allocation-buffers-part-1/">3 part blog post</a> by Todd Lipcon and <a href="#gcpause">Long GC pauses</a> above.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.resources"><a class="anchor" href="#trouble.resources"></a>104. Resources</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.resources.searchhadoop"><a class="anchor" href="#trouble.resources.searchhadoop"></a>104.1. search-hadoop.com</h3> <div class="paragraph"> <p><a href="http://search-hadoop.com">search-hadoop.com</a> indexes all the mailing lists and is great for historical searches. Search here first when you have an issue as its more than likely someone has already had your problem.</p> </div> </div> <div class="sect2"> <h3 id="trouble.resources.lists"><a class="anchor" href="#trouble.resources.lists"></a>104.2. Mailing Lists</h3> <div class="paragraph"> <p>Ask a question on the <a href="http://hbase.apache.org/mail-lists.html">Apache HBase mailing lists</a>. The 'dev' mailing list is aimed at the community of developers actually building Apache HBase and for features currently under development, and 'user' is generally used for questions on released versions of Apache HBase. Before going to the mailing list, make sure your question has not already been answered by searching the mailing list archives first. Use <a href="#trouble.resources.searchhadoop">search-hadoop.com</a>. Take some time crafting your question. See <a href="http://www.mikeash.com/getting_answers.html">Getting Answers</a> for ideas on crafting good questions. A quality question that includes all context and exhibits evidence the author has tried to find answers in the manual and out on lists is more likely to get a prompt response.</p> </div> </div> <div class="sect2"> <h3 id="trouble.resources.irc"><a class="anchor" href="#trouble.resources.irc"></a>104.3. IRC</h3> <div class="paragraph"> <p>#hbase on irc.freenode.net</p> </div> </div> <div class="sect2"> <h3 id="trouble.resources.jira"><a class="anchor" href="#trouble.resources.jira"></a>104.4. JIRA</h3> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE">JIRA</a> is also really helpful when looking for Hadoop/HBase-specific issues.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.tools"><a class="anchor" href="#trouble.tools"></a>105. Tools</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.tools.builtin"><a class="anchor" href="#trouble.tools.builtin"></a>105.1. Builtin Tools</h3> <div class="sect3"> <h4 id="trouble.tools.builtin.webmaster"><a class="anchor" href="#trouble.tools.builtin.webmaster"></a>105.1.1. Master Web Interface</h4> <div class="paragraph"> <p>The Master starts a web-interface on port 16010 by default. (Up to and including 0.98 this was port 60010)</p> </div> <div class="paragraph"> <p>The Master web UI lists created tables and their definition (e.g., ColumnFamilies, blocksize, etc.). Additionally, the available RegionServers in the cluster are listed along with selected high-level metrics (requests, number of regions, usedHeap, maxHeap). The Master web UI allows navigation to each RegionServer’s web UI.</p> </div> </div> <div class="sect3"> <h4 id="trouble.tools.builtin.webregion"><a class="anchor" href="#trouble.tools.builtin.webregion"></a>105.1.2. RegionServer Web Interface</h4> <div class="paragraph"> <p>RegionServers starts a web-interface on port 16030 by default. (Up to an including 0.98 this was port 60030)</p> </div> <div class="paragraph"> <p>The RegionServer web UI lists online regions and their start/end keys, as well as point-in-time RegionServer metrics (requests, regions, storeFileIndexSize, compactionQueueSize, etc.).</p> </div> <div class="paragraph"> <p>See <a href="#hbase_metrics">[hbase_metrics]</a> for more information in metric definitions.</p> </div> </div> <div class="sect3"> <h4 id="trouble.tools.builtin.zkcli"><a class="anchor" href="#trouble.tools.builtin.zkcli"></a>105.1.3. zkcli</h4> <div class="paragraph"> <p><code>zkcli</code> is a very useful tool for investigating ZooKeeper-related issues. To invoke:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">./hbase zkcli -server host:port <cmd> <args></code></pre> </div> </div> <div class="paragraph"> <p>The commands (and arguments) are:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> connect host:port get path [watch] ls path [watch] set path data [version] delquota [-n|-b] path quit printwatches on|off create [-s] [-e] path data acl stat path [watch] close ls2 path [watch] history listquota path setAcl path acl getAcl path sync path redo cmdno addauth scheme auth delete path [version] setquota -n|-b val path</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="trouble.tools.external"><a class="anchor" href="#trouble.tools.external"></a>105.2. External Tools</h3> <div class="sect3"> <h4 id="trouble.tools.tail"><a class="anchor" href="#trouble.tools.tail"></a>105.2.1. tail</h4> <div class="paragraph"> <p><code>tail</code> is the command line tool that lets you look at the end of a file. Add the <code>-f</code> option and it will refresh when new data is available. It’s useful when you are wondering what’s happening, for example, when a cluster is taking a long time to shutdown or startup as you can just fire a new terminal and tail the master log (and maybe a few RegionServers).</p> </div> </div> <div class="sect3"> <h4 id="trouble.tools.top"><a class="anchor" href="#trouble.tools.top"></a>105.2.2. top</h4> <div class="paragraph"> <p><code>top</code> is probably one of the most important tools when first trying to see what’s running on a machine and how the resources are consumed. Here’s an example from production system:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">top - <span class="integer">14</span>:<span class="integer">46</span>:<span class="integer">59</span> up <span class="integer">39</span> days, <span class="integer">11</span>:<span class="integer">55</span>, <span class="integer">1</span> user, load average: <span class="float">3.75</span>, <span class="float">3.57</span>, <span class="float">3.84</span> Tasks: <span class="integer">309</span> total, <span class="integer">1</span> running, <span class="integer">308</span> sleeping, <span class="integer">0</span> stopped, <span class="integer">0</span> zombie Cpu(s): <span class="float">4.5</span>%us, <span class="float">1.6</span>%sy, <span class="float">0.0</span>%ni, <span class="float">91.7</span>%id, <span class="float">1.4</span>%wa, <span class="float">0.1</span>%hi, <span class="float">0.6</span>%si, <span class="float">0.0</span>%st Mem: <span class="integer">24414432</span>k total, <span class="integer">24296956</span>k used, <span class="integer">117476</span>k free, <span class="integer">7196</span>k buffers Swap: <span class="integer">16008732</span>k total, <span class="integer">14348</span>k used, <span class="integer">15994384</span>k free, <span class="integer">11106908</span>k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND <span class="integer">15558</span> hadoop <span class="integer">18</span> -<span class="integer">2</span> <span class="integer">3292</span>m <span class="float">2.4</span>g <span class="integer">3556</span> S <span class="integer">79</span> <span class="float">10.4</span> <span class="integer">6523</span>:<span class="integer">52</span> java <span class="integer">13268</span> hadoop <span class="integer">18</span> -<span class="integer">2</span> <span class="integer">8967</span>m <span class="float">8.2</span>g <span class="integer">4104</span> S <span class="integer">21</span> <span class="float">35.1</span> <span class="integer">5170</span>:<span class="integer">30</span> java <span class="integer">8895</span> hadoop <span class="integer">18</span> -<span class="integer">2</span> <span class="integer">1581</span>m <span class="integer">497</span>m <span class="integer">3420</span> S <span class="integer">11</span> <span class="float">2.1</span> <span class="integer">4002</span>:<span class="integer">32</span> java <span class="error">…</span></code></pre> </div> </div> <div class="paragraph"> <p>Here we can see that the system load average during the last five minutes is 3.75, which very roughly means that on average 3.75 threads were waiting for CPU time during these 5 minutes. In general, the <em>perfect</em> utilization equals to the number of cores, under that number the machine is under utilized and over that the machine is over utilized. This is an important concept, see this article to understand it more: <a href="http://www.linuxjournal.com/article/9001" class="bare">http://www.linuxjournal.com/article/9001</a>.</p> </div> <div class="paragraph"> <p>Apart from load, we can see that the system is using almost all its available RAM but most of it is used for the OS cache (which is good). The swap only has a few KBs in it and this is wanted, high numbers would indicate swapping activity which is the nemesis of performance of Java systems. Another way to detect swapping is when the load average goes through the roof (although this could also be caused by things like a dying disk, among others).</p> </div> <div class="paragraph"> <p>The list of processes isn’t super useful by default, all we know is that 3 java processes are using about 111% of the CPUs. To know which is which, simply type <code>c</code> and each line will be expanded. Typing <code>1</code> will give you the detail of how each CPU is used instead of the average for all of them like shown here.</p> </div> </div> <div class="sect3"> <h4 id="trouble.tools.jps"><a class="anchor" href="#trouble.tools.jps"></a>105.2.3. jps</h4> <div class="paragraph"> <p><code>jps</code> is shipped with every JDK and gives the java process ids for the current user (if root, then it gives the ids for all users). Example:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">hadoop@sv4borg12:~$ jps 1322 TaskTracker 17789 HRegionServer 27862 Child 1158 DataNode 25115 HQuorumPeer 2950 Jps 19750 ThriftServer 18776 jmx</code></pre> </div> </div> <div class="paragraph"> <p>In order, we see a:</p> </div> <div class="ulist"> <ul> <li> <p>Hadoop TaskTracker, manages the local Childs</p> </li> <li> <p>HBase RegionServer, serves regions</p> </li> <li> <p>Child, its MapReduce task, cannot tell which type exactly</p> </li> <li> <p>Hadoop TaskTracker, manages the local Childs</p> </li> <li> <p>Hadoop DataNode, serves blocks</p> </li> <li> <p>HQuorumPeer, a ZooKeeper ensemble member</p> </li> <li> <p>Jps, well… it’s the current process</p> </li> <li> <p>ThriftServer, it’s a special one will be running only if thrift was started</p> </li> <li> <p>jmx, this is a local process that’s part of our monitoring platform ( poorly named maybe). You probably don’t have that.</p> </li> </ul> </div> <div class="paragraph"> <p>You can then do stuff like checking out the full command line that started the process:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">hadoop@sv4borg12:~$ ps aux | grep HRegionServer hadoop 17789 155 35.2 9067824 8604364 ? S&lt;l Mar04 9855:48 /usr/java/jdk1.6.0_14/bin/java -Xmx8000m -XX:+DoEscapeAnalysis -XX:+AggressiveOpts -XX:+UseConcMarkSweepGC -XX:NewSize=64m -XX:MaxNewSize=64m -XX:CMSInitiatingOccupancyFraction=88 -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -Xloggc:/export1/hadoop/logs/gc-hbase.log -Dcom.sun.management.jmxremote.port=10102 -Dcom.sun.management.jmxremote.authenticate=true -Dcom.sun.management.jmxremote.ssl=false -Dcom.sun.management.jmxremote.password.file=/home/hadoop/hbase/conf/jmxremote.password -Dcom.sun.management.jmxremote -Dhbase.log.dir=/export1/hadoop/logs -Dhbase.log.file=hbase-hadoop-regionserver-sv4borg12.log -Dhbase.home.dir=/home/hadoop/hbase -Dhbase.id.str=hadoop -Dhbase.root.logger=INFO,DRFA -Djava.library.path=/home/hadoop/hbase/lib/native/Linux-amd64-64 -classpath /home/hadoop/hbase/bin/../conf:[many jars]:/home/hadoop/hadoop/conf org.apache.hadoop.hbase.regionserver.HRegionServer start</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="trouble.tools.jstack"><a class="anchor" href="#trouble.tools.jstack"></a>105.2.4. jstack</h4> <div class="paragraph"> <p><code>jstack</code> is one of the most important tools when trying to figure out what a java process is doing apart from looking at the logs. It has to be used in conjunction with jps in order to give it a process id. It shows a list of threads, each one has a name, and they appear in the order that they were created (so the top ones are the most recent threads). Here are a few example:</p> </div> <div class="paragraph"> <p>The main thread of a RegionServer waiting for something to do from the master:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">regionserver60020</span><span class="delimiter">"</span></span> prio=<span class="integer">10</span> tid=<span class="hex">0x0000000040ab4000</span> nid=<span class="hex">0x45cf</span> waiting on condition [<span class="hex">0x00007f16b6a96000</span>.<span class="float">.0</span>x00007f16b6a96a70] java.lang.Thread.State: TIMED_WAITING (parking) at sun.misc.Unsafe.park(Native <span class="predefined-type">Method</span>) - parking to wait <span class="keyword">for</span> <<span class="hex">0x00007f16cd5c2f30</span>> (a java.util.concurrent.locks.AbstractQueuedSynchronizer<span class="error">$</span>ConditionObject) at java.util.concurrent.locks.LockSupport.parkNanos(<span class="predefined-type">LockSupport</span>.java:<span class="integer">198</span>) at java.util.concurrent.locks.AbstractQueuedSynchronizer<span class="error">$</span>ConditionObject.awaitNanos(<span class="predefined-type">AbstractQueuedSynchronizer</span>.java:<span class="integer">1963</span>) at java.util.concurrent.LinkedBlockingQueue.poll(<span class="predefined-type">LinkedBlockingQueue</span>.java:<span class="integer">395</span>) at org.apache.hadoop.hbase.regionserver.HRegionServer.run(HRegionServer.java:<span class="integer">647</span>) at java.lang.Thread.run(<span class="predefined-type">Thread</span>.java:<span class="integer">619</span>)</code></pre> </div> </div> <div class="paragraph"> <p>The MemStore flusher thread that is currently flushing to a file:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">regionserver60020.cacheFlusher</span><span class="delimiter">"</span></span> daemon prio=<span class="integer">10</span> tid=<span class="hex">0x0000000040f4e000</span> nid=<span class="hex">0x45eb</span> in <span class="predefined-type">Object</span>.wait() [<span class="hex">0x00007f16b5b86000</span>.<span class="float">.0</span>x00007f16b5b87af0] java.lang.Thread.State: WAITING (on object monitor) at java.lang.Object.wait(Native <span class="predefined-type">Method</span>) at java.lang.Object.wait(<span class="predefined-type">Object</span>.java:<span class="integer">485</span>) at org.apache.hadoop.ipc.Client.call(Client.java:<span class="integer">803</span>) - locked <<span class="hex">0x00007f16cb14b3a8</span>> (a org.apache.hadoop.ipc.Client<span class="error">$</span>Call) at org.apache.hadoop.ipc.RPC<span class="error">$</span>Invoker.invoke(RPC.java:<span class="integer">221</span>) at <span class="error">$</span>Proxy1.complete(Unknown <span class="predefined-type">Source</span>) at sun.reflect.GeneratedMethodAccessor38.invoke(Unknown <span class="predefined-type">Source</span>) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:<span class="integer">25</span>) at java.lang.reflect.Method.invoke(<span class="predefined-type">Method</span>.java:<span class="integer">597</span>) at org.apache.hadoop.io.retry.RetryInvocationHandler.invokeMethod(RetryInvocationHandler.java:<span class="integer">82</span>) at org.apache.hadoop.io.retry.RetryInvocationHandler.invoke(RetryInvocationHandler.java:<span class="integer">59</span>) at <span class="error">$</span>Proxy1.complete(Unknown <span class="predefined-type">Source</span>) at org.apache.hadoop.hdfs.DFSClient<span class="error">$</span>DFSOutputStream.closeInternal(DFSClient.java:<span class="integer">3390</span>) - locked <<span class="hex">0x00007f16cb14b470</span>> (a org.apache.hadoop.hdfs.DFSClient<span class="error">$</span>DFSOutputStream) at org.apache.hadoop.hdfs.DFSClient<span class="error">$</span>DFSOutputStream.close(DFSClient.java:<span class="integer">3304</span>) at org.apache.hadoop.fs.FSDataOutputStream<span class="error">$</span>PositionCache.close(FSDataOutputStream.java:<span class="integer">61</span>) at org.apache.hadoop.fs.FSDataOutputStream.close(FSDataOutputStream.java:<span class="integer">86</span>) at org.apache.hadoop.hbase.io.hfile.HFile<span class="error">$</span><span class="predefined-type">Writer</span>.close(HFile.java:<span class="integer">650</span>) at org.apache.hadoop.hbase.regionserver.StoreFile<span class="error">$</span><span class="predefined-type">Writer</span>.close(StoreFile.java:<span class="integer">853</span>) at org.apache.hadoop.hbase.regionserver.Store.internalFlushCache(Store.java:<span class="integer">467</span>) - locked <<span class="hex">0x00007f16d00e6f08</span>> (a java.lang.Object) at org.apache.hadoop.hbase.regionserver.Store.flushCache(Store.java:<span class="integer">427</span>) at org.apache.hadoop.hbase.regionserver.Store.access<span class="error">$</span><span class="integer">100</span>(Store.java:<span class="integer">80</span>) at org.apache.hadoop.hbase.regionserver.Store<span class="error">$</span>StoreFlusherImpl.flushCache(Store.java:<span class="integer">1359</span>) at org.apache.hadoop.hbase.regionserver.HRegion.internalFlushcache(HRegion.java:<span class="integer">907</span>) at org.apache.hadoop.hbase.regionserver.HRegion.internalFlushcache(HRegion.java:<span class="integer">834</span>) at org.apache.hadoop.hbase.regionserver.HRegion.flushcache(HRegion.java:<span class="integer">786</span>) at org.apache.hadoop.hbase.regionserver.MemStoreFlusher.flushRegion(MemStoreFlusher.java:<span class="integer">250</span>) at org.apache.hadoop.hbase.regionserver.MemStoreFlusher.flushRegion(MemStoreFlusher.java:<span class="integer">224</span>) at org.apache.hadoop.hbase.regionserver.MemStoreFlusher.run(MemStoreFlusher.java:<span class="integer">146</span>)</code></pre> </div> </div> <div class="paragraph"> <p>A handler thread that’s waiting for stuff to do (like put, delete, scan, etc):</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">IPC Server handler 16 on 60020</span><span class="delimiter">"</span></span> daemon prio=<span class="integer">10</span> tid=<span class="hex">0x00007f16b011d800</span> nid=<span class="hex">0x4a5e</span> waiting on condition [<span class="hex">0x00007f16afefd000</span>.<span class="float">.0</span>x00007f16afefd9f0] java.lang.Thread.State: WAITING (parking) at sun.misc.Unsafe.park(Native <span class="predefined-type">Method</span>) - parking to wait <span class="keyword">for</span> <<span class="hex">0x00007f16cd3f8dd8</span>> (a java.util.concurrent.locks.AbstractQueuedSynchronizer<span class="error">$</span>ConditionObject) at java.util.concurrent.locks.LockSupport.park(<span class="predefined-type">LockSupport</span>.java:<span class="integer">158</span>) at java.util.concurrent.locks.AbstractQueuedSynchronizer<span class="error">$</span>ConditionObject.await(<span class="predefined-type">AbstractQueuedSynchronizer</span>.java:<span class="integer">1925</span>) at java.util.concurrent.LinkedBlockingQueue.take(<span class="predefined-type">LinkedBlockingQueue</span>.java:<span class="integer">358</span>) at org.apache.hadoop.hbase.ipc.HBaseServer<span class="error">$</span><span class="predefined-type">Handler</span>.run(HBaseServer.java:<span class="integer">1013</span>)</code></pre> </div> </div> <div class="paragraph"> <p>And one that’s busy doing an increment of a counter (it’s in the phase where it’s trying to create a scanner in order to read the last value):</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">IPC Server handler 66 on 60020</span><span class="delimiter">"</span></span> daemon prio=<span class="integer">10</span> tid=<span class="hex">0x00007f16b006e800</span> nid=<span class="hex">0x4a90</span> runnable [<span class="hex">0x00007f16acb77000</span>.<span class="float">.0</span>x00007f16acb77cf0] java.lang.Thread.State: RUNNABLE at org.apache.hadoop.hbase.regionserver.KeyValueHeap.<init>(KeyValueHeap.java:<span class="integer">56</span>) at org.apache.hadoop.hbase.regionserver.StoreScanner.<init>(StoreScanner.java:<span class="integer">79</span>) at org.apache.hadoop.hbase.regionserver.Store.getScanner(Store.java:<span class="integer">1202</span>) at org.apache.hadoop.hbase.regionserver.HRegion<span class="error">$</span>RegionScanner.<init>(HRegion.java:<span class="integer">2209</span>) at org.apache.hadoop.hbase.regionserver.HRegion.instantiateInternalScanner(HRegion.java:<span class="integer">1063</span>) at org.apache.hadoop.hbase.regionserver.HRegion.getScanner(HRegion.java:<span class="integer">1055</span>) at org.apache.hadoop.hbase.regionserver.HRegion.getScanner(HRegion.java:<span class="integer">1039</span>) at org.apache.hadoop.hbase.regionserver.HRegion.getLastIncrement(HRegion.java:<span class="integer">2875</span>) at org.apache.hadoop.hbase.regionserver.HRegion.incrementColumnValue(HRegion.java:<span class="integer">2978</span>) at org.apache.hadoop.hbase.regionserver.HRegionServer.incrementColumnValue(HRegionServer.java:<span class="integer">2433</span>) at sun.reflect.GeneratedMethodAccessor20.invoke(Unknown <span class="predefined-type">Source</span>) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:<span class="integer">25</span>) at java.lang.reflect.Method.invoke(<span class="predefined-type">Method</span>.java:<span class="integer">597</span>) at org.apache.hadoop.hbase.ipc.HBaseRPC<span class="error">$</span>Server.call(HBaseRPC.java:<span class="integer">560</span>) at org.apache.hadoop.hbase.ipc.HBaseServer<span class="error">$</span><span class="predefined-type">Handler</span>.run(HBaseServer.java:<span class="integer">1027</span>)</code></pre> </div> </div> <div class="paragraph"> <p>A thread that receives data from HDFS:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">IPC Client (47) connection to sv4borg9/10.4.24.40:9000 from hadoop</span><span class="delimiter">"</span></span> daemon prio=<span class="integer">10</span> tid=<span class="hex">0x00007f16a02d0000</span> nid=<span class="hex">0x4fa3</span> runnable [<span class="hex">0x00007f16b517d000</span>.<span class="float">.0</span>x00007f16b517dbf0] java.lang.Thread.State: RUNNABLE at sun.nio.ch.EPollArrayWrapper.epollWait(Native <span class="predefined-type">Method</span>) at sun.nio.ch.EPollArrayWrapper.poll(EPollArrayWrapper.java:<span class="integer">215</span>) at sun.nio.ch.EPollSelectorImpl.doSelect(EPollSelectorImpl.java:<span class="integer">65</span>) at sun.nio.ch.SelectorImpl.lockAndDoSelect(SelectorImpl.java:<span class="integer">69</span>) - locked <<span class="hex">0x00007f17d5b68c00</span>> (a sun.nio.ch.Util<span class="error">$</span><span class="integer">1</span>) - locked <<span class="hex">0x00007f17d5b68be8</span>> (a java.util.Collections<span class="error">$</span>UnmodifiableSet) - locked <<span class="hex">0x00007f1877959b50</span>> (a sun.nio.ch.EPollSelectorImpl) at sun.nio.ch.SelectorImpl.select(SelectorImpl.java:<span class="integer">80</span>) at org.apache.hadoop.net.SocketIOWithTimeout<span class="error">$</span>SelectorPool.select(SocketIOWithTimeout.java:<span class="integer">332</span>) at org.apache.hadoop.net.SocketIOWithTimeout.doIO(SocketIOWithTimeout.java:<span class="integer">157</span>) at org.apache.hadoop.net.SocketInputStream.read(SocketInputStream.java:<span class="integer">155</span>) at org.apache.hadoop.net.SocketInputStream.read(SocketInputStream.java:<span class="integer">128</span>) at java.io.FilterInputStream.read(<span class="predefined-type">FilterInputStream</span>.java:<span class="integer">116</span>) at org.apache.hadoop.ipc.Client<span class="error">$</span><span class="predefined-type">Connection</span><span class="error">$</span>PingInputStream.read(Client.java:<span class="integer">304</span>) at java.io.BufferedInputStream.fill(<span class="predefined-type">BufferedInputStream</span>.java:<span class="integer">218</span>) at java.io.BufferedInputStream.read(<span class="predefined-type">BufferedInputStream</span>.java:<span class="integer">237</span>) - locked <<span class="hex">0x00007f1808539178</span>> (a java.io.BufferedInputStream) at java.io.DataInputStream.readInt(<span class="predefined-type">DataInputStream</span>.java:<span class="integer">370</span>) at org.apache.hadoop.ipc.Client<span class="error">$</span><span class="predefined-type">Connection</span>.receiveResponse(Client.java:<span class="integer">569</span>) at org.apache.hadoop.ipc.Client<span class="error">$</span><span class="predefined-type">Connection</span>.run(Client.java:<span class="integer">477</span>)</code></pre> </div> </div> <div class="paragraph"> <p>And here is a master trying to recover a lease after a RegionServer died:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="string"><span class="delimiter">"</span><span class="content">LeaseChecker</span><span class="delimiter">"</span></span> daemon prio=<span class="integer">10</span> tid=<span class="hex">0x00000000407ef800</span> nid=<span class="hex">0x76cd</span> waiting on condition [<span class="hex">0x00007f6d0eae2000</span>.<span class="float">.0</span>x00007f6d0eae2a70] -- java.lang.Thread.State: WAITING (on object monitor) at java.lang.Object.wait(Native <span class="predefined-type">Method</span>) at java.lang.Object.wait(<span class="predefined-type">Object</span>.java:<span class="integer">485</span>) at org.apache.hadoop.ipc.Client.call(Client.java:<span class="integer">726</span>) - locked <<span class="hex">0x00007f6d1cd28f80</span>> (a org.apache.hadoop.ipc.Client<span class="error">$</span>Call) at org.apache.hadoop.ipc.RPC<span class="error">$</span>Invoker.invoke(RPC.java:<span class="integer">220</span>) at <span class="error">$</span>Proxy1.recoverBlock(Unknown <span class="predefined-type">Source</span>) at org.apache.hadoop.hdfs.DFSClient<span class="error">$</span>DFSOutputStream.processDatanodeError(DFSClient.java:<span class="integer">2636</span>) at org.apache.hadoop.hdfs.DFSClient<span class="error">$</span>DFSOutputStream.<init>(DFSClient.java:<span class="integer">2832</span>) at org.apache.hadoop.hdfs.DFSClient.append(DFSClient.java:<span class="integer">529</span>) at org.apache.hadoop.hdfs.DistributedFileSystem.append(DistributedFileSystem.java:<span class="integer">186</span>) at org.apache.hadoop.fs.FileSystem.append(FileSystem.java:<span class="integer">530</span>) at org.apache.hadoop.hbase.util.FSUtils.recoverFileLease(FSUtils.java:<span class="integer">619</span>) at org.apache.hadoop.hbase.regionserver.wal.HLog.splitLog(HLog.java:<span class="integer">1322</span>) at org.apache.hadoop.hbase.regionserver.wal.HLog.splitLog(HLog.java:<span class="integer">1210</span>) at org.apache.hadoop.hbase.master.HMaster.splitLogAfterStartup(HMaster.java:<span class="integer">648</span>) at org.apache.hadoop.hbase.master.HMaster.joinCluster(HMaster.java:<span class="integer">572</span>) at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:<span class="integer">503</span>)</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="trouble.tools.opentsdb"><a class="anchor" href="#trouble.tools.opentsdb"></a>105.2.5. OpenTSDB</h4> <div class="paragraph"> <p><a href="http://opentsdb.net">OpenTSDB</a> is an excellent alternative to Ganglia as it uses Apache HBase to store all the time series and doesn’t have to downsample. Monitoring your own HBase cluster that hosts OpenTSDB is a good exercise.</p> </div> <div class="paragraph"> <p>Here’s an example of a cluster that’s suffering from hundreds of compactions launched almost all around the same time, which severely affects the IO performance: (TODO: insert graph plotting compactionQueueSize)</p> </div> <div class="paragraph"> <p>It’s a good practice to build dashboards with all the important graphs per machine and per cluster so that debugging issues can be done with a single quick look. For example, at StumbleUpon there’s one dashboard per cluster with the most important metrics from both the OS and Apache HBase. You can then go down at the machine level and get even more detailed metrics.</p> </div> </div> <div class="sect3"> <h4 id="trouble.tools.clustersshtop"><a class="anchor" href="#trouble.tools.clustersshtop"></a>105.2.6. clusterssh+top</h4> <div class="paragraph"> <p>clusterssh+top, it’s like a poor man’s monitoring system and it can be quite useful when you have only a few machines as it’s very easy to setup. Starting clusterssh will give you one terminal per machine and another terminal in which whatever you type will be retyped in every window. This means that you can type <code>top</code> once and it will start it for all of your machines at the same time giving you full view of the current state of your cluster. You can also tail all the logs at the same time, edit files, etc.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.client"><a class="anchor" href="#trouble.client"></a>106. Client</h2> <div class="sectionbody"> <div class="paragraph"> <p>For more information on the HBase client, see <a href="#client">client</a>.</p> </div> <div class="sect2"> <h3 id="trouble.client.scantimeout"><a class="anchor" href="#trouble.client.scantimeout"></a>106.1. ScannerTimeoutException or UnknownScannerException</h3> <div class="paragraph"> <p>This is thrown if the time between RPC calls from the client to RegionServer exceeds the scan timeout. For example, if <code>Scan.setCaching</code> is set to 500, then there will be an RPC call to fetch the next batch of rows every 500 <code>.next()</code> calls on the ResultScanner because data is being transferred in blocks of 500 rows to the client. Reducing the setCaching value may be an option, but setting this value too low makes for inefficient processing on numbers of rows.</p> </div> <div class="paragraph"> <p>See <a href="#perf.hbase.client.caching">Scan Caching</a>.</p> </div> </div> <div class="sect2"> <h3 id="_performance_differences_in_thrift_and_java_apis"><a class="anchor" href="#_performance_differences_in_thrift_and_java_apis"></a>106.2. Performance Differences in Thrift and Java APIs</h3> <div class="paragraph"> <p>Poor performance, or even <code>ScannerTimeoutExceptions</code>, can occur if <code>Scan.setCaching</code> is too high, as discussed in <a href="#trouble.client.scantimeout">ScannerTimeoutException or UnknownScannerException</a>. If the Thrift client uses the wrong caching settings for a given workload, performance can suffer compared to the Java API. To set caching for a given scan in the Thrift client, use the <code>scannerGetList(scannerId, numRows)</code> method, where <code>numRows</code> is an integer representing the number of rows to cache. In one case, it was found that reducing the cache for Thrift scans from 1000 to 100 increased performance to near parity with the Java API given the same queries.</p> </div> <div class="paragraph"> <p>See also Jesse Andersen’s <a href="http://blog.cloudera.com/blog/2014/04/how-to-use-the-hbase-thrift-interface-part-3-using-scans/">blog post</a> about using Scans with Thrift.</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.lease.exception"><a class="anchor" href="#trouble.client.lease.exception"></a>106.3. <code>LeaseException</code> when calling <code>Scanner.next</code></h3> <div class="paragraph"> <p>In some situations clients that fetch data from a RegionServer get a LeaseException instead of the usual <a href="#trouble.client.scantimeout">ScannerTimeoutException or UnknownScannerException</a>. Usually the source of the exception is <code>org.apache.hadoop.hbase.regionserver.Leases.removeLease(Leases.java:230)</code> (line number may vary). It tends to happen in the context of a slow/freezing <code>RegionServer#next</code> call. It can be prevented by having <code>hbase.rpc.timeout</code> > <code>hbase.regionserver.lease.period</code>. Harsh J investigated the issue as part of the mailing list thread <a href="http://mail-archives.apache.org/mod_mbox/hbase-user/201209.mbox/%3CCAOcnVr3R-LqtKhFsk8Bhrm-YW2i9O6J6Fhjz2h7q6_sxvwd2yw%40mail.gmail.com%3E">HBase, mail # user - Lease does not exist exceptions</a></p> </div> </div> <div class="sect2"> <h3 id="trouble.client.scarylogs"><a class="anchor" href="#trouble.client.scarylogs"></a>106.4. Shell or client application throws lots of scary exceptions during normal operation</h3> <div class="paragraph"> <p>Since 0.20.0 the default log level for `org.apache.hadoop.hbase.*`is DEBUG.</p> </div> <div class="paragraph"> <p>On your clients, edit <em>$HBASE_HOME/conf/log4j.properties</em> and change this: <code>log4j.logger.org.apache.hadoop.hbase=DEBUG</code> to this: <code>log4j.logger.org.apache.hadoop.hbase=INFO</code>, or even <code>log4j.logger.org.apache.hadoop.hbase=WARN</code>.</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.longpauseswithcompression"><a class="anchor" href="#trouble.client.longpauseswithcompression"></a>106.5. Long Client Pauses With Compression</h3> <div class="paragraph"> <p>This is a fairly frequent question on the Apache HBase dist-list. The scenario is that a client is typically inserting a lot of data into a relatively un-optimized HBase cluster. Compression can exacerbate the pauses, although it is not the source of the problem.</p> </div> <div class="paragraph"> <p>See <a href="#precreate.regions">Table Creation: Pre-Creating Regions</a> on the pattern for pre-creating regions and confirm that the table isn’t starting with a single region.</p> </div> <div class="paragraph"> <p>See <a href="#perf.configurations">HBase Configurations</a> for cluster configuration, particularly <code>hbase.hstore.blockingStoreFiles</code>, <code>hbase.hregion.memstore.block.multiplier</code>, <code>MAX_FILESIZE</code> (region size), and <code>MEMSTORE_FLUSHSIZE.</code></p> </div> <div class="paragraph"> <p>A slightly longer explanation of why pauses can happen is as follows: Puts are sometimes blocked on the MemStores which are blocked by the flusher thread which is blocked because there are too many files to compact because the compactor is given too many small files to compact and has to compact the same data repeatedly. This situation can occur even with minor compactions. Compounding this situation, Apache HBase doesn’t compress data in memory. Thus, the 64MB that lives in the MemStore could become a 6MB file after compression - which results in a smaller StoreFile. The upside is that more data is packed into the same region, but performance is achieved by being able to write larger files - which is why HBase waits until the flushsize before writing a new StoreFile. And smaller StoreFiles become targets for compaction. Without compression the files are much bigger and don’t need as much compaction, however this is at the expense of I/O.</p> </div> <div class="paragraph"> <p>For additional information, see this thread on <a href="http://search-hadoop.com/m/WUnLM6ojHm1/Long+client+pauses+with+compression&subj=Long+client+pauses+with+compression">Long client pauses with compression</a>.</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.security.rpc.krb"><a class="anchor" href="#trouble.client.security.rpc.krb"></a>106.6. Secure Client Connect ([Caused by GSSException: No valid credentials provided…])</h3> <div class="paragraph"> <p>You may encounter the following error:</p> </div> <div class="listingblock"> <div class="content"> <pre>Secure Client Connect ([Caused by GSSException: No valid credentials provided (Mechanism level: Request is a replay (34) V PROCESS_TGS)])</pre> </div> </div> <div class="paragraph"> <p>This issue is caused by bugs in the MIT Kerberos replay_cache component, <a href="http://krbdev.mit.edu/rt/Ticket/Display.html?id=1201">#1201</a> and <a href="http://krbdev.mit.edu/rt/Ticket/Display.html?id=5924">#5924</a>. These bugs caused the old version of krb5-server to erroneously block subsequent requests sent from a Principal. This caused krb5-server to block the connections sent from one Client (one HTable instance with multi-threading connection instances for each RegionServer); Messages, such as <code>Request is a replay (34)</code>, are logged in the client log You can ignore the messages, because HTable will retry 5 * 10 (50) times for each failed connection by default. HTable will throw IOException if any connection to the RegionServer fails after the retries, so that the user client code for HTable instance can handle it further. NOTE: <code>HTable</code> is deprecated in HBase 1.0, in favor of <code>Table</code>.</p> </div> <div class="paragraph"> <p>Alternatively, update krb5-server to a version which solves these issues, such as krb5-server-1.10.3. See JIRA <a href="https://issues.apache.org/jira/browse/HBASE-10379">HBASE-10379</a> for more details.</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.zookeeper"><a class="anchor" href="#trouble.client.zookeeper"></a>106.7. ZooKeeper Client Connection Errors</h3> <div class="paragraph"> <p>Errors like this…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="integer">11</span>/<span class="octal">07</span>/<span class="octal">05</span> <span class="integer">11</span>:<span class="integer">26</span>:<span class="integer">41</span> WARN zookeeper.ClientCnxn: Session <span class="hex">0x0</span> <span class="keyword">for</span> server <span class="predefined-constant">null</span>, unexpected error, closing socket connection and attempting reconnect java.net.ConnectException: <span class="predefined-type">Connection</span> refused: no further information at sun.nio.ch.SocketChannelImpl.checkConnect(Native <span class="predefined-type">Method</span>) at sun.nio.ch.SocketChannelImpl.finishConnect(Unknown <span class="predefined-type">Source</span>) at org.apache.zookeeper.ClientCnxn<span class="error">$</span>SendThread.run(ClientCnxn.java:<span class="integer">1078</span>) <span class="integer">11</span>/<span class="octal">07</span>/<span class="octal">05</span> <span class="integer">11</span>:<span class="integer">26</span>:<span class="integer">43</span> INFO zookeeper.ClientCnxn: Opening socket connection to server localhost/<span class="float">127.0</span><span class="float">.0</span><span class="float">.1</span>:<span class="integer">2181</span> <span class="integer">11</span>/<span class="octal">07</span>/<span class="octal">05</span> <span class="integer">11</span>:<span class="integer">26</span>:<span class="integer">44</span> WARN zookeeper.ClientCnxn: Session <span class="hex">0x0</span> <span class="keyword">for</span> server <span class="predefined-constant">null</span>, unexpected error, closing socket connection and attempting reconnect java.net.ConnectException: <span class="predefined-type">Connection</span> refused: no further information at sun.nio.ch.SocketChannelImpl.checkConnect(Native <span class="predefined-type">Method</span>) at sun.nio.ch.SocketChannelImpl.finishConnect(Unknown <span class="predefined-type">Source</span>) at org.apache.zookeeper.ClientCnxn<span class="error">$</span>SendThread.run(ClientCnxn.java:<span class="integer">1078</span>) <span class="integer">11</span>/<span class="octal">07</span>/<span class="octal">05</span> <span class="integer">11</span>:<span class="integer">26</span>:<span class="integer">45</span> INFO zookeeper.ClientCnxn: Opening socket connection to server localhost/<span class="float">127.0</span><span class="float">.0</span><span class="float">.1</span>:<span class="integer">2181</span></code></pre> </div> </div> <div class="paragraph"> <p>…are either due to ZooKeeper being down, or unreachable due to network issues.</p> </div> <div class="paragraph"> <p>The utility <a href="#trouble.tools.builtin.zkcli">zkcli</a> may help investigate ZooKeeper issues.</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.oome.directmemory.leak"><a class="anchor" href="#trouble.client.oome.directmemory.leak"></a>106.8. Client running out of memory though heap size seems to be stable (but the off-heap/direct heap keeps growing)</h3> <div class="paragraph"> <p>You are likely running into the issue that is described and worked through in the mail thread <a href="http://search-hadoop.com/m/ubhrX8KvcH/Suspected+memory+leak&subj=Re+Suspected+memory+leak">HBase, mail # user - Suspected memory leak</a> and continued over in <a href="http://search-hadoop.com/m/p2Agc1Zy7Va/MaxDirectMemorySize+Was%253A+Suspected+memory+leak&subj=Re+FeedbackRe+Suspected+memory+leak">HBase, mail # dev - FeedbackRe: Suspected memory leak</a>. A workaround is passing your client-side JVM a reasonable value for <code>-XX:MaxDirectMemorySize</code>. By default, the <code>MaxDirectMemorySize</code> is equal to your <code>-Xmx</code> max heapsize setting (if <code>-Xmx</code> is set). Try setting it to something smaller (for example, one user had success setting it to <code>1g</code> when they had a client-side heap of <code>12g</code>). If you set it too small, it will bring on <code>FullGCs</code> so keep it a bit hefty. You want to make this setting client-side only especially if you are running the new experimental server-side off-heap cache since this feature depends on being able to use big direct buffers (You may have to keep separate client-side and server-side config dirs).</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.slowdown.admin"><a class="anchor" href="#trouble.client.slowdown.admin"></a>106.9. Client Slowdown When Calling Admin Methods (flush, compact, etc.)</h3> <div class="paragraph"> <p>This is a client issue fixed by <a href="https://issues.apache.org/jira/browse/HBASE-5073">HBASE-5073</a> in 0.90.6. There was a ZooKeeper leak in the client and the client was getting pummeled by ZooKeeper events with each additional invocation of the admin API.</p> </div> </div> <div class="sect2"> <h3 id="trouble.client.security.rpc"><a class="anchor" href="#trouble.client.security.rpc"></a>106.10. Secure Client Cannot Connect ([Caused by GSSException: No valid credentials provided(Mechanism level: Failed to find any Kerberos tgt)])</h3> <div class="paragraph"> <p>There can be several causes that produce this symptom.</p> </div> <div class="paragraph"> <p>First, check that you have a valid Kerberos ticket. One is required in order to set up communication with a secure Apache HBase cluster. Examine the ticket currently in the credential cache, if any, by running the <code>klist</code> command line utility. If no ticket is listed, you must obtain a ticket by running the <code>kinit</code> command with either a keytab specified, or by interactively entering a password for the desired principal.</p> </div> <div class="paragraph"> <p>Then, consult the <a href="http://docs.oracle.com/javase/1.5.0/docs/guide/security/jgss/tutorials/Troubleshooting.html">Java Security Guide troubleshooting section</a>. The most common problem addressed there is resolved by setting <code>javax.security.auth.useSubjectCredsOnly</code> system property value to <code>false</code>.</p> </div> <div class="paragraph"> <p>Because of a change in the format in which MIT Kerberos writes its credentials cache, there is a bug in the Oracle JDK 6 Update 26 and earlier that causes Java to be unable to read the Kerberos credentials cache created by versions of MIT Kerberos 1.8.1 or higher. If you have this problematic combination of components in your environment, to work around this problem, first log in with <code>kinit</code> and then immediately refresh the credential cache with <code>kinit -R</code>. The refresh will rewrite the credential cache without the problematic formatting.</p> </div> <div class="paragraph"> <p>Finally, depending on your Kerberos configuration, you may need to install the <a href="http://docs.oracle.com/javase/1.4.2/docs/guide/security/jce/JCERefGuide.html">Java Cryptography Extension</a>, or JCE. Insure the JCE jars are on the classpath on both server and client systems.</p> </div> <div class="paragraph"> <p>You may also need to download the <a href="http://www.oracle.com/technetwork/java/javase/downloads/jce-6-download-429243.html">unlimited strength JCE policy files</a>. Uncompress and extract the downloaded file, and install the policy jars into <em><java-home>/lib/security</em>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.mapreduce"><a class="anchor" href="#trouble.mapreduce"></a>107. MapReduce</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.mapreduce.local"><a class="anchor" href="#trouble.mapreduce.local"></a>107.1. You Think You’re On The Cluster, But You’re Actually Local</h3> <div class="paragraph"> <p>This following stacktrace happened using <code>ImportTsv</code>, but things like this can happen on any job with a mis-configuration.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="text"> WARN mapred.LocalJobRunner: job_local_0001 java.lang.IllegalArgumentException: Can't read partitions file at org.apache.hadoop.hbase.mapreduce.hadoopbackport.TotalOrderPartitioner.setConf(TotalOrderPartitioner.java:111) at org.apache.hadoop.util.ReflectionUtils.setConf(ReflectionUtils.java:62) at org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:117) at org.apache.hadoop.mapred.MapTask$NewOutputCollector.<init>(MapTask.java:560) at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:639) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:323) at org.apache.hadoop.mapred.LocalJobRunner$Job.run(LocalJobRunner.java:210) Caused by: java.io.FileNotFoundException: File _partition.lst does not exist. at org.apache.hadoop.fs.RawLocalFileSystem.getFileStatus(RawLocalFileSystem.java:383) at org.apache.hadoop.fs.FilterFileSystem.getFileStatus(FilterFileSystem.java:251) at org.apache.hadoop.fs.FileSystem.getLength(FileSystem.java:776) at org.apache.hadoop.io.SequenceFile$Reader.<init>(SequenceFile.java:1424) at org.apache.hadoop.io.SequenceFile$Reader.<init>(SequenceFile.java:1419) at org.apache.hadoop.hbase.mapreduce.hadoopbackport.TotalOrderPartitioner.readPartitions(TotalOrderPartitioner.java:296)</code></pre> </div> </div> <div class="paragraph"> <p>…see the critical portion of the stack? It’s…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">at org.apache.hadoop.mapred.LocalJobRunner<span class="error">$</span>Job.run(LocalJobRunner.java:<span class="integer">210</span>)</code></pre> </div> </div> <div class="paragraph"> <p>LocalJobRunner means the job is running locally, not on the cluster.</p> </div> <div class="paragraph"> <p>To solve this problem, you should run your MR job with your <code>HADOOP_CLASSPATH</code> set to include the HBase dependencies. The "hbase classpath" utility can be used to do this easily. For example (substitute VERSION with your HBase version):</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">HADOOP_CLASSPATH=`hbase classpath` hadoop jar $HBASE_HOME/hbase-server-VERSION.jar rowcounter usertable</code></pre> </div> </div> <div class="paragraph"> <p>See <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/package-summary.html#classpathfor" class="bare">http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/package-summary.html#classpathfor</a> more information on HBase MapReduce jobs and classpaths.</p> </div> </div> <div class="sect2"> <h3 id="trouble.hbasezerocopybytestring"><a class="anchor" href="#trouble.hbasezerocopybytestring"></a>107.2. Launching a job, you get java.lang.IllegalAccessError: com/google/protobuf/HBaseZeroCopyByteString or class com.google.protobuf.ZeroCopyLiteralByteString cannot access its superclass com.google.protobuf.LiteralByteString</h3> <div class="paragraph"> <p>See <a href="https://issues.apache.org/jira/browse/HBASE-10304">HBASE-10304 Running an hbase job jar: IllegalAccessError: class com.google.protobuf.ZeroCopyLiteralByteString cannot access its superclass com.google.protobuf.LiteralByteString</a> and <a href="https://issues.apache.org/jira/browse/HBASE-11118">HBASE-11118 non environment variable solution for "IllegalAccessError: class com.google.protobuf.ZeroCopyLiteralByteString cannot access its superclass com.google.protobuf.LiteralByteString"</a>. The issue can also show up when trying to run spark jobs. See <a href="https://issues.apache.org/jira/browse/HBASE-10877">HBASE-10877 HBase non-retriable exception list should be expanded</a>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.namenode"><a class="anchor" href="#trouble.namenode"></a>108. NameNode</h2> <div class="sectionbody"> <div class="paragraph"> <p>For more information on the NameNode, see <a href="#arch.hdfs">HDFS</a>.</p> </div> <div class="sect2"> <h3 id="trouble.namenode.disk"><a class="anchor" href="#trouble.namenode.disk"></a>108.1. HDFS Utilization of Tables and Regions</h3> <div class="paragraph"> <p>To determine how much space HBase is using on HDFS use the <code>hadoop</code> shell commands from the NameNode. For example…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">hadoop fs -dus /hbase/</code></pre> </div> </div> <div class="paragraph"> <p>…returns the summarized disk utilization for all HBase objects.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">hadoop fs -dus /hbase/myTable</code></pre> </div> </div> <div class="paragraph"> <p>…returns the summarized disk utilization for the HBase table 'myTable'.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">hadoop fs -du /hbase/myTable</code></pre> </div> </div> <div class="paragraph"> <p>…returns a list of the regions under the HBase table 'myTable' and their disk utilization.</p> </div> <div class="paragraph"> <p>For more information on HDFS shell commands, see the <a href="http://hadoop.apache.org/common/docs/current/file_system_shell.html">HDFS FileSystem Shell documentation</a>.</p> </div> </div> <div class="sect2"> <h3 id="trouble.namenode.hbase.objects"><a class="anchor" href="#trouble.namenode.hbase.objects"></a>108.2. Browsing HDFS for HBase Objects</h3> <div class="paragraph"> <p>Sometimes it will be necessary to explore the HBase objects that exist on HDFS. These objects could include the WALs (Write Ahead Logs), tables, regions, StoreFiles, etc. The easiest way to do this is with the NameNode web application that runs on port 50070. The NameNode web application will provide links to the all the DataNodes in the cluster so that they can be browsed seamlessly.</p> </div> <div class="paragraph"> <p>The HDFS directory structure of HBase tables in the cluster is…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">/hbase /<Table> (Tables in the cluster) /<<span class="predefined-type">Region</span>> (Regions <span class="keyword">for</span> the table) /<ColumnFamily> (ColumnFamilies <span class="keyword">for</span> the <span class="predefined-type">Region</span> <span class="keyword">for</span> the table) /<StoreFile> (StoreFiles <span class="keyword">for</span> the ColumnFamily <span class="keyword">for</span> the Regions <span class="keyword">for</span> the table)</code></pre> </div> </div> <div class="paragraph"> <p>The HDFS directory structure of HBase WAL is..</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">/hbase /.logs /<RegionServer> (RegionServers) /<WAL> (WAL files <span class="keyword">for</span> the RegionServer)</code></pre> </div> </div> <div class="paragraph"> <p>See the <a href="http://hadoop.apache.org/common/docs/current/hdfs_user_guide.html">HDFS User Guide</a> for other non-shell diagnostic utilities like <code>fsck</code>.</p> </div> <div class="sect3"> <h4 id="trouble.namenode.0size.hlogs"><a class="anchor" href="#trouble.namenode.0size.hlogs"></a>108.2.1. Zero size WALs with data in them</h4> <div class="paragraph"> <p>Problem: when getting a listing of all the files in a RegionServer’s <em>.logs</em> directory, one file has a size of 0 but it contains data.</p> </div> <div class="paragraph"> <p>Answer: It’s an HDFS quirk. A file that’s currently being written to will appear to have a size of 0 but once it’s closed it will show its true size</p> </div> </div> <div class="sect3"> <h4 id="trouble.namenode.uncompaction"><a class="anchor" href="#trouble.namenode.uncompaction"></a>108.2.2. Use Cases</h4> <div class="paragraph"> <p>Two common use-cases for querying HDFS for HBase objects is research the degree of uncompaction of a table. If there are a large number of StoreFiles for each ColumnFamily it could indicate the need for a major compaction. Additionally, after a major compaction if the resulting StoreFile is "small" it could indicate the need for a reduction of ColumnFamilies for the table.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.network"><a class="anchor" href="#trouble.network"></a>109. Network</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.network.spikes"><a class="anchor" href="#trouble.network.spikes"></a>109.1. Network Spikes</h3> <div class="paragraph"> <p>If you are seeing periodic network spikes you might want to check the <code>compactionQueues</code> to see if major compactions are happening.</p> </div> <div class="paragraph"> <p>See <a href="#managed.compactions">Managed Compactions</a> for more information on managing compactions.</p> </div> </div> <div class="sect2"> <h3 id="trouble.network.loopback"><a class="anchor" href="#trouble.network.loopback"></a>109.2. Loopback IP</h3> <div class="paragraph"> <p>HBase expects the loopback IP Address to be 127.0.0.1. See the Getting Started section on <a href="#loopback.ip">Loopback IP - HBase 0.94.x and earlier</a>.</p> </div> </div> <div class="sect2"> <h3 id="trouble.network.ints"><a class="anchor" href="#trouble.network.ints"></a>109.3. Network Interfaces</h3> <div class="paragraph"> <p>Are all the network interfaces functioning correctly? Are you sure? See the Troubleshooting Case Study in <a href="#trouble.casestudy">Case Studies</a>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.rs"><a class="anchor" href="#trouble.rs"></a>110. RegionServer</h2> <div class="sectionbody"> <div class="paragraph"> <p>For more information on the RegionServers, see <a href="#regionserver.arch">RegionServer</a>.</p> </div> <div class="sect2"> <h3 id="trouble.rs.startup"><a class="anchor" href="#trouble.rs.startup"></a>110.1. Startup Errors</h3> <div class="sect3"> <h4 id="trouble.rs.startup.master_no_region"><a class="anchor" href="#trouble.rs.startup.master_no_region"></a>110.1.1. Master Starts, But RegionServers Do Not</h4> <div class="paragraph"> <p>The Master believes the RegionServers have the IP of 127.0.0.1 - which is localhost and resolves to the master’s own localhost.</p> </div> <div class="paragraph"> <p>The RegionServers are erroneously informing the Master that their IP addresses are 127.0.0.1.</p> </div> <div class="paragraph"> <p>Modify <em>/etc/hosts</em> on the region servers, from…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="error">#</span> Do not remove the following line, or various programs <span class="error">#</span> that require network functionality will fail. <span class="float">127.0</span><span class="float">.0</span><span class="float">.1</span> fully.qualified.regionservername regionservername localhost.localdomain localhost ::<span class="integer">1</span> localhost6.localdomain6 localhost6</code></pre> </div> </div> <div class="paragraph"> <p>... to (removing the master node’s name from localhost)…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="error">#</span> Do not remove the following line, or various programs <span class="error">#</span> that require network functionality will fail. <span class="float">127.0</span><span class="float">.0</span><span class="float">.1</span> localhost.localdomain localhost ::<span class="integer">1</span> localhost6.localdomain6 localhost6</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="trouble.rs.startup.compression"><a class="anchor" href="#trouble.rs.startup.compression"></a>110.1.2. Compression Link Errors</h4> <div class="paragraph"> <p>Since compression algorithms such as LZO need to be installed and configured on each cluster this is a frequent source of startup error. If you see messages like this…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="integer">11</span>/<span class="octal">02</span>/<span class="integer">20</span> <span class="octal">01</span>:<span class="integer">32</span>:<span class="integer">15</span> ERROR lzo.GPLNativeCodeLoader: Could not load <span class="directive">native</span> gpl library java.lang.UnsatisfiedLinkError: no gplcompression in java.library.path at java.lang.ClassLoader.loadLibrary(<span class="predefined-type">ClassLoader</span>.java:<span class="integer">1734</span>) at java.lang.Runtime.loadLibrary0(<span class="predefined-type">Runtime</span>.java:<span class="integer">823</span>) at java.lang.System.loadLibrary(<span class="predefined-type">System</span>.java:<span class="integer">1028</span>)</code></pre> </div> </div> <div class="paragraph"> <p>... then there is a path issue with the compression libraries. See the Configuration section on link:[LZO compression configuration].</p> </div> </div> </div> <div class="sect2"> <h3 id="trouble.rs.runtime"><a class="anchor" href="#trouble.rs.runtime"></a>110.2. Runtime Errors</h3> <div class="sect3"> <h4 id="trouble.rs.runtime.hang"><a class="anchor" href="#trouble.rs.runtime.hang"></a>110.2.1. RegionServer Hanging</h4> <div class="paragraph"> <p>Are you running an old JVM (< 1.6.0_u21?)? When you look at a thread dump, does it look like threads are BLOCKED but no one holds the lock all are blocked on? See <a href="https://issues.apache.org/jira/browse/HBASE-3622">HBASE 3622 Deadlock in HBaseServer (JVM bug?)</a>. Adding <code>-XX:+UseMembar</code> to the HBase <code>HBASE_OPTS</code> in <em>conf/hbase-env.sh</em> may fix it.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.filehandles"><a class="anchor" href="#trouble.rs.runtime.filehandles"></a>110.2.2. java.io.IOException…(Too many open files)</h4> <div class="paragraph"> <p>If you see log messages like this…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="integer">2010</span>-<span class="integer">09</span>-<span class="integer">13</span> <span class="octal">01</span>:<span class="integer">24</span>:<span class="integer">17</span>,<span class="integer">336</span> WARN org.apache.hadoop.hdfs.server.datanode.DataNode: Disk-related <span class="exception">IOException</span> in BlockReceiver constructor. Cause is java.io.IOException: Too many open files at java.io.UnixFileSystem.createFileExclusively(Native <span class="predefined-type">Method</span>) at java.io.File.createNewFile(<span class="predefined-type">File</span>.java:<span class="integer">883</span>)</code></pre> </div> </div> <div class="paragraph"> <p>... see the Getting Started section on link:[ulimit and nproc configuration].</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.xceivers"><a class="anchor" href="#trouble.rs.runtime.xceivers"></a>110.2.3. xceiverCount 258 exceeds the limit of concurrent xcievers 256</h4> <div class="paragraph"> <p>This typically shows up in the DataNode logs.</p> </div> <div class="paragraph"> <p>See the Getting Started section on link:[xceivers configuration].</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.oom_nt"><a class="anchor" href="#trouble.rs.runtime.oom_nt"></a>110.2.4. System instability, and the presence of "java.lang.OutOfMemoryError: unable to createnew native thread in exceptions" HDFS DataNode logs or that of any system daemon</h4> <div class="paragraph"> <p>See the Getting Started section on ulimit and nproc configuration. The default on recent Linux distributions is 1024 - which is far too low for HBase.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.gc"><a class="anchor" href="#trouble.rs.runtime.gc"></a>110.2.5. DFS instability and/or RegionServer lease timeouts</h4> <div class="paragraph"> <p>If you see warning messages like this…</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="integer">2009</span>-<span class="octal">02</span>-<span class="integer">24</span> <span class="integer">10</span>:<span class="octal">01</span>:<span class="integer">33</span>,<span class="integer">516</span> WARN org.apache.hadoop.hbase.util.Sleeper: We slept xxx ms, ten times longer than scheduled: <span class="integer">10000</span> <span class="integer">2009</span>-<span class="octal">02</span>-<span class="integer">24</span> <span class="integer">10</span>:<span class="octal">01</span>:<span class="integer">33</span>,<span class="integer">516</span> WARN org.apache.hadoop.hbase.util.Sleeper: We slept xxx ms, ten times longer than scheduled: <span class="integer">15000</span> <span class="integer">2009</span>-<span class="octal">02</span>-<span class="integer">24</span> <span class="integer">10</span>:<span class="octal">01</span>:<span class="integer">36</span>,<span class="integer">472</span> WARN org.apache.hadoop.hbase.regionserver.HRegionServer: unable to report to master <span class="keyword">for</span> xxx milliseconds - retrying</code></pre> </div> </div> <div class="paragraph"> <p>... or see full GC compactions then you may be experiencing full GC’s.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.nolivenodes"><a class="anchor" href="#trouble.rs.runtime.nolivenodes"></a>110.2.6. "No live nodes contain current block" and/or YouAreDeadException</h4> <div class="paragraph"> <p>These errors can happen either when running out of OS file handles or in periods of severe network problems where the nodes are unreachable.</p> </div> <div class="paragraph"> <p>See the Getting Started section on ulimit and nproc configuration and check your network.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.zkexpired"><a class="anchor" href="#trouble.rs.runtime.zkexpired"></a>110.2.7. ZooKeeper SessionExpired events</h4> <div class="paragraph"> <p>Master or RegionServers shutting down with messages like those in the logs:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">WARN org.apache.zookeeper.ClientCnxn: <span class="exception">Exception</span> closing session <span class="hex">0x278bd16a96000f</span> to sun.nio.ch.SelectionKeyImpl<span class="error">@</span><span class="integer">355811</span>ec java.io.IOException: TIMED OUT at org.apache.zookeeper.ClientCnxn<span class="error">$</span>SendThread.run(ClientCnxn.java:<span class="integer">906</span>) WARN org.apache.hadoop.hbase.util.Sleeper: We slept <span class="integer">79410</span>ms, ten times longer than scheduled: <span class="integer">5000</span> INFO org.apache.zookeeper.ClientCnxn: Attempting connection to server hostname/IP:PORT INFO org.apache.zookeeper.ClientCnxn: Priming connection to java.nio.channels.SocketChannel[connected local=/IP:PORT remote=hostname/IP:PORT] INFO org.apache.zookeeper.ClientCnxn: Server connection successful WARN org.apache.zookeeper.ClientCnxn: <span class="exception">Exception</span> closing session <span class="hex">0x278bd16a96000d</span> to sun.nio.ch.SelectionKeyImpl<span class="error">@</span><span class="float">3544d</span><span class="integer">65</span>e java.io.IOException: Session Expired at org.apache.zookeeper.ClientCnxn<span class="error">$</span>SendThread.readConnectResult(ClientCnxn.java:<span class="integer">589</span>) at org.apache.zookeeper.ClientCnxn<span class="error">$</span>SendThread.doIO(ClientCnxn.java:<span class="integer">709</span>) at org.apache.zookeeper.ClientCnxn<span class="error">$</span>SendThread.run(ClientCnxn.java:<span class="integer">945</span>) ERROR org.apache.hadoop.hbase.regionserver.HRegionServer: ZooKeeper session expired</code></pre> </div> </div> <div class="paragraph"> <p>The JVM is doing a long running garbage collecting which is pausing every threads (aka "stop the world"). Since the RegionServer’s local ZooKeeper client cannot send heartbeats, the session times out. By design, we shut down any node that isn’t able to contact the ZooKeeper ensemble after getting a timeout so that it stops serving data that may already be assigned elsewhere.</p> </div> <div class="ulist"> <ul> <li> <p>Make sure you give plenty of RAM (in <em>hbase-env.sh</em>), the default of 1GB won’t be able to sustain long running imports.</p> </li> <li> <p>Make sure you don’t swap, the JVM never behaves well under swapping.</p> </li> <li> <p>Make sure you are not CPU starving the RegionServer thread. For example, if you are running a MapReduce job using 6 CPU-intensive tasks on a machine with 4 cores, you are probably starving the RegionServer enough to create longer garbage collection pauses.</p> </li> <li> <p>Increase the ZooKeeper session timeout</p> </li> </ul> </div> <div class="paragraph"> <p>If you wish to increase the session timeout, add the following to your <em>hbase-site.xml</em> to increase the timeout from the default of 60 seconds to 120 seconds.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>zookeeper.session.timeout<span class="tag"></name></span> <span class="tag"><value></span>1200000<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.property.tickTime<span class="tag"></name></span> <span class="tag"><value></span>6000<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>Be aware that setting a higher timeout means that the regions served by a failed RegionServer will take at least that amount of time to be transferred to another RegionServer. For a production system serving live requests, we would instead recommend setting it lower than 1 minute and over-provision your cluster in order the lower the memory load on each machines (hence having less garbage to collect per machine).</p> </div> <div class="paragraph"> <p>If this is happening during an upload which only happens once (like initially loading all your data into HBase), consider bulk loading.</p> </div> <div class="paragraph"> <p>See <a href="#trouble.zookeeper.general">ZooKeeper, The Cluster Canary</a> for other general information about ZooKeeper troubleshooting.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.notservingregion"><a class="anchor" href="#trouble.rs.runtime.notservingregion"></a>110.2.8. NotServingRegionException</h4> <div class="paragraph"> <p>This exception is "normal" when found in the RegionServer logs at DEBUG level. This exception is returned back to the client and then the client goes back to <code>hbase:meta</code> to find the new location of the moved region.</p> </div> <div class="paragraph"> <p>However, if the NotServingRegionException is logged ERROR, then the client ran out of retries and something probably wrong.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.double_listed_regions"><a class="anchor" href="#trouble.rs.runtime.double_listed_regions"></a>110.2.9. Regions listed by domain name, then IP</h4> <div class="paragraph"> <p>Fix your DNS. In versions of Apache HBase before 0.92.x, reverse DNS needs to give same answer as forward lookup. See <a href="https://issues.apache.org/jira/browse/HBASE-3431">HBASE 3431 RegionServer is not using the name given it by the master; double entry in master listing of servers</a> for gorey details.</p> </div> </div> <div class="sect3"> <h4 id="brand.new.compressor"><a class="anchor" href="#brand.new.compressor"></a>110.2.10. Logs flooded with '2011-01-10 12:40:48,407 INFO org.apache.hadoop.io.compress.CodecPool: Gotbrand-new compressor' messages</h4> <div class="paragraph"> <p>We are not using the native versions of compression libraries. See <a href="https://issues.apache.org/jira/browse/HBASE-1900">HBASE-1900 Put back native support when hadoop 0.21 is released</a>. Copy the native libs from hadoop under HBase lib dir or symlink them into place and the message should go away.</p> </div> </div> <div class="sect3"> <h4 id="trouble.rs.runtime.client_went_away"><a class="anchor" href="#trouble.rs.runtime.client_went_away"></a>110.2.11. Server handler X on 60020 caught: java.nio.channels.ClosedChannelException</h4> <div class="paragraph"> <p>If you see this type of message it means that the region server was trying to read/send data from/to a client but it already went away. Typical causes for this are if the client was killed (you see a storm of messages like this when a MapReduce job is killed or fails) or if the client receives a SocketTimeoutException. It’s harmless, but you should consider digging in a bit more if you aren’t doing something to trigger them.</p> </div> </div> </div> <div class="sect2"> <h3 id="_snapshot_errors_due_to_reverse_dns"><a class="anchor" href="#_snapshot_errors_due_to_reverse_dns"></a>110.3. Snapshot Errors Due to Reverse DNS</h3> <div class="paragraph"> <p>Several operations within HBase, including snapshots, rely on properly configured reverse DNS. Some environments, such as Amazon EC2, have trouble with reverse DNS. If you see errors like the following on your RegionServers, check your reverse DNS configuration:</p> </div> <div class="listingblock"> <div class="content"> <pre>2013-05-01 00:04:56,356 DEBUG org.apache.hadoop.hbase.procedure.Subprocedure: Subprocedure 'backup1' coordinator notified of 'acquire', waiting on 'reached' or 'abort' from coordinator.</pre> </div> </div> <div class="paragraph"> <p>In general, the hostname reported by the RegionServer needs to be the same as the hostname the Master is trying to reach. You can see a hostname mismatch by looking for the following type of message in the RegionServer’s logs at start-up.</p> </div> <div class="listingblock"> <div class="content"> <pre>2013-05-01 00:03:00,614 INFO org.apache.hadoop.hbase.regionserver.HRegionServer: Master passed us hostname to use. Was=myhost-1234, Now=ip-10-55-88-99.ec2.internal</pre> </div> </div> </div> <div class="sect2"> <h3 id="trouble.rs.shutdown"><a class="anchor" href="#trouble.rs.shutdown"></a>110.4. Shutdown Errors</h3> </div> </div> </div> <div class="sect1"> <h2 id="trouble.master"><a class="anchor" href="#trouble.master"></a>111. Master</h2> <div class="sectionbody"> <div class="paragraph"> <p>For more information on the Master, see <a href="#master">master</a>.</p> </div> <div class="sect2"> <h3 id="trouble.master.startup"><a class="anchor" href="#trouble.master.startup"></a>111.1. Startup Errors</h3> <div class="sect3"> <h4 id="trouble.master.startup.migration"><a class="anchor" href="#trouble.master.startup.migration"></a>111.1.1. Master says that you need to run the HBase migrations script</h4> <div class="paragraph"> <p>Upon running that, the HBase migrations script says no files in root directory.</p> </div> <div class="paragraph"> <p>HBase expects the root directory to either not exist, or to have already been initialized by HBase running a previous time. If you create a new directory for HBase using Hadoop DFS, this error will occur. Make sure the HBase root directory does not currently exist or has been initialized by a previous run of HBase. Sure fire solution is to just use Hadoop dfs to delete the HBase root and let HBase create and initialize the directory itself.</p> </div> </div> <div class="sect3"> <h4 id="trouble.master.startup.zk.buffer"><a class="anchor" href="#trouble.master.startup.zk.buffer"></a>111.1.2. Packet len6080218 is out of range!</h4> <div class="paragraph"> <p>If you have many regions on your cluster and you see an error like that reported above in this sections title in your logs, see <a href="https://issues.apache.org/jira/browse/HBASE-4246">HBASE-4246 Cluster with too many regions cannot withstand some master failover scenarios</a>.</p> </div> </div> </div> <div class="sect2"> <h3 id="trouble.master.shutdown"><a class="anchor" href="#trouble.master.shutdown"></a>111.2. Shutdown Errors</h3> </div> </div> </div> <div class="sect1"> <h2 id="trouble.zookeeper"><a class="anchor" href="#trouble.zookeeper"></a>112. ZooKeeper</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.zookeeper.startup"><a class="anchor" href="#trouble.zookeeper.startup"></a>112.1. Startup Errors</h3> <div class="sect3"> <h4 id="trouble.zookeeper.startup.address"><a class="anchor" href="#trouble.zookeeper.startup.address"></a>112.1.1. Could not find my address: xyz in list of ZooKeeper quorum servers</h4> <div class="paragraph"> <p>A ZooKeeper server wasn’t able to start, throws that error. xyz is the name of your server.</p> </div> <div class="paragraph"> <p>This is a name lookup problem. HBase tries to start a ZooKeeper server on some machine but that machine isn’t able to find itself in the <code>hbase.zookeeper.quorum</code> configuration.</p> </div> <div class="paragraph"> <p>Use the hostname presented in the error message instead of the value you used. If you have a DNS server, you can set <code>hbase.zookeeper.dns.interface</code> and <code>hbase.zookeeper.dns.nameserver</code> in <em>hbase-site.xml</em> to make sure it resolves to the correct FQDN.</p> </div> </div> </div> <div class="sect2"> <h3 id="trouble.zookeeper.general"><a class="anchor" href="#trouble.zookeeper.general"></a>112.2. ZooKeeper, The Cluster Canary</h3> <div class="paragraph"> <p>ZooKeeper is the cluster’s "canary in the mineshaft". It’ll be the first to notice issues if any so making sure its happy is the short-cut to a humming cluster.</p> </div> <div class="paragraph"> <p>See the <a href="http://wiki.apache.org/hadoop/ZooKeeper/Troubleshooting">ZooKeeper Operating Environment Troubleshooting</a> page. It has suggestions and tools for checking disk and networking performance; i.e. the operating environment your ZooKeeper and HBase are running in.</p> </div> <div class="paragraph"> <p>Additionally, the utility <a href="#trouble.tools.builtin.zkcli">zkcli</a> may help investigate ZooKeeper issues.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.ec2"><a class="anchor" href="#trouble.ec2"></a>113. Amazon EC2</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.ec2.zookeeper"><a class="anchor" href="#trouble.ec2.zookeeper"></a>113.1. ZooKeeper does not seem to work on Amazon EC2</h3> <div class="paragraph"> <p>HBase does not start when deployed as Amazon EC2 instances. Exceptions like the below appear in the Master and/or RegionServer logs:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> <span class="integer">2009</span>-<span class="integer">10</span>-<span class="integer">19</span> <span class="integer">11</span>:<span class="integer">52</span>:<span class="integer">27</span>,<span class="octal">030</span> INFO org.apache.zookeeper.ClientCnxn: Attempting connection to server ec2-<span class="integer">174</span>-<span class="integer">129</span>-<span class="integer">15</span>-<span class="integer">236</span>.compute-<span class="integer">1</span>.amazonaws.com/<span class="float">10.244</span><span class="float">.9</span><span class="float">.171</span>:<span class="integer">2181</span> <span class="integer">2009</span>-<span class="integer">10</span>-<span class="integer">19</span> <span class="integer">11</span>:<span class="integer">52</span>:<span class="integer">27</span>,<span class="octal">032</span> WARN org.apache.zookeeper.ClientCnxn: <span class="exception">Exception</span> closing session <span class="hex">0x0</span> to sun.nio.ch.SelectionKeyImpl<span class="error">@</span><span class="float">656d</span>c861 java.net.ConnectException: <span class="predefined-type">Connection</span> refused</code></pre> </div> </div> <div class="paragraph"> <p>Security group policy is blocking the ZooKeeper port on a public address. Use the internal EC2 host names when configuring the ZooKeeper quorum peer list.</p> </div> </div> <div class="sect2"> <h3 id="trouble.ec2.instability"><a class="anchor" href="#trouble.ec2.instability"></a>113.2. Instability on Amazon EC2</h3> <div class="paragraph"> <p>Questions on HBase and Amazon EC2 come up frequently on the HBase dist-list. Search for old threads using <a href="http://search-hadoop.com/">Search Hadoop</a></p> </div> </div> <div class="sect2"> <h3 id="trouble.ec2.connection"><a class="anchor" href="#trouble.ec2.connection"></a>113.3. Remote Java Connection into EC2 Cluster Not Working</h3> <div class="paragraph"> <p>See Andrew’s answer here, up on the user list: <a href="http://search-hadoop.com/m/sPdqNFAwyg2">Remote Java client connection into EC2 instance</a>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.versions"><a class="anchor" href="#trouble.versions"></a>114. HBase and Hadoop version issues</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.versions.205"><a class="anchor" href="#trouble.versions.205"></a>114.1. <code>NoClassDefFoundError</code> when trying to run 0.90.x on hadoop-0.20.205.x (or hadoop-1.0.x)</h3> <div class="paragraph"> <p>Apache HBase 0.90.x does not ship with hadoop-0.20.205.x, etc. To make it run, you need to replace the hadoop jars that Apache HBase shipped with in its <em>lib</em> directory with those of the Hadoop you want to run HBase on. If even after replacing Hadoop jars you get the below exception:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">sv4r6s38: <span class="exception">Exception</span> in thread <span class="string"><span class="delimiter">"</span><span class="content">main</span><span class="delimiter">"</span></span> java.lang.NoClassDefFoundError: org/apache/commons/configuration/<span class="predefined-type">Configuration</span> sv4r6s38: at org.apache.hadoop.metrics2.lib.DefaultMetricsSystem.<init>(DefaultMetricsSystem.java:<span class="integer">37</span>) sv4r6s38: at org.apache.hadoop.metrics2.lib.DefaultMetricsSystem.<clinit>(DefaultMetricsSystem.java:<span class="integer">34</span>) sv4r6s38: at org.apache.hadoop.security.UgiInstrumentation.create(UgiInstrumentation.java:<span class="integer">51</span>) sv4r6s38: at org.apache.hadoop.security.UserGroupInformation.initialize(UserGroupInformation.java:<span class="integer">209</span>) sv4r6s38: at org.apache.hadoop.security.UserGroupInformation.ensureInitialized(UserGroupInformation.java:<span class="integer">177</span>) sv4r6s38: at org.apache.hadoop.security.UserGroupInformation.isSecurityEnabled(UserGroupInformation.java:<span class="integer">229</span>) sv4r6s38: at org.apache.hadoop.security.KerberosName.<clinit>(KerberosName.java:<span class="integer">83</span>) sv4r6s38: at org.apache.hadoop.security.UserGroupInformation.initialize(UserGroupInformation.java:<span class="integer">202</span>) sv4r6s38: at org.apache.hadoop.security.UserGroupInformation.ensureInitialized(UserGroupInformation.java:<span class="integer">177</span>)</code></pre> </div> </div> <div class="paragraph"> <p>you need to copy under <em>hbase/lib</em>, the <em>commons-configuration-X.jar</em> you find in your Hadoop’s <em>lib</em> directory. That should fix the above complaint.</p> </div> </div> <div class="sect2"> <h3 id="trouble.wrong.version"><a class="anchor" href="#trouble.wrong.version"></a>114.2. …cannot communicate with client version…</h3> <div class="paragraph"> <p>If you see something like the following in your logs <span class="computeroutput">... 2012-09-24 10:20:52,168 FATAL org.apache.hadoop.hbase.master.HMaster: Unhandled exception. Starting shutdown. org.apache.hadoop.ipc.RemoteException: Server IPC version 7 cannot communicate with client version 4 ...</span> …are you trying to talk to an Hadoop 2.0.x from an HBase that has an Hadoop 1.0.x client? Use the HBase built against Hadoop 2.0 or rebuild your HBase passing the -Dhadoop.profile=2.0 attribute to Maven (See <a href="#maven.build.hadoop">Building against various hadoop versions.</a> for more).</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_ipc_configuration_conflicts_with_hadoop"><a class="anchor" href="#_ipc_configuration_conflicts_with_hadoop"></a>115. IPC Configuration Conflicts with Hadoop</h2> <div class="sectionbody"> <div class="paragraph"> <p>If the Hadoop configuration is loaded after the HBase configuration, and you have configured custom IPC settings in both HBase and Hadoop, the Hadoop values may overwrite the HBase values. There is normally no need to change these settings for HBase, so this problem is an edge case. However, <a href="https://issues.apache.org/jira/browse/HBASE-11492">HBASE-11492</a> renames these settings for HBase to remove the chance of a conflict. Each of the setting names have been prefixed with <code>hbase.</code>, as shown in the following table. No action is required related to these changes unless you are already experiencing a conflict.</p> </div> <div class="paragraph"> <p>These changes were backported to HBase 0.98.x and apply to all newer versions.</p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Pre-0.98.x</th> <th class="tableblock halign-left valign-top">0.98-x And Newer</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.listen.queue.size</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.listen.queue.size</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.max.callqueue.size</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.max.callqueue.size</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.callqueue.handler.factor</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.callqueue.handler.factor</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.callqueue.read.share</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.callqueue.read.share</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.callqueue.type</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.callqueue.type</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.queue.max.call.delay</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.queue.max.call.delay</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.max.callqueue.length</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.max.callqueue.length</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.read.threadpool.size</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.read.threadpool.size</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.tcpkeepalive</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.tcpkeepalive</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.tcpnodelay</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.tcpnodelay</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.client.call.purge.timeout</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.client.call.purge.timeout</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.client.connection.maxidletime</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.client.connection.maxidletime</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.client.idlethreshold</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.client.idlethreshold</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.client.kill.max</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.client.kill.max</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">ipc.server.scan.vtime.weight</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.ipc.server.scan.vtime.weight</p></td> </tr> </tbody> </table> </div> </div> <div class="sect1"> <h2 id="_hbase_and_hdfs"><a class="anchor" href="#_hbase_and_hdfs"></a>116. HBase and HDFS</h2> <div class="sectionbody"> <div class="paragraph"> <p>General configuration guidance for Apache HDFS is out of the scope of this guide. Refer to the documentation available at <a href="http://hadoop.apache.org/" class="bare">http://hadoop.apache.org/</a> for extensive information about configuring HDFS. This section deals with HDFS in terms of HBase.</p> </div> <div class="paragraph"> <p>In most cases, HBase stores its data in Apache HDFS. This includes the HFiles containing the data, as well as the write-ahead logs (WALs) which store data before it is written to the HFiles and protect against RegionServer crashes. HDFS provides reliability and protection to data in HBase because it is distributed. To operate with the most efficiency, HBase needs data to be available locally. Therefore, it is a good practice to run an HDFS DataNode on each RegionServer.</p> </div> <div class="dlist"> <div class="title">Important Information and Guidelines for HBase and HDFS</div> <dl> <dt class="hdlist1">HBase is a client of HDFS.</dt> <dd> <p>HBase is an HDFS client, using the HDFS <code>DFSClient</code> class, and references to this class appear in HBase logs with other HDFS client log messages.</p> </dd> <dt class="hdlist1">Configuration is necessary in multiple places.</dt> <dd> <p>Some HDFS configurations relating to HBase need to be done at the HDFS (server) side. Others must be done within HBase (at the client side). Other settings need to be set at both the server and client side.</p> </dd> <dt class="hdlist1">Write errors which affect HBase may be logged in the HDFS logs rather than HBase logs.</dt> <dd> <p>When writing, HDFS pipelines communications from one DataNode to another. HBase communicates to both the HDFS NameNode and DataNode, using the HDFS client classes. Communication problems between DataNodes are logged in the HDFS logs, not the HBase logs.</p> </dd> <dt class="hdlist1">HBase communicates with HDFS using two different ports.</dt> <dd> <p>HBase communicates with DataNodes using the <code>ipc.Client</code> interface and the <code>DataNode</code> class. References to these will appear in HBase logs. Each of these communication channels use a different port (50010 and 50020 by default). The ports are configured in the HDFS configuration, via the <code>dfs.datanode.address</code> and <code>dfs.datanode.ipc.address</code> parameters.</p> </dd> <dt class="hdlist1">Errors may be logged in HBase, HDFS, or both.</dt> <dd> <p>When troubleshooting HDFS issues in HBase, check logs in both places for errors.</p> </dd> <dt class="hdlist1">HDFS takes a while to mark a node as dead. You can configure HDFS to avoid using stale DataNodes.</dt> <dd> <p>By default, HDFS does not mark a node as dead until it is unreachable for 630 seconds. In Hadoop 1.1 and Hadoop 2.x, this can be alleviated by enabling checks for stale DataNodes, though this check is disabled by default. You can enable the check for reads and writes separately, via <code>dfs.namenode.avoid.read.stale.datanode</code> and <code>dfs.namenode.avoid.write.stale.datanode settings</code>. A stale DataNode is one that has not been reachable for <code>dfs.namenode.stale.datanode.interval</code> (default is 30 seconds). Stale datanodes are avoided, and marked as the last possible target for a read or write operation. For configuration details, see the HDFS documentation.</p> </dd> <dt class="hdlist1">Settings for HDFS retries and timeouts are important to HBase.</dt> <dd> <p>You can configure settings for various retries and timeouts. Always refer to the HDFS documentation for current recommendations and defaults. Some of the settings important to HBase are listed here. Defaults are current as of Hadoop 2.3. Check the Hadoop documentation for the most current values and recommendations.</p> </dd> </dl> </div> <div class="paragraph"> <div class="title">Connection Timeouts</div> <p>Connection timeouts occur between the client (HBASE) and the HDFS DataNode. They may occur when establishing a connection, attempting to read, or attempting to write. The two settings below are used in combination, and affect connections between the DFSClient and the DataNode, the ipc.cClient and the DataNode, and communication between two DataNodes.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1"><code>dfs.client.socket-timeout</code> (default: 60000)</dt> <dd> <p>The amount of time before a client connection times out when establishing a connection or reading. The value is expressed in milliseconds, so the default is 60 seconds.</p> </dd> <dt class="hdlist1"><code>dfs.datanode.socket.write.timeout</code> (default: 480000)</dt> <dd> <p>The amount of time before a write operation times out. The default is 8 minutes, expressed as milliseconds.</p> </dd> </dl> </div> <div class="paragraph"> <div class="title">Typical Error Logs</div> <p>The following types of errors are often seen in the logs.</p> </div> <div class="paragraph"> <p><code>INFO HDFS.DFSClient: Failed to connect to /xxx50010, add to deadNodes and continue java.net.SocketTimeoutException: 60000 millis timeout while waiting for channel to be ready for connect. ch : java.nio.channels.SocketChannel[connection-pending remote=/region-server-1:50010]</code>:: All DataNodes for a block are dead, and recovery is not possible. Here is the sequence of events that leads to this error:</p> </div> <div class="paragraph"> <p><code>INFO org.apache.hadoop.HDFS.DFSClient: Exception in createBlockOutputStream java.net.SocketTimeoutException: 69000 millis timeout while waiting for channel to be ready for connect. ch : java.nio.channels.SocketChannel[connection-pending remote=/ xxx:50010]</code>:: This type of error indicates a write issue. In this case, the master wants to split the log. It does not have a local DataNodes so it tries to connect to a remote DataNode, but the DataNode is dead.</p> </div> </div> </div> <div class="sect1"> <h2 id="trouble.tests"><a class="anchor" href="#trouble.tests"></a>117. Running unit or integration tests</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.hdfs_2556"><a class="anchor" href="#trouble.hdfs_2556"></a>117.1. Runtime exceptions from MiniDFSCluster when running tests</h3> <div class="paragraph"> <p>If you see something like the following</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... java.lang.NullPointerException: <span class="predefined-constant">null</span> at org.apache.hadoop.hdfs.MiniDFSCluster.startDataNodes at org.apache.hadoop.hdfs.MiniDFSCluster.<init> at org.apache.hadoop.hbase.MiniHBaseCluster.<init> at org.apache.hadoop.hbase.HBaseTestingUtility.startMiniDFSCluster at org.apache.hadoop.hbase.HBaseTestingUtility.startMiniCluster ...</code></pre> </div> </div> <div class="paragraph"> <p>or</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... java.io.IOException: Shutting down at org.apache.hadoop.hbase.MiniHBaseCluster.init at org.apache.hadoop.hbase.MiniHBaseCluster.<init> at org.apache.hadoop.hbase.MiniHBaseCluster.<init> at org.apache.hadoop.hbase.HBaseTestingUtility.startMiniHBaseCluster at org.apache.hadoop.hbase.HBaseTestingUtility.startMiniCluster ...</code></pre> </div> </div> <div class="paragraph"> <p>... then try issuing the command umask 022 before launching tests. This is a workaround for <a href="https://issues.apache.org/jira/browse/HDFS-2556">HDFS-2556</a></p> </div> </div> </div> </div> <div class="sect1"> <h2 id="trouble.casestudy"><a class="anchor" href="#trouble.casestudy"></a>118. Case Studies</h2> <div class="sectionbody"> <div class="paragraph"> <p>For Performance and Troubleshooting Case Studies, see <a href="#casestudies">Apache HBase Case Studies</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="trouble.crypto"><a class="anchor" href="#trouble.crypto"></a>119. Cryptographic Features</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="trouble.crypto.hbase_10132"><a class="anchor" href="#trouble.crypto.hbase_10132"></a>119.1. sun.security.pkcs11.wrapper.PKCS11Exception: CKR_ARGUMENTS_BAD</h3> <div class="paragraph"> <p>This problem manifests as exceptions ultimately caused by:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Caused by: sun.security.pkcs11.wrapper.PKCS11Exception: CKR_ARGUMENTS_BAD at sun.security.pkcs11.wrapper.PKCS11.C_DecryptUpdate(Native <span class="predefined-type">Method</span>) at sun.security.pkcs11.P11Cipher.implDoFinal(P11Cipher.java:<span class="integer">795</span>)</code></pre> </div> </div> <div class="paragraph"> <p>This problem appears to affect some versions of OpenJDK 7 shipped by some Linux vendors. NSS is configured as the default provider. If the host has an x86_64 architecture, depending on if the vendor packages contain the defect, the NSS provider will not function correctly.</p> </div> <div class="paragraph"> <p>To work around this problem, find the JRE home directory and edit the file <em>lib/security/java.security</em>. Edit the file to comment out the line:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">security.provider<span class="float">.1</span>=sun.security.pkcs11.SunPKCS11 <span class="error">$</span>{java.home}/lib/security/nss.cfg</code></pre> </div> </div> <div class="paragraph"> <p>Then renumber the remaining providers accordingly.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_operating_system_specific_issues"><a class="anchor" href="#_operating_system_specific_issues"></a>120. Operating System Specific Issues</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_page_allocation_failure"><a class="anchor" href="#_page_allocation_failure"></a>120.1. Page Allocation Failure</h3> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> This issue is known to affect CentOS 6.2 and possibly CentOS 6.5. It may also affect some versions of Red Hat Enterprise Linux, according to <a href="https://bugzilla.redhat.com/show_bug.cgi?id=770545" class="bare">https://bugzilla.redhat.com/show_bug.cgi?id=770545</a>. </td> </tr> </table> </div> <div class="paragraph"> <p>Some users have reported seeing the following error:</p> </div> <div class="listingblock"> <div class="content"> <pre>kernel: java: page allocation failure. order:4, mode:0x20</pre> </div> </div> <div class="paragraph"> <p>Raising the value of <code>min_free_kbytes</code> was reported to fix this problem. This parameter is set to a percentage of the amount of RAM on your system, and is described in more detail at <a href="http://www.centos.org/docs/5/html/5.1/Deployment_Guide/s3-proc-sys-vm.html" class="bare">http://www.centos.org/docs/5/html/5.1/Deployment_Guide/s3-proc-sys-vm.html</a>.</p> </div> <div class="paragraph"> <p>To find the current value on your system, run the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre>[user@host]# cat /proc/sys/vm/min_free_kbytes</pre> </div> </div> <div class="paragraph"> <p>Next, raise the value. Try doubling, then quadrupling the value. Note that setting the value too low or too high could have detrimental effects on your system. Consult your operating system vendor for specific recommendations.</p> </div> <div class="paragraph"> <p>Use the following command to modify the value of <code>min_free_kbytes</code>, substituting <em><value></em> with your intended value:</p> </div> <div class="listingblock"> <div class="content"> <pre>[user@host]# echo <value> > /proc/sys/vm/min_free_kbytes</pre> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="_jdk_issues"><a class="anchor" href="#_jdk_issues"></a>121. JDK Issues</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_nosuchmethoderror_java_util_concurrent_concurrenthashmap_keyset"><a class="anchor" href="#_nosuchmethoderror_java_util_concurrent_concurrenthashmap_keyset"></a>121.1. NoSuchMethodError: java.util.concurrent.ConcurrentHashMap.keySet</h3> <div class="paragraph"> <p>If you see this in your logs:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Caused by: java.lang.NoSuchMethodError: java.util.concurrent.ConcurrentHashMap.keySet()Ljava/util/concurrent/<span class="predefined-type">ConcurrentHashMap</span><span class="error">$</span>KeySetView; at org.apache.hadoop.hbase.master.ServerManager.findServerWithSameHostnamePortWithLock(ServerManager.java:<span class="integer">393</span>) at org.apache.hadoop.hbase.master.ServerManager.checkAndRecordNewServer(ServerManager.java:<span class="integer">307</span>) at org.apache.hadoop.hbase.master.ServerManager.regionServerStartup(ServerManager.java:<span class="integer">244</span>) at org.apache.hadoop.hbase.master.MasterRpcServices.regionServerStartup(MasterRpcServices.java:<span class="integer">304</span>) at org.apache.hadoop.hbase.protobuf.generated.RegionServerStatusProtos<span class="error">$</span>RegionServerStatusService<span class="error">$</span><span class="integer">2</span>.callBlockingMethod(RegionServerStatusProtos.java:<span class="integer">7910</span>) at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:<span class="integer">2020</span>) ... <span class="integer">4</span> more</code></pre> </div> </div> <div class="paragraph"> <p>then check if you compiled with jdk8 and tried to run it on jdk7. If so, this won’t work. Run on jdk8 or recompile with jdk7. See <a href="https://issues.apache.org/jira/browse/HBASE-10607">HBASE-10607 JDK8 NoSuchMethodError involving ConcurrentHashMap.keySet if running on JRE 7</a>.</p> </div> </div> </div> </div> <h1 id="casestudies" class="sect0"><a class="anchor" href="#casestudies"></a>Apache HBase Case Studies</h1> <div class="sect1"> <h2 id="casestudies.overview"><a class="anchor" href="#casestudies.overview"></a>122. Overview</h2> <div class="sectionbody"> <div class="paragraph"> <p>This chapter will describe a variety of performance and troubleshooting case studies that can provide a useful blueprint on diagnosing Apache HBase cluster issues.</p> </div> <div class="paragraph"> <p>For more information on Performance and Troubleshooting, see <a href="#performance">Apache HBase Performance Tuning</a> and <a href="#trouble">Troubleshooting and Debugging Apache HBase</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="casestudies.schema"><a class="anchor" href="#casestudies.schema"></a>123. Schema Design</h2> <div class="sectionbody"> <div class="paragraph"> <p>See the schema design case studies here: <a href="#schema.casestudies">Schema Design Case Studies</a></p> </div> </div> </div> <div class="sect1"> <h2 id="casestudies.perftroub"><a class="anchor" href="#casestudies.perftroub"></a>124. Performance/Troubleshooting</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="casestudies.slownode"><a class="anchor" href="#casestudies.slownode"></a>124.1. Case Study #1 (Performance Issue On A Single Node)</h3> <div class="sect3"> <h4 id="_scenario"><a class="anchor" href="#_scenario"></a>124.1.1. Scenario</h4> <div class="paragraph"> <p>Following a scheduled reboot, one data node began exhibiting unusual behavior. Routine MapReduce jobs run against HBase tables which regularly completed in five or six minutes began taking 30 or 40 minutes to finish. These jobs were consistently found to be waiting on map and reduce tasks assigned to the troubled data node (e.g., the slow map tasks all had the same Input Split). The situation came to a head during a distributed copy, when the copy was severely prolonged by the lagging node.</p> </div> </div> <div class="sect3"> <h4 id="_hardware"><a class="anchor" href="#_hardware"></a>124.1.2. Hardware</h4> <div class="ulist"> <div class="title">Datanodes:</div> <ul> <li> <p>Two 12-core processors</p> </li> <li> <p>Six Enerprise SATA disks</p> </li> <li> <p>24GB of RAM</p> </li> <li> <p>Two bonded gigabit NICs</p> </li> </ul> </div> <div class="ulist"> <div class="title">Network:</div> <ul> <li> <p>10 Gigabit top-of-rack switches</p> </li> <li> <p>20 Gigabit bonded interconnects between racks.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="_hypotheses"><a class="anchor" href="#_hypotheses"></a>124.1.3. Hypotheses</h4> <div class="sect4"> <h5 id="_hbase_hot_spot_region"><a class="anchor" href="#_hbase_hot_spot_region"></a>HBase "Hot Spot" Region</h5> <div class="paragraph"> <p>We hypothesized that we were experiencing a familiar point of pain: a "hot spot" region in an HBase table, where uneven key-space distribution can funnel a huge number of requests to a single HBase region, bombarding the RegionServer process and cause slow response time. Examination of the HBase Master status page showed that the number of HBase requests to the troubled node was almost zero. Further, examination of the HBase logs showed that there were no region splits, compactions, or other region transitions in progress. This effectively ruled out a "hot spot" as the root cause of the observed slowness.</p> </div> </div> <div class="sect4"> <h5 id="_hbase_region_with_non_local_data"><a class="anchor" href="#_hbase_region_with_non_local_data"></a>HBase Region With Non-Local Data</h5> <div class="paragraph"> <p>Our next hypothesis was that one of the MapReduce tasks was requesting data from HBase that was not local to the DataNode, thus forcing HDFS to request data blocks from other servers over the network. Examination of the DataNode logs showed that there were very few blocks being requested over the network, indicating that the HBase region was correctly assigned, and that the majority of the necessary data was located on the node. This ruled out the possibility of non-local data causing a slowdown.</p> </div> </div> <div class="sect4"> <h5 id="_excessive_i_o_wait_due_to_swapping_or_an_over_worked_or_failing_hard_disk"><a class="anchor" href="#_excessive_i_o_wait_due_to_swapping_or_an_over_worked_or_failing_hard_disk"></a>Excessive I/O Wait Due To Swapping Or An Over-Worked Or Failing Hard Disk</h5> <div class="paragraph"> <p>After concluding that the Hadoop and HBase were not likely to be the culprits, we moved on to troubleshooting the DataNode’s hardware. Java, by design, will periodically scan its entire memory space to do garbage collection. If system memory is heavily overcommitted, the Linux kernel may enter a vicious cycle, using up all of its resources swapping Java heap back and forth from disk to RAM as Java tries to run garbage collection. Further, a failing hard disk will often retry reads and/or writes many times before giving up and returning an error. This can manifest as high iowait, as running processes wait for reads and writes to complete. Finally, a disk nearing the upper edge of its performance envelope will begin to cause iowait as it informs the kernel that it cannot accept any more data, and the kernel queues incoming data into the dirty write pool in memory. However, using <code>vmstat(1)</code> and <code>free(1)</code>, we could see that no swap was being used, and the amount of disk IO was only a few kilobytes per second.</p> </div> </div> <div class="sect4"> <h5 id="_slowness_due_to_high_processor_usage"><a class="anchor" href="#_slowness_due_to_high_processor_usage"></a>Slowness Due To High Processor Usage</h5> <div class="paragraph"> <p>Next, we checked to see whether the system was performing slowly simply due to very high computational load. <code>top(1)</code> showed that the system load was higher than normal, but <code>vmstat(1)</code> and <code>mpstat(1)</code> showed that the amount of processor being used for actual computation was low.</p> </div> </div> <div class="sect4"> <h5 id="_network_saturation_the_winner"><a class="anchor" href="#_network_saturation_the_winner"></a>Network Saturation (The Winner)</h5> <div class="paragraph"> <p>Since neither the disks nor the processors were being utilized heavily, we moved on to the performance of the network interfaces. The DataNode had two gigabit ethernet adapters, bonded to form an active-standby interface. <code>ifconfig(8)</code> showed some unusual anomalies, namely interface errors, overruns, framing errors. While not unheard of, these kinds of errors are exceedingly rare on modern hardware which is operating as it should:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ /sbin/ifconfig bond0 bond0 Link encap:Ethernet HWaddr 00:00:00:00:00:00 inet addr:10.x.x.x Bcast:10.x.x.255 Mask:255.255.255.0 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1 RX packets:2990700159 errors:12 dropped:0 overruns:1 frame:6 <--- Look Here! Errors! TX packets:3443518196 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:2416328868676 (2.4 TB) TX bytes:3464991094001 (3.4 TB)</pre> </div> </div> <div class="paragraph"> <p>These errors immediately lead us to suspect that one or more of the ethernet interfaces might have negotiated the wrong line speed. This was confirmed both by running an ICMP ping from an external host and observing round-trip-time in excess of 700ms, and by running <code>ethtool(8)</code> on the members of the bond interface and discovering that the active interface was operating at 100Mbs/, full duplex.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ sudo ethtool eth0 Settings for eth0: Supported ports: [ TP ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Advertised pause frame use: No Advertised auto-negotiation: Yes Link partner advertised link modes: Not reported Link partner advertised pause frame use: No Link partner advertised auto-negotiation: No Speed: 100Mb/s <--- Look Here! Should say 1000Mb/s! Duplex: Full Port: Twisted Pair PHYAD: 1 Transceiver: internal Auto-negotiation: on MDI-X: Unknown Supports Wake-on: umbg Wake-on: g Current message level: 0x00000003 (3) Link detected: yes</pre> </div> </div> <div class="paragraph"> <p>In normal operation, the ICMP ping round trip time should be around 20ms, and the interface speed and duplex should read, "1000MB/s", and, "Full", respectively.</p> </div> </div> </div> <div class="sect3"> <h4 id="_resolution"><a class="anchor" href="#_resolution"></a>124.1.4. Resolution</h4> <div class="paragraph"> <p>After determining that the active ethernet adapter was at the incorrect speed, we used the <code>ifenslave(8)</code> command to make the standby interface the active interface, which yielded an immediate improvement in MapReduce performance, and a 10 times improvement in network throughput:</p> </div> <div class="paragraph"> <p>On the next trip to the datacenter, we determined that the line speed issue was ultimately caused by a bad network cable, which was replaced.</p> </div> </div> </div> <div class="sect2"> <h3 id="casestudies.perf.1"><a class="anchor" href="#casestudies.perf.1"></a>124.2. Case Study #2 (Performance Research 2012)</h3> <div class="paragraph"> <p>Investigation results of a self-described "we’re not sure what’s wrong, but it seems slow" problem. <a href="http://gbif.blogspot.com/2012/03/hbase-performance-evaluation-continued.html" class="bare">http://gbif.blogspot.com/2012/03/hbase-performance-evaluation-continued.html</a></p> </div> </div> <div class="sect2"> <h3 id="casestudies.perf.2"><a class="anchor" href="#casestudies.perf.2"></a>124.3. Case Study #3 (Performance Research 2010))</h3> <div class="paragraph"> <p>Investigation results of general cluster performance from 2010. Although this research is on an older version of the codebase, this writeup is still very useful in terms of approach. <a href="http://hstack.org/hbase-performance-testing/" class="bare">http://hstack.org/hbase-performance-testing/</a></p> </div> </div> <div class="sect2"> <h3 id="casestudies.max.transfer.threads"><a class="anchor" href="#casestudies.max.transfer.threads"></a>124.4. Case Study #4 (max.transfer.threads Config)</h3> <div class="paragraph"> <p>Case study of configuring <code>max.transfer.threads</code> (previously known as <code>xcievers</code>) and diagnosing errors from misconfigurations. <a href="http://www.larsgeorge.com/2012/03/hadoop-hbase-and-xceivers.html" class="bare">http://www.larsgeorge.com/2012/03/hadoop-hbase-and-xceivers.html</a></p> </div> <div class="paragraph"> <p>See also <a href="#dfs.datanode.max.transfer.threads"><code>dfs.datanode.max.transfer.threads</code> </a>.</p> </div> </div> </div> </div> <h1 id="ops_mgt" class="sect0"><a class="anchor" href="#ops_mgt"></a>Apache HBase Operational Management</h1> <div class="openblock partintro"> <div class="content"> This chapter will cover operational tools and practices required of a running Apache HBase cluster. The subject of operations is related to the topics of <a href="#trouble">Troubleshooting and Debugging Apache HBase</a>, <a href="#performance">Apache HBase Performance Tuning</a>, and <a href="#configuration">Apache HBase Configuration</a> but is a distinct topic in itself. </div> </div> <div class="sect1"> <h2 id="tools"><a class="anchor" href="#tools"></a>125. HBase Tools and Utilities</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase provides several tools for administration, analysis, and debugging of your cluster. The entry-point to most of these tools is the <em>bin/hbase</em> command, though some tools are available in the <em>dev-support/</em> directory.</p> </div> <div class="paragraph"> <p>To see usage instructions for <em>bin/hbase</em> command, run it with no arguments, or with the <code>-h</code> argument. These are the usage instructions for HBase 0.98.x. Some commands, such as <code>version</code>, <code>pe</code>, <code>ltt</code>, <code>clean</code>, are not available in previous versions.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase Usage: hbase [<options>] <command> [<args>] Options: --config DIR Configuration direction to use. Default: ./conf --hosts HOSTS Override the list in 'regionservers' file Commands: Some commands take arguments. Pass no args or -h for usage. shell Run the HBase shell hbck Run the hbase 'fsck' tool wal Write-ahead-log analyzer hfile Store file analyzer zkcli Run the ZooKeeper shell upgrade Upgrade hbase master Run an HBase HMaster node regionserver Run an HBase HRegionServer node zookeeper Run a Zookeeper server rest Run an HBase REST server thrift Run the HBase Thrift server thrift2 Run the HBase Thrift2 server clean Run the HBase clean up script classpath Dump hbase CLASSPATH mapredcp Dump CLASSPATH entries required by mapreduce pe Run PerformanceEvaluation ltt Run LoadTestTool version Print the version CLASSNAME Run the class named CLASSNAME</pre> </div> </div> <div class="paragraph"> <p>Some of the tools and utilities below are Java classes which are passed directly to the <em>bin/hbase</em> command, as referred to in the last line of the usage instructions. Others, such as <code>hbase shell</code> (<a href="#shell">The Apache HBase Shell</a>), <code>hbase upgrade</code> (<a href="#upgrading">Upgrading</a>), and <code>hbase thrift</code> (<a href="#thrift">Thrift API and Filter Language</a>), are documented elsewhere in this guide.</p> </div> <div class="sect2"> <h3 id="_canary"><a class="anchor" href="#_canary"></a>125.1. Canary</h3> <div class="paragraph"> <p>There is a Canary class can help users to canary-test the HBase cluster status, with every column-family for every regions or RegionServer’s granularity. To see the usage, use the <code>--help</code> parameter.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -help Usage: hbase canary [opts] [table1 [table2]...] | [regionserver1 [regionserver2]..] where [opts] are: -help Show this help and exit. -regionserver replace the table argument to regionserver, which means to enable regionserver mode -daemon Continuous check at defined intervals. -interval <N> Interval between checks (sec) -e Use region/regionserver as regular expression which means the region/regionserver is regular expression pattern -f <B> stop whole program if first error occurs, default is true -t <N> timeout for a check, default is 600000 (milliseconds) -writeSniffing enable the write sniffing in canary -treatFailureAsError treats read / write failure as error -writeTable The table used for write sniffing. Default is hbase:canary -D<configProperty>=<value> assigning or override the configuration params</pre> </div> </div> <div class="paragraph"> <p>This tool will return non zero error codes to user for collaborating with other monitoring tools, such as Nagios. The error code definitions are:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">int</span> USAGE_EXIT_CODE = <span class="integer">1</span>; <span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">int</span> INIT_ERROR_EXIT_CODE = <span class="integer">2</span>; <span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">int</span> TIMEOUT_ERROR_EXIT_CODE = <span class="integer">3</span>; <span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">int</span> ERROR_EXIT_CODE = <span class="integer">4</span>;</code></pre> </div> </div> <div class="paragraph"> <p>Here are some examples based on the following given case. There are two Table objects called test-01 and test-02, they have two column family cf1 and cf2 respectively, and deployed on the 3 RegionServers. see following table.</p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">RegionServer</th> <th class="tableblock halign-left valign-top">test-01</th> <th class="tableblock halign-left valign-top">test-02</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">rs1</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">r1</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">r2</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">rs2</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">r2</p></td> <td class="tableblock halign-left valign-top"></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">rs3</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">r2</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">r1</p></td> </tr> </tbody> </table> <div class="paragraph"> <p>Following are some examples based on the previous given case.</p> </div> <div class="sect3"> <h4 id="_canary_test_for_every_column_family_store_of_every_region_of_every_table"><a class="anchor" href="#_canary_test_for_every_column_family_store_of_every_region_of_every_table"></a>125.1.1. Canary test for every column family (store) of every region of every table</h4> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary 3/12/09 03:26:32 INFO tool.Canary: read from region test-01,,1386230156732.0e3c7d77ffb6361ea1b996ac1042ca9a. column family cf1 in 2ms 13/12/09 03:26:32 INFO tool.Canary: read from region test-01,,1386230156732.0e3c7d77ffb6361ea1b996ac1042ca9a. column family cf2 in 2ms 13/12/09 03:26:32 INFO tool.Canary: read from region test-01,0004883,1386230156732.87b55e03dfeade00f441125159f8ca87. column family cf1 in 4ms 13/12/09 03:26:32 INFO tool.Canary: read from region test-01,0004883,1386230156732.87b55e03dfeade00f441125159f8ca87. column family cf2 in 1ms ... 13/12/09 03:26:32 INFO tool.Canary: read from region test-02,,1386559511167.aa2951a86289281beee480f107bb36ee. column family cf1 in 5ms 13/12/09 03:26:32 INFO tool.Canary: read from region test-02,,1386559511167.aa2951a86289281beee480f107bb36ee. column family cf2 in 3ms 13/12/09 03:26:32 INFO tool.Canary: read from region test-02,0004883,1386559511167.cbda32d5e2e276520712d84eaaa29d84. column family cf1 in 31ms 13/12/09 03:26:32 INFO tool.Canary: read from region test-02,0004883,1386559511167.cbda32d5e2e276520712d84eaaa29d84. column family cf2 in 8ms</pre> </div> </div> <div class="paragraph"> <p>So you can see, table test-01 has two regions and two column families, so the Canary tool will pick 4 small piece of data from 4 (2 region * 2 store) different stores. This is a default behavior of the this tool does.</p> </div> </div> <div class="sect3"> <h4 id="_canary_test_for_every_column_family_store_of_every_region_of_specific_table_s"><a class="anchor" href="#_canary_test_for_every_column_family_store_of_every_region_of_specific_table_s"></a>125.1.2. Canary test for every column family (store) of every region of specific table(s)</h4> <div class="paragraph"> <p>You can also test one or more specific tables.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary test-01 test-02</pre> </div> </div> </div> <div class="sect3"> <h4 id="_canary_test_with_regionserver_granularity"><a class="anchor" href="#_canary_test_with_regionserver_granularity"></a>125.1.3. Canary test with RegionServer granularity</h4> <div class="paragraph"> <p>This will pick one small piece of data from each RegionServer, and can also put your RegionServer name as input options for canary-test specific RegionServer.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -regionserver 13/12/09 06:05:17 INFO tool.Canary: Read from table:test-01 on region server:rs2 in 72ms 13/12/09 06:05:17 INFO tool.Canary: Read from table:test-02 on region server:rs3 in 34ms 13/12/09 06:05:17 INFO tool.Canary: Read from table:test-01 on region server:rs1 in 56ms</pre> </div> </div> </div> <div class="sect3"> <h4 id="_canary_test_with_regular_expression_pattern"><a class="anchor" href="#_canary_test_with_regular_expression_pattern"></a>125.1.4. Canary test with regular expression pattern</h4> <div class="paragraph"> <p>This will test both table test-01 and test-02.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -e test-0[1-2]</pre> </div> </div> </div> <div class="sect3"> <h4 id="_run_canary_test_as_daemon_mode"><a class="anchor" href="#_run_canary_test_as_daemon_mode"></a>125.1.5. Run canary test as daemon mode</h4> <div class="paragraph"> <p>Run repeatedly with interval defined in option <code>-interval</code> whose default value is 60 seconds. This daemon will stop itself and return non-zero error code if any error occurs, due to the default value of option -f is true.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -daemon</pre> </div> </div> <div class="paragraph"> <p>Run repeatedly with internal 5 seconds and will not stop itself even if errors occur in the test.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -daemon -interval 50000 -f false</pre> </div> </div> </div> <div class="sect3"> <h4 id="_force_timeout_if_canary_test_stuck"><a class="anchor" href="#_force_timeout_if_canary_test_stuck"></a>125.1.6. Force timeout if canary test stuck</h4> <div class="paragraph"> <p>In some cases the request is stuck and no response is sent back to the client. This can happen with dead RegionServers which the master has not yet noticed. Because of this we provide a timeout option to kill the canary test and return a non-zero error code. This run sets the timeout value to 60 seconds, the default value is 600 seconds.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -t 600000</pre> </div> </div> </div> <div class="sect3"> <h4 id="_enable_write_sniffing_in_canary"><a class="anchor" href="#_enable_write_sniffing_in_canary"></a>125.1.7. Enable write sniffing in canary</h4> <div class="paragraph"> <p>By default, the canary tool only check the read operations, it’s hard to find the problem in the write path. To enable the write sniffing, you can run canary with the <code>-writeSniffing</code> option. When the write sniffing is enabled, the canary tool will create a hbase table and make sure the regions of the table distributed on all region servers. In each sniffing period, the canary will try to put data to these regions to check the write availability of each region server.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -writeSniffing</pre> </div> </div> <div class="paragraph"> <p>The default write table is <code>hbase:canary</code> and can be specified by the option <code>-writeTable</code>.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary -writeSniffing -writeTable ns:canary</pre> </div> </div> <div class="paragraph"> <p>The default value size of each put is 10 bytes and you can set it by the config key: <code>hbase.canary.write.value.size</code>.</p> </div> </div> <div class="sect3"> <h4 id="_treat_read_write_failure_as_error"><a class="anchor" href="#_treat_read_write_failure_as_error"></a>125.1.8. Treat read / write failure as error</h4> <div class="paragraph"> <p>By default, the canary tool only logs read failure, due to e.g. RetriesExhaustedException, while returning normal exit code. To treat read / write failure as error, you can run canary with the <code>-treatFailureAsError</code> option. When enabled, read / write failure would result in error exit code.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ${HBASE_HOME}/bin/hbase canary --treatFailureAsError</pre> </div> </div> </div> <div class="sect3"> <h4 id="_running_canary_in_a_kerberos_enabled_cluster"><a class="anchor" href="#_running_canary_in_a_kerberos_enabled_cluster"></a>125.1.9. Running Canary in a Kerberos-enabled Cluster</h4> <div class="paragraph"> <p>To run Canary in a Kerberos-enabled cluster, configure the following two properties in <em>hbase-site.xml</em>:</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.client.keytab.file</code></p> </li> <li> <p><code>hbase.client.kerberos.principal</code></p> </li> </ul> </div> <div class="paragraph"> <p>Kerberos credentials are refreshed every 30 seconds when Canary runs in daemon mode.</p> </div> <div class="paragraph"> <p>To configure the DNS interface for the client, configure the following optional properties in <em>hbase-site.xml</em>.</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.client.dns.interface</code></p> </li> <li> <p><code>hbase.client.dns.nameserver</code></p> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 47. Canary in a Kerberos-Enabled Cluster</div> <div class="content"> <div class="paragraph"> <p>This example shows each of the properties with valid values.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.client.kerberos.principal<span class="tag"></name></span> <span class="tag"><value></span>hbase/_HOST@YOUR-REALM.COM<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.client.keytab.file<span class="tag"></name></span> <span class="tag"><value></span>/etc/hbase/conf/keytab.krb5<span class="tag"></value></span> <span class="tag"></property></span> <span class="comment"></span> property<span class="error">></span> <span class="tag"><name></span>hbase.client.dns.interface<span class="tag"></name></span> <span class="tag"><value></span>default<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.client.dns.nameserver<span class="tag"></name></span> <span class="tag"><value></span>default<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> </div> </div> <div class="sect2"> <h3 id="health.check"><a class="anchor" href="#health.check"></a>125.2. Health Checker</h3> <div class="paragraph"> <p>You can configure HBase to run a script periodically and if it fails N times (configurable), have the server exit. See <em>HBASE-7351 Periodic health check script</em> for configurations and detail.</p> </div> </div> <div class="sect2"> <h3 id="_driver"><a class="anchor" href="#_driver"></a>125.3. Driver</h3> <div class="paragraph"> <p>Several frequently-accessed utilities are provided as <code>Driver</code> classes, and executed by the <em>bin/hbase</em> command. These utilities represent MapReduce jobs which run on your cluster. They are run in the following way, replacing <em>UtilityName</em> with the utility you want to run. This command assumes you have set the environment variable <code>HBASE_HOME</code> to the directory where HBase is unpacked on your server.</p> </div> <div class="listingblock"> <div class="content"> <pre>${HBASE_HOME}/bin/hbase org.apache.hadoop.hbase.mapreduce.UtilityName</pre> </div> </div> <div class="paragraph"> <p>The following utilities are available:</p> </div> <div class="dlist"> <dl> <dt class="hdlist1"><code>LoadIncrementalHFiles</code></dt> <dd> <p>Complete a bulk data load.</p> </dd> <dt class="hdlist1"><code>CopyTable</code></dt> <dd> <p>Export a table from the local cluster to a peer cluster.</p> </dd> <dt class="hdlist1"><code>Export</code></dt> <dd> <p>Write table data to HDFS.</p> </dd> <dt class="hdlist1"><code>Import</code></dt> <dd> <p>Import data written by a previous <code>Export</code> operation.</p> </dd> <dt class="hdlist1"><code>ImportTsv</code></dt> <dd> <p>Import data in TSV format.</p> </dd> <dt class="hdlist1"><code>RowCounter</code></dt> <dd> <p>Count rows in an HBase table.</p> </dd> <dt class="hdlist1"><code>replication.VerifyReplication</code></dt> <dd> <p>Compare the data from tables in two different clusters. WARNING: It doesn’t work for incrementColumnValues’d cells since the timestamp is changed. Note that this command is in a different package than the others.</p> </dd> </dl> </div> <div class="paragraph"> <p>Each command except <code>RowCounter</code> accepts a single <code>--help</code> argument to print usage instructions.</p> </div> </div> <div class="sect2"> <h3 id="hbck"><a class="anchor" href="#hbck"></a>125.4. HBase <code>hbck</code></h3> <div class="paragraph"> <p>To run <code>hbck</code> against your HBase cluster run <code>$./bin/hbase hbck</code>. At the end of the command’s output it prints <code>OK</code> or <code>INCONSISTENCY</code>. If your cluster reports inconsistencies, pass <code>-details</code> to see more detail emitted. If inconsistencies, run <code>hbck</code> a few times because the inconsistency may be transient (e.g. cluster is starting up or a region is splitting). Passing <code>-fix</code> may correct the inconsistency (This is an experimental feature).</p> </div> <div class="paragraph"> <p>For more information, see <a href="#hbck.in.depth">hbck In Depth</a>.</p> </div> </div> <div class="sect2"> <h3 id="hfile_tool2"><a class="anchor" href="#hfile_tool2"></a>125.5. HFile Tool</h3> <div class="paragraph"> <p>See <a href="#hfile_tool">[hfile_tool]</a>.</p> </div> </div> <div class="sect2"> <h3 id="_wal_tools"><a class="anchor" href="#_wal_tools"></a>125.6. WAL Tools</h3> <div class="sect3"> <h4 id="hlog_tool"><a class="anchor" href="#hlog_tool"></a>125.6.1. <code>FSHLog</code> tool</h4> <div class="paragraph"> <p>The main method on <code>FSHLog</code> offers manual split and dump facilities. Pass it WALs or the product of a split, the content of the <em>recovered.edits</em>. directory.</p> </div> <div class="paragraph"> <p>You can get a textual dump of a WAL file content by doing the following:</p> </div> <div class="listingblock"> <div class="content"> <pre> $ ./bin/hbase org.apache.hadoop.hbase.regionserver.wal.FSHLog --dump hdfs://example.org:8020/hbase/.logs/example.org,60020,1283516293161/10.10.21.10%3A60020.1283973724012</pre> </div> </div> <div class="paragraph"> <p>The return code will be non-zero if there are any issues with the file so you can test wholesomeness of file by redirecting <code>STDOUT</code> to <code>/dev/null</code> and testing the program return.</p> </div> <div class="paragraph"> <p>Similarly you can force a split of a log file directory by doing:</p> </div> <div class="listingblock"> <div class="content"> <pre> $ ./bin/hbase org.apache.hadoop.hbase.regionserver.wal.FSHLog --split hdfs://example.org:8020/hbase/.logs/example.org,60020,1283516293161/</pre> </div> </div> <div class="sect4"> <h5 id="hlog_tool.prettyprint"><a class="anchor" href="#hlog_tool.prettyprint"></a>WAL Pretty Printer</h5> <div class="paragraph"> <p>The WAL Pretty Printer is a tool with configurable options to print the contents of a WAL. You can invoke it via the HBase cli with the 'wal' command.</p> </div> <div class="listingblock"> <div class="content"> <pre> $ ./bin/hbase wal hdfs://example.org:8020/hbase/.logs/example.org,60020,1283516293161/10.10.21.10%3A60020.1283973724012</pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">WAL Printing in older versions of HBase</div> <div class="paragraph"> <p>Prior to version 2.0, the WAL Pretty Printer was called the <code>HLogPrettyPrinter</code>, after an internal name for HBase’s write ahead log. In those versions, you can pring the contents of a WAL using the same configuration as above, but with the 'hlog' command.</p> </div> <div class="listingblock"> <div class="content"> <pre> $ ./bin/hbase hlog hdfs://example.org:8020/hbase/.logs/example.org,60020,1283516293161/10.10.21.10%3A60020.1283973724012</pre> </div> </div> </td> </tr> </table> </div> </div> </div> </div> <div class="sect2"> <h3 id="compression.tool"><a class="anchor" href="#compression.tool"></a>125.7. Compression Tool</h3> <div class="paragraph"> <p>See <a href="#compression.test">compression.test</a>.</p> </div> </div> <div class="sect2"> <h3 id="_copytable"><a class="anchor" href="#_copytable"></a>125.8. CopyTable</h3> <div class="paragraph"> <p>CopyTable is a utility that can copy part or of all of a table, either to the same cluster or another cluster. The target table must first exist. The usage is as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase org.apache.hadoop.hbase.mapreduce.CopyTable --help /bin/hbase org.apache.hadoop.hbase.mapreduce.CopyTable --help Usage: CopyTable [general options] [--starttime=X] [--endtime=Y] [--new.name=NEW] [--peer.adr=ADR] <tablename> Options: rs.class hbase.regionserver.class of the peer cluster, specify if different from current cluster rs.impl hbase.regionserver.impl of the peer cluster, startrow the start row stoprow the stop row starttime beginning of the time range (unixtime in millis) without endtime means from starttime to forever endtime end of the time range. Ignored if no starttime specified. versions number of cell versions to copy new.name new table's name peer.adr Address of the peer cluster given in the format hbase.zookeeer.quorum:hbase.zookeeper.client.port:zookeeper.znode.parent families comma-separated list of families to copy To copy from cf1 to cf2, give sourceCfName:destCfName. To keep the same name, just give "cfName" all.cells also copy delete markers and deleted cells Args: tablename Name of the table to copy Examples: To copy 'TestTable' to a cluster that uses replication for a 1 hour window: $ bin/hbase org.apache.hadoop.hbase.mapreduce.CopyTable --starttime=1265875194289 --endtime=1265878794289 --peer.adr=server1,server2,server3:2181:/hbase --families=myOldCf:myNewCf,cf2,cf3 TestTable For performance consider the following general options: It is recommended that you set the following to >=100. A higher value uses more memory but decreases the round trip time to the server and may increase performance. -Dhbase.client.scanner.caching=100 The following should always be set to false, to prevent writing data twice, which may produce inaccurate results. -Dmapred.map.tasks.speculative.execution=false</pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Scanner Caching</div> <div class="paragraph"> <p>Caching for the input Scan is configured via <code>hbase.client.scanner.caching</code> in the job configuration.</p> </div> </td> </tr> </table> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Versions</div> <div class="paragraph"> <p>By default, CopyTable utility only copies the latest version of row cells unless <code>--versions=n</code> is explicitly specified in the command.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>See Jonathan Hsieh’s <a href="http://www.cloudera.com/blog/2012/06/online-hbase-backups-with-copytable-2/">Online HBase Backups with CopyTable</a> blog post for more on <code>CopyTable</code>.</p> </div> </div> <div class="sect2"> <h3 id="_export"><a class="anchor" href="#_export"></a>125.9. Export</h3> <div class="paragraph"> <p>Export is a utility that will dump the contents of table to HDFS in a sequence file. Invoke via:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.Export <tablename> <outputdir> [<versions> [<starttime> [<endtime>]]]</pre> </div> </div> <div class="paragraph"> <p>By default, the <code>Export</code> tool only exports the newest version of a given cell, regardless of the number of versions stored. To export more than one version, replace <strong><em><versions></em></strong> with the desired number of versions.</p> </div> <div class="paragraph"> <p>Note: caching for the input Scan is configured via <code>hbase.client.scanner.caching</code> in the job configuration.</p> </div> </div> <div class="sect2"> <h3 id="_import"><a class="anchor" href="#_import"></a>125.10. Import</h3> <div class="paragraph"> <p>Import is a utility that will load data that has been exported back into HBase. Invoke via:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.Import <tablename> <inputdir></pre> </div> </div> <div class="paragraph"> <p>To import 0.94 exported files in a 0.96 cluster or onwards, you need to set system property "hbase.import.version" when running the import command as below:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase -Dhbase.import.version=0.94 org.apache.hadoop.hbase.mapreduce.Import <tablename> <inputdir></pre> </div> </div> </div> <div class="sect2"> <h3 id="_importtsv"><a class="anchor" href="#_importtsv"></a>125.11. ImportTsv</h3> <div class="paragraph"> <p>ImportTsv is a utility that will load data in TSV format into HBase. It has two distinct usages: loading data from TSV format in HDFS into HBase via Puts, and preparing StoreFiles to be loaded via the <code>completebulkload</code>.</p> </div> <div class="paragraph"> <p>To load data via Puts (i.e., non-bulk loading):</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.ImportTsv -Dimporttsv.columns=a,b,c <tablename> <hdfs-inputdir></pre> </div> </div> <div class="paragraph"> <p>To generate StoreFiles for bulk-loading:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ bin/hbase org.apache.hadoop.hbase.mapreduce.ImportTsv -Dimporttsv.columns=a,b,c -Dimporttsv.bulk.output=hdfs://storefile-outputdir <tablename> <hdfs-data-inputdir></code></pre> </div> </div> <div class="paragraph"> <p>These generated StoreFiles can be loaded into HBase via <a href="#completebulkload">completebulkload</a>.</p> </div> <div class="sect3"> <h4 id="importtsv.options"><a class="anchor" href="#importtsv.options"></a>125.11.1. ImportTsv Options</h4> <div class="paragraph"> <p>Running <code>ImportTsv</code> with no arguments prints brief usage information:</p> </div> <div class="listingblock"> <div class="content"> <pre>Usage: importtsv -Dimporttsv.columns=a,b,c <tablename> <inputdir> Imports the given input directory of TSV data into the specified table. The column names of the TSV data must be specified using the -Dimporttsv.columns option. This option takes the form of comma-separated column names, where each column name is either a simple column family, or a columnfamily:qualifier. The special column name HBASE_ROW_KEY is used to designate that this column should be used as the row key for each imported record. You must specify exactly one column to be the row key, and you must specify a column name for every column that exists in the input data. By default importtsv will load data directly into HBase. To instead generate HFiles of data to prepare for a bulk data load, pass the option: -Dimporttsv.bulk.output=/path/for/output Note: the target table will be created with default column family descriptors if it does not already exist. Other options that may be specified with -D include: -Dimporttsv.skip.bad.lines=false - fail if encountering an invalid line '-Dimporttsv.separator=|' - eg separate on pipes instead of tabs -Dimporttsv.timestamp=currentTimeAsLong - use the specified timestamp for the import -Dimporttsv.mapper.class=my.Mapper - A user-defined Mapper to use instead of org.apache.hadoop.hbase.mapreduce.TsvImporterMapper</pre> </div> </div> </div> <div class="sect3"> <h4 id="importtsv.example"><a class="anchor" href="#importtsv.example"></a>125.11.2. ImportTsv Example</h4> <div class="paragraph"> <p>For example, assume that we are loading data into a table called 'datatsv' with a ColumnFamily called 'd' with two columns "c1" and "c2".</p> </div> <div class="paragraph"> <p>Assume that an input file exists as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre>row1 c1 c2 row2 c1 c2 row3 c1 c2 row4 c1 c2 row5 c1 c2 row6 c1 c2 row7 c1 c2 row8 c1 c2 row9 c1 c2 row10 c1 c2</pre> </div> </div> <div class="paragraph"> <p>For ImportTsv to use this imput file, the command line needs to look like this:</p> </div> <div class="listingblock"> <div class="content"> <pre> HADOOP_CLASSPATH=`${HBASE_HOME}/bin/hbase classpath` ${HADOOP_HOME}/bin/hadoop jar ${HBASE_HOME}/hbase-server-VERSION.jar importtsv -Dimporttsv.columns=HBASE_ROW_KEY,d:c1,d:c2 -Dimporttsv.bulk.output=hdfs://storefileoutput datatsv hdfs://inputfile</pre> </div> </div> <div class="paragraph"> <p>... and in this example the first column is the rowkey, which is why the HBASE_ROW_KEY is used. The second and third columns in the file will be imported as "d:c1" and "d:c2", respectively.</p> </div> </div> <div class="sect3"> <h4 id="importtsv.warning"><a class="anchor" href="#importtsv.warning"></a>125.11.3. ImportTsv Warning</h4> <div class="paragraph"> <p>If you have preparing a lot of data for bulk loading, make sure the target HBase table is pre-split appropriately.</p> </div> </div> <div class="sect3"> <h4 id="importtsv.also"><a class="anchor" href="#importtsv.also"></a>125.11.4. See Also</h4> <div class="paragraph"> <p>For more information about bulk-loading HFiles into HBase, see <a href="#arch.bulk.load">arch.bulk.load</a></p> </div> </div> </div> <div class="sect2"> <h3 id="_completebulkload"><a class="anchor" href="#_completebulkload"></a>125.12. CompleteBulkLoad</h3> <div class="paragraph"> <p>The <code>completebulkload</code> utility will move generated StoreFiles into an HBase table. This utility is often used in conjunction with output from <a href="#importtsv">importtsv</a>.</p> </div> <div class="paragraph"> <p>There are two ways to invoke this utility, with explicit classname and via the driver:</p> </div> <div class="listingblock"> <div class="title">Explicit Classname</div> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.LoadIncrementalHFiles <hdfs://storefileoutput> <tablename></pre> </div> </div> <div class="listingblock"> <div class="title">Driver</div> <div class="content"> <pre>HADOOP_CLASSPATH=`${HBASE_HOME}/bin/hbase classpath` ${HADOOP_HOME}/bin/hadoop jar ${HBASE_HOME}/hbase-server-VERSION.jar completebulkload <hdfs://storefileoutput> <tablename></pre> </div> </div> <div class="sect3"> <h4 id="completebulkload.warning"><a class="anchor" href="#completebulkload.warning"></a>125.12.1. CompleteBulkLoad Warning</h4> <div class="paragraph"> <p>Data generated via MapReduce is often created with file permissions that are not compatible with the running HBase process. Assuming you’re running HDFS with permissions enabled, those permissions will need to be updated before you run CompleteBulkLoad.</p> </div> <div class="paragraph"> <p>For more information about bulk-loading HFiles into HBase, see <a href="#arch.bulk.load">arch.bulk.load</a>.</p> </div> </div> </div> <div class="sect2"> <h3 id="_walplayer"><a class="anchor" href="#_walplayer"></a>125.13. WALPlayer</h3> <div class="paragraph"> <p>WALPlayer is a utility to replay WAL files into HBase.</p> </div> <div class="paragraph"> <p>The WAL can be replayed for a set of tables or all tables, and a timerange can be provided (in milliseconds). The WAL is filtered to this set of tables. The output can optionally be mapped to another set of tables.</p> </div> <div class="paragraph"> <p>WALPlayer can also generate HFiles for later bulk importing, in that case only a single table and no mapping can be specified.</p> </div> <div class="paragraph"> <p>Invoke via:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.WALPlayer [options] <wal inputdir> <tables> [<tableMappings>]></pre> </div> </div> <div class="paragraph"> <p>For example:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.WALPlayer /backuplogdir oldTable1,oldTable2 newTable1,newTable2</pre> </div> </div> <div class="paragraph"> <p>WALPlayer, by default, runs as a mapreduce job. To NOT run WALPlayer as a mapreduce job on your cluster, force it to run all in the local process by adding the flags <code>-Dmapreduce.jobtracker.address=local</code> on the command line.</p> </div> </div> <div class="sect2"> <h3 id="rowcounter"><a class="anchor" href="#rowcounter"></a>125.14. RowCounter and CellCounter</h3> <div class="paragraph"> <p><a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/RowCounter.html">RowCounter</a> is a mapreduce job to count all the rows of a table. This is a good utility to use as a sanity check to ensure that HBase can read all the blocks of a table if there are any concerns of metadata inconsistency. It will run the mapreduce all in a single process but it will run faster if you have a MapReduce cluster in place for it to exploit.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.RowCounter <tablename> [<column1> <column2>...]</pre> </div> </div> <div class="paragraph"> <p>RowCounter only counts one version per cell.</p> </div> <div class="paragraph"> <p>Note: caching for the input Scan is configured via <code>hbase.client.scanner.caching</code> in the job configuration.</p> </div> <div class="paragraph"> <p>HBase ships another diagnostic mapreduce job called <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/mapreduce/CellCounter.html">CellCounter</a>. Like RowCounter, it gathers more fine-grained statistics about your table. The statistics gathered by RowCounter are more fine-grained and include:</p> </div> <div class="ulist"> <ul> <li> <p>Total number of rows in the table.</p> </li> <li> <p>Total number of CFs across all rows.</p> </li> <li> <p>Total qualifiers across all rows.</p> </li> <li> <p>Total occurrence of each CF.</p> </li> <li> <p>Total occurrence of each qualifier.</p> </li> <li> <p>Total number of versions of each qualifier.</p> </li> </ul> </div> <div class="paragraph"> <p>The program allows you to limit the scope of the run. Provide a row regex or prefix to limit the rows to analyze. Use <code>hbase.mapreduce.scan.column.family</code> to specify scanning a single column family.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.mapreduce.CellCounter <tablename> <outputDir> [regex or prefix]</pre> </div> </div> <div class="paragraph"> <p>Note: just like RowCounter, caching for the input Scan is configured via <code>hbase.client.scanner.caching</code> in the job configuration.</p> </div> </div> <div class="sect2"> <h3 id="_mlockall"><a class="anchor" href="#_mlockall"></a>125.15. mlockall</h3> <div class="paragraph"> <p>It is possible to optionally pin your servers in physical memory making them less likely to be swapped out in oversubscribed environments by having the servers call <a href="http://linux.die.net/man/2/mlockall">mlockall</a> on startup. See <a href="https://issues.apache.org/jira/browse/HBASE-4391">HBASE-4391 Add ability to start RS as root and call mlockall</a> for how to build the optional library and have it run on startup.</p> </div> </div> <div class="sect2"> <h3 id="compaction.tool"><a class="anchor" href="#compaction.tool"></a>125.16. Offline Compaction Tool</h3> <div class="paragraph"> <p>See the usage for the <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/regionserver/CompactionTool.html">Compaction Tool</a>. Run it like this ./bin/hbase org.apache.hadoop.hbase.regionserver.CompactionTool</p> </div> </div> <div class="sect2"> <h3 id="__code_hbase_clean_code"><a class="anchor" href="#__code_hbase_clean_code"></a>125.17. <code>hbase clean</code></h3> <div class="paragraph"> <p>The <code>hbase clean</code> command cleans HBase data from ZooKeeper, HDFS, or both. It is appropriate to use for testing. Run it with no options for usage instructions. The <code>hbase clean</code> command was introduced in HBase 0.98.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase clean Usage: hbase clean (--cleanZk|--cleanHdfs|--cleanAll) Options: --cleanZk cleans hbase related data from zookeeper. --cleanHdfs cleans hbase related data from hdfs. --cleanAll cleans hbase related data from both zookeeper and hdfs.</pre> </div> </div> </div> <div class="sect2"> <h3 id="__code_hbase_pe_code"><a class="anchor" href="#__code_hbase_pe_code"></a>125.18. <code>hbase pe</code></h3> <div class="paragraph"> <p>The <code>hbase pe</code> command is a shortcut provided to run the <code>org.apache.hadoop.hbase.PerformanceEvaluation</code> tool, which is used for testing. The <code>hbase pe</code> command was introduced in HBase 0.98.4.</p> </div> <div class="paragraph"> <p>The PerformanceEvaluation tool accepts many different options and commands. For usage instructions, run the command with no options.</p> </div> <div class="paragraph"> <p>To run PerformanceEvaluation prior to HBase 0.98.4, issue the command <code>hbase org.apache.hadoop.hbase.PerformanceEvaluation</code>.</p> </div> <div class="paragraph"> <p>The PerformanceEvaluation tool has received many updates in recent HBase releases, including support for namespaces, support for tags, cell-level ACLs and visibility labels, multiget support for RPC calls, increased sampling sizes, an option to randomly sleep during testing, and ability to "warm up" the cluster before testing starts.</p> </div> </div> <div class="sect2"> <h3 id="__code_hbase_ltt_code"><a class="anchor" href="#__code_hbase_ltt_code"></a>125.19. <code>hbase ltt</code></h3> <div class="paragraph"> <p>The <code>hbase ltt</code> command is a shortcut provided to run the <code>org.apache.hadoop.hbase.util.LoadTestTool</code> utility, which is used for testing. The <code>hbase ltt</code> command was introduced in HBase 0.98.4.</p> </div> <div class="paragraph"> <p>You must specify either <code>-write</code> or <code>-update-read</code> as the first option. For general usage instructions, pass the <code>-h</code> option.</p> </div> <div class="paragraph"> <p>To run LoadTestTool prior to HBase 0.98.4, issue the command hbase org.apache.hadoop.hbase.util.LoadTestTool.</p> </div> <div class="paragraph"> <p>The LoadTestTool has received many updates in recent HBase releases, including support for namespaces, support for tags, cell-level ACLS and visibility labels, testing security-related features, ability to specify the number of regions per server, tests for multi-get RPC calls, and tests relating to replication.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="ops.regionmgt"><a class="anchor" href="#ops.regionmgt"></a>126. Region Management</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="ops.regionmgt.majorcompact"><a class="anchor" href="#ops.regionmgt.majorcompact"></a>126.1. Major Compaction</h3> <div class="paragraph"> <p>Major compactions can be requested via the HBase shell or <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Admin.html#majorCompact%28java.lang.String%29">Admin.majorCompact</a>.</p> </div> <div class="paragraph"> <p>Note: major compactions do NOT do region merges. See <a href="#compaction">compaction</a> for more information about compactions.</p> </div> </div> <div class="sect2"> <h3 id="ops.regionmgt.merge"><a class="anchor" href="#ops.regionmgt.merge"></a>126.2. Merge</h3> <div class="paragraph"> <p>Merge is a utility that can merge adjoining regions in the same table (see org.apache.hadoop.hbase.util.Merge).</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ bin/hbase org.apache.hadoop.hbase.util.Merge <tablename> <region1> <region2></code></pre> </div> </div> <div class="paragraph"> <p>If you feel you have too many regions and want to consolidate them, Merge is the utility you need. Merge must run be done when the cluster is down. See the <a href="http://ofps.oreilly.com/titles/9781449396107/performance.html">O’Reilly HBase Book</a> for an example of usage.</p> </div> <div class="paragraph"> <p>You will need to pass 3 parameters to this application. The first one is the table name. The second one is the fully qualified name of the first region to merge, like "table_name,\x0A,1342956111995.7cef47f192318ba7ccc75b1bbf27a82b.". The third one is the fully qualified name for the second region to merge.</p> </div> <div class="paragraph"> <p>Additionally, there is a Ruby script attached to <a href="https://issues.apache.org/jira/browse/HBASE-1621">HBASE-1621</a> for region merging.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="node.management"><a class="anchor" href="#node.management"></a>127. Node Management</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="decommission"><a class="anchor" href="#decommission"></a>127.1. Node Decommission</h3> <div class="paragraph"> <p>You can stop an individual RegionServer by running the following script in the HBase directory on the particular node:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase-daemon.sh stop regionserver</pre> </div> </div> <div class="paragraph"> <p>The RegionServer will first close all regions and then shut itself down. On shutdown, the RegionServer’s ephemeral node in ZooKeeper will expire. The master will notice the RegionServer gone and will treat it as a 'crashed' server; it will reassign the nodes the RegionServer was carrying.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Disable the Load Balancer before Decommissioning a node</div> <div class="paragraph"> <p>If the load balancer runs while a node is shutting down, then there could be contention between the Load Balancer and the Master’s recovery of the just decommissioned RegionServer. Avoid any problems by disabling the balancer first. See <a href="#lb">lb</a> below.</p> </div> </td> </tr> </table> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Kill Node Tool</div> <div class="paragraph"> <p>In hbase-2.0, in the bin directory, we added a script named <em>considerAsDead.sh</em> that can be used to kill a regionserver. Hardware issues could be detected by specialized monitoring tools before the zookeeper timeout has expired. <em>considerAsDead.sh</em> is a simple function to mark a RegionServer as dead. It deletes all the znodes of the server, starting the recovery process. Plug in the script into your monitoring/fault detection tools to initiate faster failover. Be careful how you use this disruptive tool. Copy the script if you need to make use of it in a version of hbase previous to hbase-2.0.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>A downside to the above stop of a RegionServer is that regions could be offline for a good period of time. Regions are closed in order. If many regions on the server, the first region to close may not be back online until all regions close and after the master notices the RegionServer’s znode gone. In Apache HBase 0.90.2, we added facility for having a node gradually shed its load and then shutdown itself down. Apache HBase 0.90.2 added the <em>graceful_stop.sh</em> script. Here is its usage:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/graceful_stop.sh Usage: graceful_stop.sh [--config &conf-dir>] [--restart] [--reload] [--thrift] [--rest] &hostname> thrift If we should stop/start thrift before/after the hbase stop/start rest If we should stop/start rest before/after the hbase stop/start restart If we should restart after graceful stop reload Move offloaded regions back on to the stopped server debug Move offloaded regions back on to the stopped server hostname Hostname of server we are to stop</pre> </div> </div> <div class="paragraph"> <p>To decommission a loaded RegionServer, run the following: $ ./bin/graceful_stop.sh HOSTNAME where <code>HOSTNAME</code> is the host carrying the RegionServer you would decommission.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">On <code>HOSTNAME</code></div> <div class="paragraph"> <p>The <code>HOSTNAME</code> passed to <em>graceful_stop.sh</em> must match the hostname that hbase is using to identify RegionServers. Check the list of RegionServers in the master UI for how HBase is referring to servers. Its usually hostname but can also be FQDN. Whatever HBase is using, this is what you should pass the <em>graceful_stop.sh</em> decommission script. If you pass IPs, the script is not yet smart enough to make a hostname (or FQDN) of it and so it will fail when it checks if server is currently running; the graceful unloading of regions will not run.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>The <em>graceful_stop.sh</em> script will move the regions off the decommissioned RegionServer one at a time to minimize region churn. It will verify the region deployed in the new location before it will moves the next region and so on until the decommissioned server is carrying zero regions. At this point, the <em>graceful_stop.sh</em> tells the RegionServer <code>stop</code>. The master will at this point notice the RegionServer gone but all regions will have already been redeployed and because the RegionServer went down cleanly, there will be no WAL logs to split.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Load Balancer</div> <div class="paragraph"> <p>It is assumed that the Region Load Balancer is disabled while the <code>graceful_stop</code> script runs (otherwise the balancer and the decommission script will end up fighting over region deployments). Use the shell to disable the balancer:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">001</span>:<span class="integer">0</span>> balance_switch <span class="predefined-constant">false</span> <span class="predefined-constant">true</span> <span class="integer">0</span> row(s) in <span class="float">0.3590</span> seconds</code></pre> </div> </div> <div class="paragraph"> <p>This turns the balancer OFF. To reenable, do:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">001</span>:<span class="integer">0</span>> balance_switch <span class="predefined-constant">true</span> <span class="predefined-constant">false</span> <span class="integer">0</span> row(s) in <span class="float">0.3590</span> seconds</code></pre> </div> </div> <div class="paragraph"> <p>The <code>graceful_stop</code> will check the balancer and if enabled, will turn it off before it goes to work. If it exits prematurely because of error, it will not have reset the balancer. Hence, it is better to manage the balancer apart from <code>graceful_stop</code> reenabling it after you are done w/ graceful_stop.</p> </div> </td> </tr> </table> </div> <div class="sect3"> <h4 id="draining.servers"><a class="anchor" href="#draining.servers"></a>127.1.1. Decommissioning several Regions Servers concurrently</h4> <div class="paragraph"> <p>If you have a large cluster, you may want to decommission more than one machine at a time by gracefully stopping mutiple RegionServers concurrently. To gracefully drain multiple regionservers at the same time, RegionServers can be put into a "draining" state. This is done by marking a RegionServer as a draining node by creating an entry in ZooKeeper under the <em>hbase_root/draining</em> znode. This znode has format <code>name,port,startcode</code> just like the regionserver entries under <em>hbase_root/rs</em> znode.</p> </div> <div class="paragraph"> <p>Without this facility, decommissioning mulitple nodes may be non-optimal because regions that are being drained from one region server may be moved to other regionservers that are also draining. Marking RegionServers to be in the draining state prevents this from happening. See this <a href="http://inchoate-clatter.blogspot.com/2012/03/hbase-ops-automation.html">blog post</a> for more details.</p> </div> </div> <div class="sect3"> <h4 id="bad.disk"><a class="anchor" href="#bad.disk"></a>127.1.2. Bad or Failing Disk</h4> <div class="paragraph"> <p>It is good having <a href="#dfs.datanode.failed.volumes.tolerated">dfs.datanode.failed.volumes.tolerated</a> set if you have a decent number of disks per machine for the case where a disk plain dies. But usually disks do the "John Wayne" — i.e. take a while to go down spewing errors in <em>dmesg</em> — or for some reason, run much slower than their companions. In this case you want to decommission the disk. You have two options. You can <a href="http://wiki.apache.org/hadoop/FAQ#I_want_to_make_a_large_cluster_smaller_by_taking_out_a_bunch_of_nodes_simultaneously._How_can_this_be_done.3F">decommission the datanode</a> or, less disruptive in that only the bad disks data will be rereplicated, can stop the datanode, unmount the bad volume (You can’t umount a volume while the datanode is using it), and then restart the datanode (presuming you have set dfs.datanode.failed.volumes.tolerated > 0). The regionserver will throw some errors in its logs as it recalibrates where to get its data from — it will likely roll its WAL log too — but in general but for some latency spikes, it should keep on chugging.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Short Circuit Reads</div> <div class="paragraph"> <p>If you are doing short-circuit reads, you will have to move the regions off the regionserver before you stop the datanode; when short-circuiting reading, though chmod’d so regionserver cannot have access, because it already has the files open, it will be able to keep reading the file blocks from the bad disk even though the datanode is down. Move the regions back after you restart the datanode.</p> </div> </td> </tr> </table> </div> </div> </div> <div class="sect2"> <h3 id="rolling"><a class="anchor" href="#rolling"></a>127.2. Rolling Restart</h3> <div class="paragraph"> <p>Some cluster configuration changes require either the entire cluster, or the RegionServers, to be restarted in order to pick up the changes. In addition, rolling restarts are supported for upgrading to a minor or maintenance release, and to a major release if at all possible. See the release notes for release you want to upgrade to, to find out about limitations to the ability to perform a rolling upgrade.</p> </div> <div class="paragraph"> <p>There are multiple ways to restart your cluster nodes, depending on your situation. These methods are detailed below.</p> </div> <div class="sect3"> <h4 id="_using_the_code_rolling_restart_sh_code_script"><a class="anchor" href="#_using_the_code_rolling_restart_sh_code_script"></a>127.2.1. Using the <code>rolling-restart.sh</code> Script</h4> <div class="paragraph"> <p>HBase ships with a script, <em>bin/rolling-restart.sh</em>, that allows you to perform rolling restarts on the entire cluster, the master only, or the RegionServers only. The script is provided as a template for your own script, and is not explicitly tested. It requires password-less SSH login to be configured and assumes that you have deployed using a tarball. The script requires you to set some environment variables before running it. Examine the script and modify it to suit your needs.</p> </div> <div class="exampleblock"> <div class="title">Example 48. <em>rolling-restart.sh</em> General Usage</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>$ ./bin/rolling-restart.sh --help Usage: rolling-restart.sh [--config <hbase-confdir>] [--rs-only] [--master-only] [--graceful] [--maxthreads xx]</pre> </div> </div> </div> </div> <div class="dlist"> <dl> <dt class="hdlist1">Rolling Restart on RegionServers Only</dt> <dd> <p>To perform a rolling restart on the RegionServers only, use the <code>--rs-only</code> option. This might be necessary if you need to reboot the individual RegionServer or if you make a configuration change that only affects RegionServers and not the other HBase processes.</p> </dd> <dt class="hdlist1">Rolling Restart on Masters Only</dt> <dd> <p>To perform a rolling restart on the active and backup Masters, use the <code>--master-only</code> option. You might use this if you know that your configuration change only affects the Master and not the RegionServers, or if you need to restart the server where the active Master is running.</p> </dd> <dt class="hdlist1">Graceful Restart</dt> <dd> <p>If you specify the <code>--graceful</code> option, RegionServers are restarted using the <em>bin/graceful_stop.sh</em> script, which moves regions off a RegionServer before restarting it. This is safer, but can delay the restart.</p> </dd> <dt class="hdlist1">Limiting the Number of Threads</dt> <dd> <p>To limit the rolling restart to using only a specific number of threads, use the <code>--maxthreads</code> option.</p> </dd> </dl> </div> </div> <div class="sect3"> <h4 id="rolling.restart.manual"><a class="anchor" href="#rolling.restart.manual"></a>127.2.2. Manual Rolling Restart</h4> <div class="paragraph"> <p>To retain more control over the process, you may wish to manually do a rolling restart across your cluster. This uses the <code>graceful-stop.sh</code> command <a href="#decommission">decommission</a>. In this method, you can restart each RegionServer individually and then move its old regions back into place, retaining locality. If you also need to restart the Master, you need to do it separately, and restart the Master before restarting the RegionServers using this method. The following is an example of such a command. You may need to tailor it to your environment. This script does a rolling restart of RegionServers only. It disables the load balancer before moving the regions.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ for i in `cat conf/regionservers|sort`; do ./bin/graceful_stop.sh --restart --reload --debug $i; done &> /tmp/log.txt &;</pre> </div> </div> <div class="paragraph"> <p>Monitor the output of the <em>/tmp/log.txt</em> file to follow the progress of the script.</p> </div> </div> <div class="sect3"> <h4 id="_logic_for_crafting_your_own_rolling_restart_script"><a class="anchor" href="#_logic_for_crafting_your_own_rolling_restart_script"></a>127.2.3. Logic for Crafting Your Own Rolling Restart Script</h4> <div class="paragraph"> <p>Use the following guidelines if you want to create your own rolling restart script.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Extract the new release, verify its configuration, and synchronize it to all nodes of your cluster using <code>rsync</code>, <code>scp</code>, or another secure synchronization mechanism.</p> </li> <li> <p>Use the hbck utility to ensure that the cluster is consistent.</p> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbck</pre> </div> </div> <div class="paragraph"> <p>Perform repairs if required. See <a href="#hbck">hbck</a> for details.</p> </div> </li> <li> <p>Restart the master first. You may need to modify these commands if your new HBase directory is different from the old one, such as for an upgrade.</p> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase-daemon.sh stop master; ./bin/hbase-daemon.sh start master</pre> </div> </div> </li> <li> <p>Gracefully restart each RegionServer, using a script such as the following, from the Master.</p> <div class="listingblock"> <div class="content"> <pre>$ for i in `cat conf/regionservers|sort`; do ./bin/graceful_stop.sh --restart --reload --debug $i; done &> /tmp/log.txt &</pre> </div> </div> <div class="paragraph"> <p>If you are running Thrift or REST servers, pass the --thrift or --rest options. For other available options, run the <code>bin/graceful-stop.sh --help</code> command.</p> </div> <div class="paragraph"> <p>It is important to drain HBase regions slowly when restarting multiple RegionServers. Otherwise, multiple regions go offline simultaneously and must be reassigned to other nodes, which may also go offline soon. This can negatively affect performance. You can inject delays into the script above, for instance, by adding a Shell command such as <code>sleep</code>. To wait for 5 minutes between each RegionServer restart, modify the above script to the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ for i in `cat conf/regionservers|sort`; do ./bin/graceful_stop.sh --restart --reload --debug $i & sleep 5m; done &> /tmp/log.txt &</pre> </div> </div> </li> <li> <p>Restart the Master again, to clear out the dead servers list and re-enable the load balancer.</p> </li> <li> <p>Run the <code>hbck</code> utility again, to be sure the cluster is consistent.</p> </li> </ol> </div> </div> </div> <div class="sect2"> <h3 id="adding.new.node"><a class="anchor" href="#adding.new.node"></a>127.3. Adding a New Node</h3> <div class="paragraph"> <p>Adding a new regionserver in HBase is essentially free, you simply start it like this: <code>$ ./bin/hbase-daemon.sh start regionserver</code> and it will register itself with the master. Ideally you also started a DataNode on the same machine so that the RS can eventually start to have local files. If you rely on ssh to start your daemons, don’t forget to add the new hostname in <em>conf/regionservers</em> on the master.</p> </div> <div class="paragraph"> <p>At this point the region server isn’t serving data because no regions have moved to it yet. If the balancer is enabled, it will start moving regions to the new RS. On a small/medium cluster this can have a very adverse effect on latency as a lot of regions will be offline at the same time. It is thus recommended to disable the balancer the same way it’s done when decommissioning a node and move the regions manually (or even better, using a script that moves them one by one).</p> </div> <div class="paragraph"> <p>The moved regions will all have 0% locality and won’t have any blocks in cache so the region server will have to use the network to serve requests. Apart from resulting in higher latency, it may also be able to use all of your network card’s capacity. For practical purposes, consider that a standard 1GigE NIC won’t be able to read much more than <em>100MB/s</em>. In this case, or if you are in a OLAP environment and require having locality, then it is recommended to major compact the moved regions.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_hbase_metrics"><a class="anchor" href="#_hbase_metrics"></a>128. HBase Metrics</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase emits metrics which adhere to the <a href="http://hadoop.apache.org/core/docs/current/api/org/apache/hadoop/metrics/package-summary.html">Hadoop metrics</a> API. Starting with HBase 0.95<sup class="footnote">[<a id="_footnoteref_4" class="footnote" href="#_footnote_4" title="View footnote.">4</a>]</sup>, HBase is configured to emit a default set of metrics with a default sampling period of every 10 seconds. You can use HBase metrics in conjunction with Ganglia. You can also filter which metrics are emitted and extend the metrics framework to capture custom metrics appropriate for your environment.</p> </div> <div class="sect2"> <h3 id="_metric_setup"><a class="anchor" href="#_metric_setup"></a>128.1. Metric Setup</h3> <div class="paragraph"> <p>For HBase 0.95 and newer, HBase ships with a default metrics configuration, or <em class="firstterm">sink</em>. This includes a wide variety of individual metrics, and emits them every 10 seconds by default. To configure metrics for a given region server, edit the <em>conf/hadoop-metrics2-hbase.properties</em> file. Restart the region server for the changes to take effect.</p> </div> <div class="paragraph"> <p>To change the sampling rate for the default sink, edit the line beginning with <code>*.period</code>. To filter which metrics are emitted or to extend the metrics framework, see link:http://hadoop.apache.org/docs/current/api/org/apache/hadoop/metrics2/package-summary.html</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">HBase Metrics and Ganglia</div> <div class="paragraph"> <p>By default, HBase emits a large number of metrics per region server. Ganglia may have difficulty processing all these metrics. Consider increasing the capacity of the Ganglia server or reducing the number of metrics emitted by HBase. See <a href="http://hadoop.apache.org/docs/current/api/org/apache/hadoop/metrics2/package-summary.html#filtering">Metrics Filtering</a>.</p> </div> </td> </tr> </table> </div> </div> <div class="sect2"> <h3 id="_disabling_metrics"><a class="anchor" href="#_disabling_metrics"></a>128.2. Disabling Metrics</h3> <div class="paragraph"> <p>To disable metrics for a region server, edit the <em>conf/hadoop-metrics2-hbase.properties</em> file and comment out any uncommented lines. Restart the region server for the changes to take effect.</p> </div> </div> <div class="sect2"> <h3 id="discovering.available.metrics"><a class="anchor" href="#discovering.available.metrics"></a>128.3. Discovering Available Metrics</h3> <div class="paragraph"> <p>Rather than listing each metric which HBase emits by default, you can browse through the available metrics, either as a JSON output or via JMX. Different metrics are exposed for the Master process and each region server process.</p> </div> <div class="olist arabic"> <div class="title">Procedure: Access a JSON Output of Available Metrics</div> <ol class="arabic"> <li> <p>After starting HBase, access the region server’s web UI, at <code><a href="http://REGIONSERVER_HOSTNAME:60030" class="bare">http://REGIONSERVER_HOSTNAME:60030</a></code> by default (or port 16030 in HBase 1.0+).</p> </li> <li> <p>Click the <span class="label">Metrics Dump</span> link near the top. The metrics for the region server are presented as a dump of the JMX bean in JSON format. This will dump out all metrics names and their values. To include metrics descriptions in the listing — this can be useful when you are exploring what is available — add a query string of <code>?description=true</code> so your URL becomes <code><a href="http://REGIONSERVER_HOSTNAME:60030/jmx?description=true" class="bare">http://REGIONSERVER_HOSTNAME:60030/jmx?description=true</a></code>. Not all beans and attributes have descriptions.</p> </li> <li> <p>To view metrics for the Master, connect to the Master’s web UI instead (defaults to <code><a href="http://localhost:60010" class="bare">http://localhost:60010</a></code> or port 16010 in HBase 1.0+) and click its <span class="label">Metrics Dump</span> link. To include metrics descriptions in the listing — this can be useful when you are exploring what is available — add a query string of <code>?description=true</code> so your URL becomes <code><a href="http://REGIONSERVER_HOSTNAME:60010/jmx?description=true" class="bare">http://REGIONSERVER_HOSTNAME:60010/jmx?description=true</a></code>. Not all beans and attributes have descriptions.</p> </li> </ol> </div> <div class="paragraph"> <p>You can use many different tools to view JMX content by browsing MBeans. This procedure uses <code>jvisualvm</code>, which is an application usually available in the JDK.</p> </div> <div class="olist arabic"> <div class="title">Procedure: Browse the JMX Output of Available Metrics</div> <ol class="arabic"> <li> <p>Start HBase, if it is not already running.</p> </li> <li> <p>Run the command <code>jvisualvm</code> command on a host with a GUI display. You can launch it from the command line or another method appropriate for your operating system.</p> </li> <li> <p>Be sure the <span class="label">VisualVM-MBeans</span> plugin is installed. Browse to <strong>Tools → Plugins</strong>. Click <span class="label">Installed</span> and check whether the plugin is listed. If not, click <span class="label">Available Plugins</span>, select it, and click <b class="button">Install</b>. When finished, click <b class="button">Close</b>.</p> </li> <li> <p>To view details for a given HBase process, double-click the process in the <span class="label">Local</span> sub-tree in the left-hand panel. A detailed view opens in the right-hand panel. Click the <span class="label">MBeans</span> tab which appears as a tab in the top of the right-hand panel.</p> </li> <li> <p>To access the HBase metrics, navigate to the appropriate sub-bean:</p> </li> <li> <p>The name of each metric and its current value is displayed in the <span class="label">Attributes</span> tab. For a view which includes more details, including the description of each attribute, click the <span class="label">Metadata</span> tab.</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="_units_of_measure_for_metrics"><a class="anchor" href="#_units_of_measure_for_metrics"></a>128.4. Units of Measure for Metrics</h3> <div class="paragraph"> <p>Different metrics are expressed in different units, as appropriate. Often, the unit of measure is in the name (as in the metric <code>shippedKBs</code>). Otherwise, use the following guidelines. When in doubt, you may need to examine the source for a given metric.</p> </div> <div class="ulist"> <ul> <li> <p>Metrics that refer to a point in time are usually expressed as a timestamp.</p> </li> <li> <p>Metrics that refer to an age (such as <code>ageOfLastShippedOp</code>) are usually expressed in milliseconds.</p> </li> <li> <p>Metrics that refer to memory sizes are in bytes.</p> </li> <li> <p>Sizes of queues (such as <code>sizeOfLogQueue</code>) are expressed as the number of items in the queue. Determine the size by multiplying by the block size (default is 64 MB in HDFS).</p> </li> <li> <p>Metrics that refer to things like the number of a given type of operations (such as <code>logEditsRead</code>) are expressed as an integer.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="master_metrics"><a class="anchor" href="#master_metrics"></a>128.5. Most Important Master Metrics</h3> <div class="paragraph"> <p>Note: Counts are usually over the last metrics reporting interval.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">hbase.master.numRegionServers</dt> <dd> <p>Number of live regionservers</p> </dd> <dt class="hdlist1">hbase.master.numDeadRegionServers</dt> <dd> <p>Number of dead regionservers</p> </dd> <dt class="hdlist1">hbase.master.ritCount </dt> <dd> <p>The number of regions in transition</p> </dd> <dt class="hdlist1">hbase.master.ritCountOverThreshold</dt> <dd> <p>The number of regions that have been in transition longer than a threshold time (default: 60 seconds)</p> </dd> <dt class="hdlist1">hbase.master.ritOldestAge</dt> <dd> <p>The age of the longest region in transition, in milliseconds</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="rs_metrics"><a class="anchor" href="#rs_metrics"></a>128.6. Most Important RegionServer Metrics</h3> <div class="paragraph"> <p>Note: Counts are usually over the last metrics reporting interval.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">hbase.regionserver.regionCount</dt> <dd> <p>The number of regions hosted by the regionserver</p> </dd> <dt class="hdlist1">hbase.regionserver.storeFileCount</dt> <dd> <p>The number of store files on disk currently managed by the regionserver</p> </dd> <dt class="hdlist1">hbase.regionserver.storeFileSize</dt> <dd> <p>Aggregate size of the store files on disk</p> </dd> <dt class="hdlist1">hbase.regionserver.hlogFileCount</dt> <dd> <p>The number of write ahead logs not yet archived</p> </dd> <dt class="hdlist1">hbase.regionserver.totalRequestCount</dt> <dd> <p>The total number of requests received</p> </dd> <dt class="hdlist1">hbase.regionserver.readRequestCount</dt> <dd> <p>The number of read requests received</p> </dd> <dt class="hdlist1">hbase.regionserver.writeRequestCount</dt> <dd> <p>The number of write requests received</p> </dd> <dt class="hdlist1">hbase.regionserver.numOpenConnections</dt> <dd> <p>The number of open connections at the RPC layer</p> </dd> <dt class="hdlist1">hbase.regionserver.numActiveHandler</dt> <dd> <p>The number of RPC handlers actively servicing requests</p> </dd> <dt class="hdlist1">hbase.regionserver.numCallsInGeneralQueue</dt> <dd> <p>The number of currently enqueued user requests</p> </dd> <dt class="hdlist1">hbase.regionserver.numCallsInReplicationQueue</dt> <dd> <p>The number of currently enqueued operations received from replication</p> </dd> <dt class="hdlist1">hbase.regionserver.numCallsInPriorityQueue</dt> <dd> <p>The number of currently enqueued priority (internal housekeeping) requests</p> </dd> <dt class="hdlist1">hbase.regionserver.flushQueueLength</dt> <dd> <p>Current depth of the memstore flush queue. If increasing, we are falling behind with clearing memstores out to HDFS.</p> </dd> <dt class="hdlist1">hbase.regionserver.updatesBlockedTime</dt> <dd> <p>Number of milliseconds updates have been blocked so the memstore can be flushed</p> </dd> <dt class="hdlist1">hbase.regionserver.compactionQueueLength</dt> <dd> <p>Current depth of the compaction request queue. If increasing, we are falling behind with storefile compaction.</p> </dd> <dt class="hdlist1">hbase.regionserver.blockCacheHitCount</dt> <dd> <p>The number of block cache hits</p> </dd> <dt class="hdlist1">hbase.regionserver.blockCacheMissCount</dt> <dd> <p>The number of block cache misses</p> </dd> <dt class="hdlist1">hbase.regionserver.blockCacheExpressHitPercent </dt> <dd> <p>The percent of the time that requests with the cache turned on hit the cache</p> </dd> <dt class="hdlist1">hbase.regionserver.percentFilesLocal</dt> <dd> <p>Percent of store file data that can be read from the local DataNode, 0-100</p> </dd> <dt class="hdlist1">hbase.regionserver.<op>_<measure></dt> <dd> <p>Operation latencies, where <op> is one of Append, Delete, Mutate, Get, Replay, Increment; and where <measure> is one of min, max, mean, median, 75th_percentile, 95th_percentile, 99th_percentile</p> </dd> <dt class="hdlist1">hbase.regionserver.slow<op>Count </dt> <dd> <p>The number of operations we thought were slow, where <op> is one of the list above</p> </dd> <dt class="hdlist1">hbase.regionserver.GcTimeMillis</dt> <dd> <p>Time spent in garbage collection, in milliseconds</p> </dd> <dt class="hdlist1">hbase.regionserver.GcTimeMillisParNew</dt> <dd> <p>Time spent in garbage collection of the young generation, in milliseconds</p> </dd> <dt class="hdlist1">hbase.regionserver.GcTimeMillisConcurrentMarkSweep</dt> <dd> <p>Time spent in garbage collection of the old generation, in milliseconds</p> </dd> <dt class="hdlist1">hbase.regionserver.authenticationSuccesses</dt> <dd> <p>Number of client connections where authentication succeeded</p> </dd> <dt class="hdlist1">hbase.regionserver.authenticationFailures</dt> <dd> <p>Number of client connection authentication failures</p> </dd> <dt class="hdlist1">hbase.regionserver.mutationsWithoutWALCount </dt> <dd> <p>Count of writes submitted with a flag indicating they should bypass the write ahead log</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="rs_meta_metrics"><a class="anchor" href="#rs_meta_metrics"></a>128.7. Meta Table Load Metrics</h3> <div class="paragraph"> <p>HBase meta table metrics collection feature is available in HBase 1.4+ but it is disabled by default, as it can affect the performance of the cluster. When it is enabled, it helps to monitor client access patterns by collecting the following statistics:</p> </div> <div class="ulist"> <ul> <li> <p>number of get, put and delete operations on the <code>hbase:meta</code> table</p> </li> <li> <p>number of get, put and delete operations made by the top-N clients</p> </li> <li> <p>number of operations related to each table</p> </li> <li> <p>number of operations related to the top-N regions</p> <div class="dlist"> <dl> <dt class="hdlist1">When to use the feature</dt> <dd> <p>This feature can help to identify hot spots in the meta table by showing the regions or tables where the meta info is modified (e.g. by create, drop, split or move tables) or retrieved most frequently. It can also help to find misbehaving client applications by showing which clients are using the meta table most heavily, which can for example suggest the lack of meta table buffering or the lack of re-using open client connections in the client application.</p> </dd> </dl> </div> </li> </ul> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> <div class="title">Possible side-effects of enabling this feature</div> <div class="paragraph"> <p>Having large number of clients and regions in the cluster can cause the registration and tracking of a large amount of metrics, which can increase the memory and CPU footprint of the HBase region server handling the <code>hbase:meta</code> table. It can also cause the significant increase of the JMX dump size, which can affect the monitoring or log aggregation system you use beside HBase. It is recommended to turn on this feature only during debugging.</p> </div> </td> </tr> </table> </div> <div class="dlist"> <dl> <dt class="hdlist1">Where to find the metrics in JMX</dt> <dd> <p>Each metric attribute name will start with the ‘MetaTable_’ prefix. For all the metrics you will see five different JMX attributes: count, mean rate, 1 minute rate, 5 minute rate and 15 minute rate. You will find these metrics in JMX under the following MBean: <code>Hadoop → HBase → RegionServer → Coprocessor.Region.CP_org.apache.hadoop.hbase.coprocessor.MetaTableMetrics</code>.</p> </dd> </dl> </div> <div class="listingblock"> <div class="title">Examples: some Meta Table metrics you can see in your JMX dump</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="json">{ <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_get_request_count</span><span class="delimiter">"</span></span>: <span class="integer">77309</span>, <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_put_request_mean_rate</span><span class="delimiter">"</span></span>: <span class="float">0.06339092997186495</span>, <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_table_MyTestTable_request_15min_rate</span><span class="delimiter">"</span></span>: <span class="float">1.1020599841623246</span>, <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_client_/172.30.65.42_lossy_request_count</span><span class="delimiter">"</span></span>: <span class="integer">1786</span> <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_client_/172.30.65.45_put_request_5min_rate</span><span class="delimiter">"</span></span>: <span class="float">0.6189810954855728</span>, <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_region_1561131112259.c66e4308d492936179352c80432ccfe0._lossy_request_count</span><span class="delimiter">"</span></span>: <span class="integer">38342</span>, <span class="key"><span class="delimiter">"</span><span class="content">MetaTable_region_1561131043640.5bdffe4b9e7e334172065c853cf0caa6._lossy_request_1min_rate</span><span class="delimiter">"</span></span>: <span class="float">0.04925099917433935</span>, }</code></pre> </div> </div> <div class="dlist"> <dl> <dt class="hdlist1">Configuration</dt> <dd> <p>To turn on this feature, you have to enable a custom coprocessor by adding the following section to hbase-site.xml. This coprocessor will run on all the HBase RegionServers, but will be active (i.e. consume memory / CPU) only on the server, where the <code>hbase:meta</code> table is located. It will produce JMX metrics which can be downloaded from the web UI of the given RegionServer or by a simple REST call. These metrics will not be present in the JMX dump of the other RegionServers.</p> </dd> </dl> </div> <div class="listingblock"> <div class="title">Enabling the Meta Table Metrics feature</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.coprocessor.region.classes<span class="tag"></name></span> <span class="tag"><value></span>org.apache.hadoop.hbase.coprocessor.MetaTableMetrics<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">How the top-N metrics are calculated?</div> <div class="paragraph"> <p>The 'top-N' type of metrics will be counted using the Lossy Counting Algorithm (as defined in <a href="http://www.vldb.org/conf/2002/S10P03.pdf">Motwani, R; Manku, G.S (2002). "Approximate frequency counts over data streams"</a>), which is designed to identify elements in a data stream whose frequency count exceed a user-given threshold. The frequency computed by this algorithm is not always accurate but has an error threshold that can be specified by the user as a configuration parameter. The run time space required by the algorithm is inversely proportional to the specified error threshold, hence larger the error parameter, the smaller the footprint and the less accurate are the metrics.</p> </div> <div class="paragraph"> <p>You can specify the error rate of the algorithm as a floating-point value between 0 and 1 (exclusive), it’s default value is 0.02. Having the error rate set to <code>E</code> and having <code>N</code> as the total number of meta table operations, then (assuming the uniform distribution of the activity of low frequency elements) at most <code>7 / E</code> meters will be kept and each kept element will have a frequency higher than <code>E * N</code>.</p> </div> <div class="paragraph"> <p>An example: Let’s assume we are interested in the HBase clients that are most active in accessing the meta table. When there was 1,000,000 operations on the meta table so far and the error rate parameter is set to 0.02, then we can assume that only at most 350 client IP address related counters will be present in JMX and each of these clients accessed the meta table at least 20,000 times.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.util.default.lossycounting.errorrate<span class="tag"></name></span> <span class="tag"><value></span>0.02<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </td> </tr> </table> </div> </div> </div> </div> <div class="sect1"> <h2 id="ops.monitoring"><a class="anchor" href="#ops.monitoring"></a>129. HBase Monitoring</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="ops.monitoring.overview"><a class="anchor" href="#ops.monitoring.overview"></a>129.1. Overview</h3> <div class="paragraph"> <p>The following metrics are arguably the most important to monitor for each RegionServer for "macro monitoring", preferably with a system like <a href="http://opentsdb.net/">OpenTSDB</a>. If your cluster is having performance issues it’s likely that you’ll see something unusual with this group.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">HBase</dt> <dd> <div class="ulist"> <ul> <li> <p>See <a href="#rs_metrics">rs metrics</a></p> </li> </ul> </div> </dd> <dt class="hdlist1">OS</dt> <dd> <div class="ulist"> <ul> <li> <p>IO Wait</p> </li> <li> <p>User CPU</p> </li> </ul> </div> </dd> <dt class="hdlist1">Java</dt> <dd> <div class="ulist"> <ul> <li> <p>GC</p> </li> </ul> </div> </dd> </dl> </div> <div class="paragraph"> <p>For more information on HBase metrics, see <a href="#hbase_metrics">hbase metrics</a>.</p> </div> </div> <div class="sect2"> <h3 id="ops.slow.query"><a class="anchor" href="#ops.slow.query"></a>129.2. Slow Query Log</h3> <div class="paragraph"> <p>The HBase slow query log consists of parseable JSON structures describing the properties of those client operations (Gets, Puts, Deletes, etc.) that either took too long to run, or produced too much output. The thresholds for "too long to run" and "too much output" are configurable, as described below. The output is produced inline in the main region server logs so that it is easy to discover further details from context with other logged events. It is also prepended with identifying tags <code>(responseTooSlow)</code>, <code>(responseTooLarge)</code>, <code>(operationTooSlow)</code>, and <code>(operationTooLarge)</code> in order to enable easy filtering with grep, in case the user desires to see only slow queries.</p> </div> <div class="sect3"> <h4 id="_configuration"><a class="anchor" href="#_configuration"></a>129.2.1. Configuration</h4> <div class="paragraph"> <p>There are two configuration knobs that can be used to adjust the thresholds for when queries are logged.</p> </div> <div class="ulist"> <ul> <li> <p><code>hbase.ipc.warn.response.time</code> Maximum number of milliseconds that a query can be run without being logged. Defaults to 10000, or 10 seconds. Can be set to -1 to disable logging by time.</p> </li> <li> <p><code>hbase.ipc.warn.response.size</code> Maximum byte size of response that a query can return without being logged. Defaults to 100 megabytes. Can be set to -1 to disable logging by size.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="_metrics"><a class="anchor" href="#_metrics"></a>129.2.2. Metrics</h4> <div class="paragraph"> <p>The slow query log exposes to metrics to JMX.</p> </div> <div class="ulist"> <ul> <li> <p><code>hadoop.regionserver_rpc_slowResponse</code> a global metric reflecting the durations of all responses that triggered logging.</p> </li> <li> <p><code>hadoop.regionserver_rpc_methodName.aboveOneSec</code> A metric reflecting the durations of all responses that lasted for more than one second.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="_output"><a class="anchor" href="#_output"></a>129.2.3. Output</h4> <div class="paragraph"> <p>The output is tagged with operation e.g. <code>(operationTooSlow)</code> if the call was a client operation, such as a Put, Get, or Delete, which we expose detailed fingerprint information for. If not, it is tagged <code>(responseTooSlow)</code> and still produces parseable JSON output, but with less verbose information solely regarding its duration and size in the RPC itself. <code>TooLarge</code> is substituted for <code>TooSlow</code> if the response size triggered the logging, with <code>TooLarge</code> appearing even in the case that both size and duration triggered logging.</p> </div> </div> <div class="sect3"> <h4 id="_example"><a class="anchor" href="#_example"></a>129.2.4. Example</h4> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="integer">2011</span>-<span class="integer">09</span>-<span class="integer">08</span> <span class="integer">10</span>:<span class="octal">01</span>:<span class="integer">25</span>,<span class="integer">824</span> WARN org.apache.hadoop.ipc.HBaseServer: (operationTooSlow): {<span class="string"><span class="delimiter">"</span><span class="content">tables</span><span class="delimiter">"</span></span>:{<span class="string"><span class="delimiter">"</span><span class="content">riley2</span><span class="delimiter">"</span></span>:{<span class="string"><span class="delimiter">"</span><span class="content">puts</span><span class="delimiter">"</span></span>:[{<span class="string"><span class="delimiter">"</span><span class="content">totalColumns</span><span class="delimiter">"</span></span>:<span class="integer">11</span>,<span class="string"><span class="delimiter">"</span><span class="content">families</span><span class="delimiter">"</span></span>:{<span class="string"><span class="delimiter">"</span><span class="content">actions</span><span class="delimiter">"</span></span>:[{<span class="string"><span class="delimiter">"</span><span class="content">timestamp</span><span class="delimiter">"</span></span>:<span class="integer">1315501284459</span>,<span class="string"><span class="delimiter">"</span><span class="content">qualifier</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">0</span><span class="delimiter">"</span></span>,<span class="string"><span class="delimiter">"</span><span class="content">vlen</span><span class="delimiter">"</span></span>:<span class="integer">9667580</span>},{<span class="string"><span class="delimiter">"</span><span class="content">timestamp</span><span class="delimiter">"</span></span>:<span class="integer">1315501284459</span>,<span class="string"><span class="delimiter">"</span><span class="content">qualifier</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">1</span><span class="delimiter">"</span></span>,<span class="string"><span class="delimiter">"</span><span class="content">vlen</span><span class="delimiter">"</span></span>:<span class="integer">10122412</span>},{<span class="string"><span class="delimiter">"</span><span class="content">timestamp</span><span class="delimiter">"</span></span>:<span class="integer">1315501284459</span>,<span class="string"><span class="delimiter">"</span><span class="content">qualifier</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">2</span><span class="delimiter">"</span></span>,<span class="string"><span class="delimiter">"</span><span class="content">vlen</span><span class="delimiter">"</span></span>:<span class="integer">11104617</span>},{<span class="string"><span class="delimiter">"</span><span class="content">timestamp</span><span class="delimiter">"</span></span>:<span class="integer">1315501284459</span>,<span class="string"><span class="delimiter">"</span><span class="content">qualifier</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">3</span><span class="delimiter">"</span></span>,<span class="string"><span class="delimiter">"</span><span class="content">vlen</span><span class="delimiter">"</span></span>:<span class="integer">13430635</span>}]},<span class="string"><span class="delimiter">"</span><span class="content">row</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">cfcd208495d565ef66e7dff9f98764da:0</span><span class="delimiter">"</span></span>}],<span class="string"><span class="delimiter">"</span><span class="content">families</span><span class="delimiter">"</span></span>:[<span class="string"><span class="delimiter">"</span><span class="content">actions</span><span class="delimiter">"</span></span>]}},<span class="string"><span class="delimiter">"</span><span class="content">processingtimems</span><span class="delimiter">"</span></span>:<span class="integer">956</span>,<span class="string"><span class="delimiter">"</span><span class="content">client</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">10.47.34.63:33623</span><span class="delimiter">"</span></span>,<span class="string"><span class="delimiter">"</span><span class="content">starttimems</span><span class="delimiter">"</span></span>:<span class="integer">1315501284456</span>,<span class="string"><span class="delimiter">"</span><span class="content">queuetimems</span><span class="delimiter">"</span></span>:<span class="integer">0</span>,<span class="string"><span class="delimiter">"</span><span class="content">totalPuts</span><span class="delimiter">"</span></span>:<span class="integer">1</span>,<span class="string"><span class="delimiter">"</span><span class="content">class</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">HRegionServer</span><span class="delimiter">"</span></span>,<span class="string"><span class="delimiter">"</span><span class="content">responsesize</span><span class="delimiter">"</span></span>:<span class="integer">0</span>,<span class="string"><span class="delimiter">"</span><span class="content">method</span><span class="delimiter">"</span></span>:<span class="string"><span class="delimiter">"</span><span class="content">multiPut</span><span class="delimiter">"</span></span>}</code></pre> </div> </div> <div class="paragraph"> <p>Note that everything inside the "tables" structure is output produced by MultiPut’s fingerprint, while the rest of the information is RPC-specific, such as processing time and client IP/port. Other client operations follow the same pattern and the same general structure, with necessary differences due to the nature of the individual operations. In the case that the call is not a client operation, that detailed fingerprint information will be completely absent.</p> </div> <div class="paragraph"> <p>This particular example, for example, would indicate that the likely cause of slowness is simply a very large (on the order of 100MB) multiput, as we can tell by the "vlen," or value length, fields of each put in the multiPut.</p> </div> </div> </div> <div class="sect2"> <h3 id="_block_cache_monitoring"><a class="anchor" href="#_block_cache_monitoring"></a>129.3. Block Cache Monitoring</h3> <div class="paragraph"> <p>Starting with HBase 0.98, the HBase Web UI includes the ability to monitor and report on the performance of the block cache. To view the block cache reports, click . Following are a few examples of the reporting capabilities.</p> </div> <div class="imageblock"> <div class="content"> <img src="images/bc_basic.png" alt="bc basic"> </div> <div class="title">Figure 5. Basic Info</div> </div> <div class="imageblock"> <div class="content"> <img src="images/bc_config.png" alt="bc config"> </div> <div class="title">Figure 6. Config</div> </div> <div class="imageblock"> <div class="content"> <img src="images/bc_stats.png" alt="bc stats"> </div> <div class="title">Figure 7. Stats</div> </div> <div class="imageblock"> <div class="content"> <img src="images/bc_l1.png" alt="bc l1"> </div> <div class="title">Figure 8. L1 and L2</div> </div> <div class="paragraph"> <p>This is not an exhaustive list of all the screens and reports available. Have a look in the Web UI.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_cluster_replication"><a class="anchor" href="#_cluster_replication"></a>130. Cluster Replication</h2> <div class="sectionbody"> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> This information was previously available at <a href="http://hbase.apache.org/replication.html">Cluster Replication</a>. </td> </tr> </table> </div> <div class="paragraph"> <p>HBase provides a cluster replication mechanism which allows you to keep one cluster’s state synchronized with that of another cluster, using the write-ahead log (WAL) of the source cluster to propagate the changes. Some use cases for cluster replication include:</p> </div> <div class="ulist"> <ul> <li> <p>Backup and disaster recovery</p> </li> <li> <p>Data aggregation</p> </li> <li> <p>Geographic data distribution</p> </li> <li> <p>Online data ingestion combined with offline data analytics</p> </li> </ul> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Replication is enabled at the granularity of the column family. Before enabling replication for a column family, create the table and all column families to be replicated, on the destination cluster. </td> </tr> </table> </div> <div class="sect2"> <h3 id="_replication_overview"><a class="anchor" href="#_replication_overview"></a>130.1. Replication Overview</h3> <div class="paragraph"> <p>Cluster replication uses a source-push methodology. An HBase cluster can be a source (also called master or active, meaning that it is the originator of new data), a destination (also called slave or passive, meaning that it receives data via replication), or can fulfill both roles at once. Replication is asynchronous, and the goal of replication is eventual consistency. When the source receives an edit to a column family with replication enabled, that edit is propagated to all destination clusters using the WAL for that for that column family on the RegionServer managing the relevant region.</p> </div> <div class="paragraph"> <p>When data is replicated from one cluster to another, the original source of the data is tracked via a cluster ID which is part of the metadata. In HBase 0.96 and newer (<a href="https://issues.apache.org/jira/browse/HBASE-7709">HBASE-7709</a>), all clusters which have already consumed the data are also tracked. This prevents replication loops.</p> </div> <div class="paragraph"> <p>The WALs for each region server must be kept in HDFS as long as they are needed to replicate data to any slave cluster. Each region server reads from the oldest log it needs to replicate and keeps track of its progress processing WALs inside ZooKeeper to simplify failure recovery. The position marker which indicates a slave cluster’s progress, as well as the queue of WALs to process, may be different for every slave cluster.</p> </div> <div class="paragraph"> <p>The clusters participating in replication can be of different sizes. The master cluster relies on randomization to attempt to balance the stream of replication on the slave clusters. It is expected that the slave cluster has storage capacity to hold the replicated data, as well as any data it is responsible for ingesting. If a slave cluster does run out of room, or is inaccessible for other reasons, it throws an error and the master retains the WAL and retries the replication at intervals.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Terminology Changes</div> <div class="paragraph"> <p>Previously, terms such as <em class="firstterm">master-master</em>, <em class="firstterm">master-slave</em>, and <em class="firstterm">cyclical</em> were used to describe replication relationships in HBase. These terms added confusion, and have been abandoned in favor of discussions about cluster topologies appropriate for different scenarios.</p> </div> </td> </tr> </table> </div> <div class="ulist"> <div class="title">Cluster Topologies</div> <ul> <li> <p>A central source cluster might propagate changes out to multiple destination clusters, for failover or due to geographic distribution.</p> </li> <li> <p>A source cluster might push changes to a destination cluster, which might also push its own changes back to the original cluster.</p> </li> <li> <p>Many different low-latency clusters might push changes to one centralized cluster for backup or resource-intensive data analytics jobs. The processed data might then be replicated back to the low-latency clusters.</p> </li> </ul> </div> <div class="paragraph"> <p>Multiple levels of replication may be chained together to suit your organization’s needs. The following diagram shows a hypothetical scenario. Use the arrows to follow the data paths.</p> </div> <div class="imageblock"> <div class="content"> <img src="images/hbase_replication_diagram.jpg" alt="hbase replication diagram"> </div> <div class="title">Figure 9. Example of a Complex Cluster Replication Configuration</div> </div> <div class="paragraph"> <p>HBase replication borrows many concepts from the <em class="firstterm">statement-based replication</em> design used by MySQL. Instead of SQL statements, entire WALEdits (consisting of multiple cell inserts coming from Put and Delete operations on the clients) are replicated in order to maintain atomicity.</p> </div> </div> <div class="sect2"> <h3 id="_managing_and_configuring_cluster_replication"><a class="anchor" href="#_managing_and_configuring_cluster_replication"></a>130.2. Managing and Configuring Cluster Replication</h3> <div class="olist arabic"> <div class="title">Cluster Configuration Overview</div> <ol class="arabic"> <li> <p>Configure and start the source and destination clusters. Create tables with the same names and column families on both the source and destination clusters, so that the destination cluster knows where to store data it will receive.</p> </li> <li> <p>All hosts in the source and destination clusters should be reachable to each other.</p> </li> <li> <p>If both clusters use the same ZooKeeper cluster, you must use a different <code>zookeeper.znode.parent</code>, because they cannot write in the same folder.</p> </li> <li> <p>Check to be sure that replication has not been disabled. <code>hbase.replication</code> defaults to <code>true</code>.</p> </li> <li> <p>On the source cluster, in HBase Shell, add the destination cluster as a peer, using the <code>add_peer</code> command.</p> </li> <li> <p>On the source cluster, in HBase Shell, enable the table replication, using the <code>enable_table_replication</code> command.</p> </li> <li> <p>Check the logs to see if replication is taking place. If so, you will see messages like the following, coming from the ReplicationSource.</p> </li> </ol> </div> <div class="listingblock"> <div class="content"> <pre>LOG.info("Replicating "+clusterId + " -> " + peerClusterId);</pre> </div> </div> <div class="dlist"> <div class="title">Cluster Management Commands</div> <dl> <dt class="hdlist1">add_peer <ID> <CLUSTER_KEY></dt> <dd> <p>Adds a replication relationship between two clusters.<br></p> <div class="ulist"> <ul> <li> <p>ID — a unique string, which must not contain a hyphen.</p> </li> <li> <p>CLUSTER_KEY: composed using the following template, with appropriate place-holders: <code>hbase.zookeeper.quorum:hbase.zookeeper.property.clientPort:zookeeper.znode.parent</code></p> </li> </ul> </div> </dd> <dt class="hdlist1">list_peers</dt> <dd> <p>list all replication relationships known by this cluster</p> </dd> <dt class="hdlist1">enable_peer <ID></dt> <dd> <p>Enable a previously-disabled replication relationship</p> </dd> <dt class="hdlist1">disable_peer <ID></dt> <dd> <p>Disable a replication relationship. HBase will no longer send edits to that peer cluster, but it still keeps track of all the new WALs that it will need to replicate if and when it is re-enabled.</p> </dd> <dt class="hdlist1">remove_peer <ID></dt> <dd> <p>Disable and remove a replication relationship. HBase will no longer send edits to that peer cluster or keep track of WALs.</p> </dd> <dt class="hdlist1">enable_table_replication <TABLE_NAME></dt> <dd> <p>Enable the table replication switch for all it’s column families. If the table is not found in the destination cluster then it will create one with the same name and column families.</p> </dd> <dt class="hdlist1">disable_table_replication <TABLE_NAME></dt> <dd> <p>Disable the table replication switch for all it’s column families.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_verifying_replicated_data"><a class="anchor" href="#_verifying_replicated_data"></a>130.3. Verifying Replicated Data</h3> <div class="paragraph"> <p>The <code>VerifyReplication</code> MapReduce job, which is included in HBase, performs a systematic comparison of replicated data between two different clusters. Run the VerifyReplication job on the master cluster, supplying it with the peer ID and table name to use for validation. You can limit the verification further by specifying a time range or specific families. The job’s short name is <code>verifyrep</code>. To run the job, use a command like the following:</p> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ HADOOP_CLASSPATH=`${HBASE_HOME}/bin/hbase classpath` "${HADOOP_HOME}/bin/hadoop" jar "${HBASE_HOME}/hbase-server-VERSION.jar" verifyrep --starttime=<timestamp> --endtime=<timestamp> --families=<myFam> <ID> <tableName></code></pre> </div> </div> <div class="paragraph"> <p>+ The <code>VerifyReplication</code> command prints out <code>GOODROWS</code> and <code>BADROWS</code> counters to indicate rows that did and did not replicate correctly.</p> </div> </div> <div class="sect2"> <h3 id="_detailed_information_about_cluster_replication"><a class="anchor" href="#_detailed_information_about_cluster_replication"></a>130.4. Detailed Information About Cluster Replication</h3> <div class="imageblock"> <div class="content"> <img src="images/replication_overview.png" alt="replication overview"> </div> <div class="title">Figure 10. Replication Architecture Overview</div> </div> <div class="sect3"> <h4 id="_life_of_a_wal_edit"><a class="anchor" href="#_life_of_a_wal_edit"></a>130.4.1. Life of a WAL Edit</h4> <div class="paragraph"> <p>A single WAL edit goes through several steps in order to be replicated to a slave cluster.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>An HBase client uses a Put or Delete operation to manipulate data in HBase.</p> </li> <li> <p>The region server writes the request to the WAL in a way allows it to be replayed if it is not written successfully.</p> </li> <li> <p>If the changed cell corresponds to a column family that is scoped for replication, the edit is added to the queue for replication.</p> </li> <li> <p>In a separate thread, the edit is read from the log, as part of a batch process. Only the KeyValues that are eligible for replication are kept. Replicable KeyValues are part of a column family whose schema is scoped GLOBAL, are not part of a catalog such as <code>hbase:meta</code>, did not originate from the target slave cluster, and have not already been consumed by the target slave cluster.</p> </li> <li> <p>The edit is tagged with the master’s UUID and added to a buffer. When the buffer is filled, or the reader reaches the end of the file, the buffer is sent to a random region server on the slave cluster.</p> </li> <li> <p>The region server reads the edits sequentially and separates them into buffers, one buffer per table. After all edits are read, each buffer is flushed using <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/client/Table.html">Table</a>, HBase’s normal client. The master’s UUID and the UUIDs of slaves which have already consumed the data are preserved in the edits they are applied, in order to prevent replication loops.</p> </li> <li> <p>In the master, the offset for the WAL that is currently being replicated is registered in ZooKeeper.</p> </li> <li> <p>The first three steps, where the edit is inserted, are identical.</p> </li> <li> <p>Again in a separate thread, the region server reads, filters, and edits the log edits in the same way as above. The slave region server does not answer the RPC call.</p> </li> <li> <p>The master sleeps and tries again a configurable number of times.</p> </li> <li> <p>If the slave region server is still not available, the master selects a new subset of region server to replicate to, and tries again to send the buffer of edits.</p> </li> <li> <p>Meanwhile, the WALs are rolled and stored in a queue in ZooKeeper. Logs that are <em class="firstterm">archived</em> by their region server, by moving them from the region server’s log directory to a central log directory, will update their paths in the in-memory queue of the replicating thread.</p> </li> <li> <p>When the slave cluster is finally available, the buffer is applied in the same way as during normal processing. The master region server will then replicate the backlog of logs that accumulated during the outage.</p> </li> </ol> </div> <div class="paragraph"> <div class="title">Spreading Queue Failover Load</div> <p>When replication is active, a subset of region servers in the source cluster is responsible for shipping edits to the sink. This responsibility must be failed over like all other region server functions should a process or node crash. The following configuration settings are recommended for maintaining an even distribution of replication activity over the remaining live servers in the source cluster:</p> </div> <div class="ulist"> <ul> <li> <p>Set <code>replication.source.maxretriesmultiplier</code> to <code>300</code>.</p> </li> <li> <p>Set <code>replication.source.sleepforretries</code> to <code>1</code> (1 second). This value, combined with the value of <code>replication.source.maxretriesmultiplier</code>, causes the retry cycle to last about 5 minutes.</p> </li> <li> <p>Set <code>replication.sleep.before.failover</code> to <code>30000</code> (30 seconds) in the source cluster site configuration.</p> </li> </ul> </div> <div class="paragraph"> <div class="title">Preserving Tags During Replication</div> <p>By default, the codec used for replication between clusters strips tags, such as cell-level ACLs, from cells. To prevent the tags from being stripped, you can use a different codec which does not strip them. Configure <code>hbase.replication.rpc.codec</code> to use <code>org.apache.hadoop.hbase.codec.KeyValueCodecWithTags</code>, on both the source and sink RegionServers involved in the replication. This option was introduced in <a href="https://issues.apache.org/jira/browse/HBASE-10322">HBASE-10322</a>.</p> </div> </div> <div class="sect3"> <h4 id="_replication_internals"><a class="anchor" href="#_replication_internals"></a>130.4.2. Replication Internals</h4> <div class="dlist"> <dl> <dt class="hdlist1">Replication State in ZooKeeper</dt> <dd> <p>HBase replication maintains its state in ZooKeeper. By default, the state is contained in the base node <em>/hbase/replication</em>. This node contains two child nodes, the <code>Peers</code> znode and the <code>RS</code> znode.</p> </dd> <dt class="hdlist1">The <code>Peers</code> Znode</dt> <dd> <p>The <code>peers</code> znode is stored in <em>/hbase/replication/peers</em> by default. It consists of a list of all peer replication clusters, along with the status of each of them. The value of each peer is its cluster key, which is provided in the HBase Shell. The cluster key contains a list of ZooKeeper nodes in the cluster’s quorum, the client port for the ZooKeeper quorum, and the base znode for HBase in HDFS on that cluster.</p> </dd> <dt class="hdlist1">The <code>RS</code> Znode</dt> <dd> <p>The <code>rs</code> znode contains a list of WAL logs which need to be replicated. This list is divided into a set of queues organized by region server and the peer cluster the region server is shipping the logs to. The rs znode has one child znode for each region server in the cluster. The child znode name is the region server’s hostname, client port, and start code. This list includes both live and dead region servers.</p> </dd> </dl> </div> </div> <div class="sect3"> <h4 id="_choosing_region_servers_to_replicate_to"><a class="anchor" href="#_choosing_region_servers_to_replicate_to"></a>130.4.3. Choosing Region Servers to Replicate To</h4> <div class="paragraph"> <p>When a master cluster region server initiates a replication source to a slave cluster, it first connects to the slave’s ZooKeeper ensemble using the provided cluster key . It then scans the <em>rs/</em> directory to discover all the available sinks (region servers that are accepting incoming streams of edits to replicate) and randomly chooses a subset of them using a configured ratio which has a default value of 10%. For example, if a slave cluster has 150 machines, 15 will be chosen as potential recipient for edits that this master cluster region server sends. Because this selection is performed by each master region server, the probability that all slave region servers are used is very high, and this method works for clusters of any size. For example, a master cluster of 10 machines replicating to a slave cluster of 5 machines with a ratio of 10% causes the master cluster region servers to choose one machine each at random.</p> </div> <div class="paragraph"> <p>A ZooKeeper watcher is placed on the <em>${zookeeper.znode.parent}/rs</em> node of the slave cluster by each of the master cluster’s region servers. This watch is used to monitor changes in the composition of the slave cluster. When nodes are removed from the slave cluster, or if nodes go down or come back up, the master cluster’s region servers will respond by selecting a new pool of slave region servers to replicate to.</p> </div> </div> <div class="sect3"> <h4 id="_keeping_track_of_logs"><a class="anchor" href="#_keeping_track_of_logs"></a>130.4.4. Keeping Track of Logs</h4> <div class="paragraph"> <p>Each master cluster region server has its own znode in the replication znodes hierarchy. It contains one znode per peer cluster (if 5 slave clusters, 5 znodes are created), and each of these contain a queue of WALs to process. Each of these queues will track the WALs created by that region server, but they can differ in size. For example, if one slave cluster becomes unavailable for some time, the WALs should not be deleted, so they need to stay in the queue while the others are processed. See <a href="#rs.failover.details">rs.failover.details</a> for an example.</p> </div> <div class="paragraph"> <p>When a source is instantiated, it contains the current WAL that the region server is writing to. During log rolling, the new file is added to the queue of each slave cluster’s znode just before it is made available. This ensures that all the sources are aware that a new log exists before the region server is able to append edits into it, but this operations is now more expensive. The queue items are discarded when the replication thread cannot read more entries from a file (because it reached the end of the last block) and there are other files in the queue. This means that if a source is up to date and replicates from the log that the region server writes to, reading up to the "end" of the current file will not delete the item in the queue.</p> </div> <div class="paragraph"> <p>A log can be archived if it is no longer used or if the number of logs exceeds <code>hbase.regionserver.maxlogs</code> because the insertion rate is faster than regions are flushed. When a log is archived, the source threads are notified that the path for that log changed. If a particular source has already finished with an archived log, it will just ignore the message. If the log is in the queue, the path will be updated in memory. If the log is currently being replicated, the change will be done atomically so that the reader doesn’t attempt to open the file when has already been moved. Because moving a file is a NameNode operation , if the reader is currently reading the log, it won’t generate any exception.</p> </div> </div> <div class="sect3"> <h4 id="_reading_filtering_and_sending_edits"><a class="anchor" href="#_reading_filtering_and_sending_edits"></a>130.4.5. Reading, Filtering and Sending Edits</h4> <div class="paragraph"> <p>By default, a source attempts to read from a WAL and ship log entries to a sink as quickly as possible. Speed is limited by the filtering of log entries Only KeyValues that are scoped GLOBAL and that do not belong to catalog tables will be retained. Speed is also limited by total size of the list of edits to replicate per slave, which is limited to 64 MB by default. With this configuration, a master cluster region server with three slaves would use at most 192 MB to store data to replicate. This does not account for the data which was filtered but not garbage collected.</p> </div> <div class="paragraph"> <p>Once the maximum size of edits has been buffered or the reader reaces the end of the WAL, the source thread stops reading and chooses at random a sink to replicate to (from the list that was generated by keeping only a subset of slave region servers). It directly issues a RPC to the chosen region server and waits for the method to return. If the RPC was successful, the source determines whether the current file has been emptied or it contains more data which needs to be read. If the file has been emptied, the source deletes the znode in the queue. Otherwise, it registers the new offset in the log’s znode. If the RPC threw an exception, the source will retry 10 times before trying to find a different sink.</p> </div> </div> <div class="sect3"> <h4 id="_cleaning_logs"><a class="anchor" href="#_cleaning_logs"></a>130.4.6. Cleaning Logs</h4> <div class="paragraph"> <p>If replication is not enabled, the master’s log-cleaning thread deletes old logs using a configured TTL. This TTL-based method does not work well with replication, because archived logs which have exceeded their TTL may still be in a queue. The default behavior is augmented so that if a log is past its TTL, the cleaning thread looks up every queue until it finds the log, while caching queues it has found. If the log is not found in any queues, the log will be deleted. The next time the cleaning process needs to look for a log, it starts by using its cached list.</p> </div> </div> <div class="sect3"> <h4 id="rs.failover.details"><a class="anchor" href="#rs.failover.details"></a>130.4.7. Region Server Failover</h4> <div class="paragraph"> <p>When no region servers are failing, keeping track of the logs in ZooKeeper adds no value. Unfortunately, region servers do fail, and since ZooKeeper is highly available, it is useful for managing the transfer of the queues in the event of a failure.</p> </div> <div class="paragraph"> <p>Each of the master cluster region servers keeps a watcher on every other region server, in order to be notified when one dies (just as the master does). When a failure happens, they all race to create a znode called <code>lock</code> inside the dead region server’s znode that contains its queues. The region server that creates it successfully then transfers all the queues to its own znode, one at a time since ZooKeeper does not support renaming queues. After queues are all transferred, they are deleted from the old location. The znodes that were recovered are renamed with the ID of the slave cluster appended with the name of the dead server.</p> </div> <div class="paragraph"> <p>Next, the master cluster region server creates one new source thread per copied queue, and each of the source threads follows the read/filter/ship pattern. The main difference is that those queues will never receive new data, since they do not belong to their new region server. When the reader hits the end of the last log, the queue’s znode is deleted and the master cluster region server closes that replication source.</p> </div> <div class="paragraph"> <p>Given a master cluster with 3 region servers replicating to a single slave with id <code>2</code>, the following hierarchy represents what the znodes layout could be at some point in time. The region servers' znodes all contain a <code>peers</code> znode which contains a single queue. The znode names in the queues represent the actual file names on HDFS in the form <code>address,port.timestamp</code>.</p> </div> <div class="listingblock"> <div class="content"> <pre>/hbase/replication/rs/ 1.1.1.1,60020,123456780/ 2/ 1.1.1.1,60020.1234 (Contains a position) 1.1.1.1,60020.1265 1.1.1.2,60020,123456790/ 2/ 1.1.1.2,60020.1214 (Contains a position) 1.1.1.2,60020.1248 1.1.1.2,60020.1312 1.1.1.3,60020, 123456630/ 2/ 1.1.1.3,60020.1280 (Contains a position)</pre> </div> </div> <div class="paragraph"> <p>Assume that 1.1.1.2 loses its ZooKeeper session. The survivors will race to create a lock, and, arbitrarily, 1.1.1.3 wins. It will then start transferring all the queues to its local peers znode by appending the name of the dead server. Right before 1.1.1.3 is able to clean up the old znodes, the layout will look like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>/hbase/replication/rs/ 1.1.1.1,60020,123456780/ 2/ 1.1.1.1,60020.1234 (Contains a position) 1.1.1.1,60020.1265 1.1.1.2,60020,123456790/ lock 2/ 1.1.1.2,60020.1214 (Contains a position) 1.1.1.2,60020.1248 1.1.1.2,60020.1312 1.1.1.3,60020,123456630/ 2/ 1.1.1.3,60020.1280 (Contains a position) 2-1.1.1.2,60020,123456790/ 1.1.1.2,60020.1214 (Contains a position) 1.1.1.2,60020.1248 1.1.1.2,60020.1312</pre> </div> </div> <div class="paragraph"> <p>Some time later, but before 1.1.1.3 is able to finish replicating the last WAL from 1.1.1.2, it dies too. Some new logs were also created in the normal queues. The last region server will then try to lock 1.1.1.3’s znode and will begin transferring all the queues. The new layout will be:</p> </div> <div class="listingblock"> <div class="content"> <pre>/hbase/replication/rs/ 1.1.1.1,60020,123456780/ 2/ 1.1.1.1,60020.1378 (Contains a position) 2-1.1.1.3,60020,123456630/ 1.1.1.3,60020.1325 (Contains a position) 1.1.1.3,60020.1401 2-1.1.1.2,60020,123456790-1.1.1.3,60020,123456630/ 1.1.1.2,60020.1312 (Contains a position) 1.1.1.3,60020,123456630/ lock 2/ 1.1.1.3,60020.1325 (Contains a position) 1.1.1.3,60020.1401 2-1.1.1.2,60020,123456790/ 1.1.1.2,60020.1312 (Contains a position)</pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="_replication_metrics"><a class="anchor" href="#_replication_metrics"></a>130.5. Replication Metrics</h3> <div class="paragraph"> <p>The following metrics are exposed at the global region server level and at the peer level:</p> </div> <div class="dlist"> <dl> <dt class="hdlist1"><code>source.sizeOfLogQueue</code></dt> <dd> <p>number of WALs to process (excludes the one which is being processed) at the Replication source</p> </dd> <dt class="hdlist1"><code>source.shippedOps</code></dt> <dd> <p>number of mutations shipped</p> </dd> <dt class="hdlist1"><code>source.logEditsRead</code></dt> <dd> <p>number of mutations read from WALs at the replication source</p> </dd> <dt class="hdlist1"><code>source.ageOfLastShippedOp</code></dt> <dd> <p>age of last batch that was shipped by the replication source</p> </dd> <dt class="hdlist1"><code>source.completedLogs</code></dt> <dd> <p>The number of write-ahead-log files that have completed their acknowledged sending to the peer associated with this source. Increments to this metric are a part of normal operation of HBase replication.</p> </dd> <dt class="hdlist1"><code>source.completedRecoverQueues</code></dt> <dd> <p>The number of recovery queues this source has completed sending to the associated peer. Increments to this metric are a part of normal recovery of HBase replication in the face of failed Region Servers.</p> </dd> <dt class="hdlist1"><code>source.uncleanlyClosedLogs</code></dt> <dd> <p>The number of write-ahead-log files the replication system considered completed after reaching the end of readable entries in the face of an uncleanly closed file.</p> </dd> <dt class="hdlist1"><code>source.ignoredUncleanlyClosedLogContentsInBytes</code></dt> <dd> <p>When a write-ahead-log file is not closed cleanly, there will likely be some entry that has been partially serialized. This metric contains the number of bytes of such entries the HBase replication system believes were remaining at the end of files skipped in the face of an uncleanly closed file. Those bytes should either be in different file or represent a client write that was not acknowledged.</p> </dd> <dt class="hdlist1"><code>source.restartedLogReading</code></dt> <dd> <p>The number of times the HBase replication system detected that it failed to correctly parse a cleanly closed write-ahead-log file. In this circumstance, the system replays the entire log from the beginning, ensuring that no edits fail to be acknowledged by the associated peer. Increments to this metric indicate that the HBase replication system is having difficulty correctly handling failures in the underlying distributed storage system. No dataloss should occur, but you should check Region Server log files for details of the failures.</p> </dd> <dt class="hdlist1"><code>source.repeatedLogFileBytes</code></dt> <dd> <p>When the HBase replication system determines that it needs to replay a given write-ahead-log file, this metric is incremented by the number of bytes the replication system believes had already been acknowledged by the associated peer prior to starting over.</p> </dd> <dt class="hdlist1"><code>source.closedLogsWithUnknownFileLength</code></dt> <dd> <p>Incremented when the HBase replication system believes it is at the end of a write-ahead-log file but it can not determine the length of that file in the underlying distributed storage system. Could indicate dataloss since the replication system is unable to determine if the end of readable entries lines up with the expected end of the file. You should check Region Server log files for details of the failures.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_replication_configuration_options"><a class="anchor" href="#_replication_configuration_options"></a>130.6. Replication Configuration Options</h3> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Option</th> <th class="tableblock halign-left valign-top">Description</th> <th class="tableblock halign-left valign-top">Default</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">zookeeper.znode.parent</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The name of the base ZooKeeper znode used for HBase</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">/hbase</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">zookeeper.znode.replication</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The name of the base znode used for replication</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">replication</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">zookeeper.znode.replication.peers</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The name of the peer znode</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">peers</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">zookeeper.znode.replication.peers.state</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The name of peer-state znode</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">peer-state</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">zookeeper.znode.replication.rs</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The name of the rs znode</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">rs</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hbase.replication</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Whether replication is enabled or disabled on a given cluster</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">false</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">eplication.sleep.before.failover</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">How many milliseconds a worker should sleep before attempting to replicate a dead region server’s WAL queues.</p></td> <td class="tableblock halign-left valign-top"></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">replication.executor.workers</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The number of region servers a given region server should attempt to failover simultaneously.</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">1</p></td> </tr> </tbody> </table> </div> <div class="sect2"> <h3 id="_monitoring_replication_status"><a class="anchor" href="#_monitoring_replication_status"></a>130.7. Monitoring Replication Status</h3> <div class="paragraph"> <p>You can use the HBase Shell command <code>status 'replication'</code> to monitor the replication status on your cluster. The command has three variations: * <code>status 'replication'</code> — prints the status of each source and its sinks, sorted by hostname. * <code>status 'replication', 'source'</code> — prints the status for each replication source, sorted by hostname. * <code>status 'replication', 'sink'</code> — prints the status for each replication sink, sorted by hostname.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="ops.backup"><a class="anchor" href="#ops.backup"></a>131. HBase Backup</h2> <div class="sectionbody"> <div class="paragraph"> <p>There are two broad strategies for performing HBase backups: backing up with a full cluster shutdown, and backing up on a live cluster. Each approach has pros and cons.</p> </div> <div class="paragraph"> <p>For additional information, see <a href="http://blog.sematext.com/2011/03/11/hbase-backup-options/">HBase Backup Options</a> over on the Sematext Blog.</p> </div> <div class="sect2"> <h3 id="ops.backup.fullshutdown"><a class="anchor" href="#ops.backup.fullshutdown"></a>131.1. Full Shutdown Backup</h3> <div class="paragraph"> <p>Some environments can tolerate a periodic full shutdown of their HBase cluster, for example if it is being used a back-end analytic capacity and not serving front-end web-pages. The benefits are that the NameNode/Master are RegionServers are down, so there is no chance of missing any in-flight changes to either StoreFiles or metadata. The obvious con is that the cluster is down. The steps include:</p> </div> <div class="sect3"> <h4 id="ops.backup.fullshutdown.stop"><a class="anchor" href="#ops.backup.fullshutdown.stop"></a>131.1.1. Stop HBase</h4> </div> <div class="sect3"> <h4 id="ops.backup.fullshutdown.distcp"><a class="anchor" href="#ops.backup.fullshutdown.distcp"></a>131.1.2. Distcp</h4> <div class="paragraph"> <p>Distcp could be used to either copy the contents of the HBase directory in HDFS to either the same cluster in another directory, or to a different cluster.</p> </div> <div class="paragraph"> <p>Note: Distcp works in this situation because the cluster is down and there are no in-flight edits to files. Distcp-ing of files in the HBase directory is not generally recommended on a live cluster.</p> </div> </div> <div class="sect3"> <h4 id="ops.backup.fullshutdown.restore"><a class="anchor" href="#ops.backup.fullshutdown.restore"></a>131.1.3. Restore (if needed)</h4> <div class="paragraph"> <p>The backup of the hbase directory from HDFS is copied onto the 'real' hbase directory via distcp. The act of copying these files creates new HDFS metadata, which is why a restore of the NameNode edits from the time of the HBase backup isn’t required for this kind of restore, because it’s a restore (via distcp) of a specific HDFS directory (i.e., the HBase part) not the entire HDFS file-system.</p> </div> </div> </div> <div class="sect2"> <h3 id="ops.backup.live.replication"><a class="anchor" href="#ops.backup.live.replication"></a>131.2. Live Cluster Backup - Replication</h3> <div class="paragraph"> <p>This approach assumes that there is a second cluster. See the HBase page on <a href="http://hbase.apache.org/replication.html">replication</a> for more information.</p> </div> </div> <div class="sect2"> <h3 id="ops.backup.live.copytable"><a class="anchor" href="#ops.backup.live.copytable"></a>131.3. Live Cluster Backup - CopyTable</h3> <div class="paragraph"> <p>The <a href="#copytable">copytable</a> utility could either be used to copy data from one table to another on the same cluster, or to copy data to another table on another cluster.</p> </div> <div class="paragraph"> <p>Since the cluster is up, there is a risk that edits could be missed in the copy process.</p> </div> </div> <div class="sect2"> <h3 id="ops.backup.live.export"><a class="anchor" href="#ops.backup.live.export"></a>131.4. Live Cluster Backup - Export</h3> <div class="paragraph"> <p>The <a href="#export">export</a> approach dumps the content of a table to HDFS on the same cluster. To restore the data, the <a href="#import">import</a> utility would be used.</p> </div> <div class="paragraph"> <p>Since the cluster is up, there is a risk that edits could be missed in the export process.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="ops.snapshots"><a class="anchor" href="#ops.snapshots"></a>132. HBase Snapshots</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase Snapshots allow you to take a snapshot of a table without too much impact on Region Servers. Snapshot, Clone and restore operations don’t involve data copying. Also, Exporting the snapshot to another cluster doesn’t have impact on the Region Servers.</p> </div> <div class="paragraph"> <p>Prior to version 0.94.6, the only way to backup or to clone a table is to use CopyTable/ExportTable, or to copy all the hfiles in HDFS after disabling the table. The disadvantages of these methods are that you can degrade region server performance (Copy/Export Table) or you need to disable the table, that means no reads or writes; and this is usually unacceptable.</p> </div> <div class="sect2"> <h3 id="ops.snapshots.configuration"><a class="anchor" href="#ops.snapshots.configuration"></a>132.1. Configuration</h3> <div class="paragraph"> <p>To turn on the snapshot support just set the <code>hbase.snapshot.enabled</code> property to true. (Snapshots are enabled by default in 0.95+ and off by default in 0.94.6+)</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> <property> <name>hbase.snapshot.enabled</name> <value><span class="predefined-constant">true</span></value> </property></code></pre> </div> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.takeasnapshot"><a class="anchor" href="#ops.snapshots.takeasnapshot"></a>132.2. Take a Snapshot</h3> <div class="paragraph"> <p>You can take a snapshot of a table regardless of whether it is enabled or disabled. The snapshot operation doesn’t involve any data copying.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase shell hbase> snapshot 'myTable', 'myTableSnapshot-122112'</pre> </div> </div> <div class="paragraph"> <div class="title">Take a Snapshot Without Flushing</div> <p>The default behavior is to perform a flush of data in memory before the snapshot is taken. This means that data in memory is included in the snapshot. In most cases, this is the desired behavior. However, if your set-up can tolerate data in memory being excluded from the snapshot, you can use the <code>SKIP_FLUSH</code> option of the <code>snapshot</code> command to disable and flushing while taking the snapshot.</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase> snapshot 'mytable', 'snapshot123', {SKIP_FLUSH => true}</pre> </div> </div> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> There is no way to determine or predict whether a very concurrent insert or update will be included in a given snapshot, whether flushing is enabled or disabled. A snapshot is only a representation of a table during a window of time. The amount of time the snapshot operation will take to reach each Region Server may vary from a few seconds to a minute, depending on the resource load and speed of the hardware or network, among other factors. There is also no way to know whether a given insert or update is in memory or has been flushed. </td> </tr> </table> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.list"><a class="anchor" href="#ops.snapshots.list"></a>132.3. Listing Snapshots</h3> <div class="paragraph"> <p>List all snapshots taken (by printing the names and relative information).</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase shell hbase> list_snapshots</pre> </div> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.delete"><a class="anchor" href="#ops.snapshots.delete"></a>132.4. Deleting Snapshots</h3> <div class="paragraph"> <p>You can remove a snapshot, and the files retained for that snapshot will be removed if no longer needed.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase shell hbase> delete_snapshot 'myTableSnapshot-122112'</pre> </div> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.clone"><a class="anchor" href="#ops.snapshots.clone"></a>132.5. Clone a table from snapshot</h3> <div class="paragraph"> <p>From a snapshot you can create a new table (clone operation) with the same data that you had when the snapshot was taken. The clone operation, doesn’t involve data copies, and a change to the cloned table doesn’t impact the snapshot or the original table.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase shell hbase> clone_snapshot 'myTableSnapshot-122112', 'myNewTestTable'</pre> </div> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.restore"><a class="anchor" href="#ops.snapshots.restore"></a>132.6. Restore a snapshot</h3> <div class="paragraph"> <p>The restore operation requires the table to be disabled, and the table will be restored to the state at the time when the snapshot was taken, changing both data and schema if required.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase shell hbase> disable 'myTable' hbase> restore_snapshot 'myTableSnapshot-122112'</pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Since Replication works at log level and snapshots at file-system level, after a restore, the replicas will be in a different state from the master. If you want to use restore, you need to stop replication and redo the bootstrap. </td> </tr> </table> </div> <div class="paragraph"> <p>In case of partial data-loss due to misbehaving client, instead of a full restore that requires the table to be disabled, you can clone the table from the snapshot and use a Map-Reduce job to copy the data that you need, from the clone to the main one.</p> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.acls"><a class="anchor" href="#ops.snapshots.acls"></a>132.7. Snapshots operations and ACLs</h3> <div class="paragraph"> <p>If you are using security with the AccessController Coprocessor (See <a href="#hbase.accesscontrol.configuration">hbase.accesscontrol.configuration</a>), only a global administrator can take, clone, or restore a snapshot, and these actions do not capture the ACL rights. This means that restoring a table preserves the ACL rights of the existing table, while cloning a table creates a new table that has no ACL rights until the administrator adds them.</p> </div> </div> <div class="sect2"> <h3 id="ops.snapshots.export"><a class="anchor" href="#ops.snapshots.export"></a>132.8. Export to another cluster</h3> <div class="paragraph"> <p>The ExportSnapshot tool copies all the data related to a snapshot (hfiles, logs, snapshot metadata) to another cluster. The tool executes a Map-Reduce job, similar to distcp, to copy files between the two clusters, and since it works at file-system level the hbase cluster does not have to be online.</p> </div> <div class="paragraph"> <p>To copy a snapshot called MySnapshot to an HBase cluster srv2 (hdfs:///srv2:8082/hbase) using 16 mappers:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ bin/hbase org.apache.hadoop.hbase.snapshot.ExportSnapshot -snapshot MySnapshot -copy-to hdfs://srv2:8082/hbase -mappers 16</code></pre> </div> </div> <div class="paragraph"> <div class="title">Limiting Bandwidth Consumption</div> <p>You can limit the bandwidth consumption when exporting a snapshot, by specifying the <code>-bandwidth</code> parameter, which expects an integer representing megabytes per second. The following example limits the above example to 200 MB/sec.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ bin/hbase org.apache.hadoop.hbase.snapshot.ExportSnapshot -snapshot MySnapshot -copy-to hdfs://srv2:8082/hbase -mappers 16 -bandwidth 200</code></pre> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="ops.capacity"><a class="anchor" href="#ops.capacity"></a>133. Capacity Planning and Region Sizing</h2> <div class="sectionbody"> <div class="paragraph"> <p>There are several considerations when planning the capacity for an HBase cluster and performing the initial configuration. Start with a solid understanding of how HBase handles data internally.</p> </div> <div class="sect2"> <h3 id="ops.capacity.nodes"><a class="anchor" href="#ops.capacity.nodes"></a>133.1. Node count and hardware/VM configuration</h3> <div class="sect3"> <h4 id="ops.capacity.nodes.datasize"><a class="anchor" href="#ops.capacity.nodes.datasize"></a>133.1.1. Physical data size</h4> <div class="paragraph"> <p>Physical data size on disk is distinct from logical size of your data and is affected by the following:</p> </div> <div class="ulist"> <ul> <li> <p>Increased by HBase overhead</p> </li> <li> <p>See <a href="#keyvalue">keyvalue</a> and <a href="#keysize">keysize</a>. At least 24 bytes per key-value (cell), can be more. Small keys/values means more relative overhead.</p> </li> <li> <p>KeyValue instances are aggregated into blocks, which are indexed. Indexes also have to be stored. Blocksize is configurable on a per-ColumnFamily basis. See <a href="#regions.arch">regions.arch</a>.</p> </li> <li> <p>Decreased by <a href="#compression">compression</a> and data block encoding, depending on data. See also <a href="http://search-hadoop.com/m/lL12B1PFVhp1">this thread</a>. You might want to test what compression and encoding (if any) make sense for your data.</p> </li> <li> <p>Increased by size of region server <a href="#wal">wal</a> (usually fixed and negligible - less than half of RS memory size, per RS).</p> </li> <li> <p>Increased by HDFS replication - usually x3.</p> </li> </ul> </div> <div class="paragraph"> <p>Aside from the disk space necessary to store the data, one RS may not be able to serve arbitrarily large amounts of data due to some practical limits on region count and size (see <a href="#ops.capacity.regions">ops.capacity.regions</a>).</p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.nodes.throughput"><a class="anchor" href="#ops.capacity.nodes.throughput"></a>133.1.2. Read/Write throughput</h4> <div class="paragraph"> <p>Number of nodes can also be driven by required thoughput for reads and/or writes. The throughput one can get per node depends a lot on data (esp. key/value sizes) and request patterns, as well as node and system configuration. Planning should be done for peak load if it is likely that the load would be the main driver of the increase of the node count. PerformanceEvaluation and <a href="#ycsb">ycsb</a> tools can be used to test single node or a test cluster.</p> </div> <div class="paragraph"> <p>For write, usually 5-15Mb/s per RS can be expected, since every region server has only one active WAL. There’s no good estimate for reads, as it depends vastly on data, requests, and cache hit rate. <a href="#perf.casestudy">perf.casestudy</a> might be helpful.</p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.nodes.gc"><a class="anchor" href="#ops.capacity.nodes.gc"></a>133.1.3. JVM GC limitations</h4> <div class="paragraph"> <p>RS cannot currently utilize very large heap due to cost of GC. There’s also no good way of running multiple RS-es per server (other than running several VMs per machine). Thus, ~20-24Gb or less memory dedicated to one RS is recommended. GC tuning is required for large heap sizes. See <a href="#gcpause">gcpause</a>, <a href="#trouble.log.gc">trouble.log.gc</a> and elsewhere (TODO: where?)</p> </div> </div> </div> <div class="sect2"> <h3 id="ops.capacity.regions"><a class="anchor" href="#ops.capacity.regions"></a>133.2. Determining region count and size</h3> <div class="paragraph"> <p>Generally less regions makes for a smoother running cluster (you can always manually split the big regions later (if necessary) to spread the data, or request load, over the cluster); 20-200 regions per RS is a reasonable range. The number of regions cannot be configured directly (unless you go for fully <a href="#disable.splitting">disable.splitting</a>); adjust the region size to achieve the target region size given table size.</p> </div> <div class="paragraph"> <p>When configuring regions for multiple tables, note that most region settings can be set on a per-table basis via <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/HTableDescriptor.html">HTableDescriptor</a>, as well as shell commands. These settings will override the ones in <code>hbase-site.xml</code>. That is useful if your tables have different workloads/use cases.</p> </div> <div class="paragraph"> <p>Also note that in the discussion of region sizes here, <em>HDFS replication factor is not (and should not be) taken into account, whereas other factors <a href="#ops.capacity.nodes.datasize">ops.capacity.nodes.datasize</a> should be.</em> So, if your data is compressed and replicated 3 ways by HDFS, "9 Gb region" means 9 Gb of compressed data. HDFS replication factor only affects your disk usage and is invisible to most HBase code.</p> </div> <div class="sect3"> <h4 id="_viewing_the_current_number_of_regions"><a class="anchor" href="#_viewing_the_current_number_of_regions"></a>133.2.1. Viewing the Current Number of Regions</h4> <div class="paragraph"> <p>You can view the current number of regions for a given table using the HMaster UI. In the <span class="label">Tables</span> section, the number of online regions for each table is listed in the <span class="label">Online Regions</span> column. This total only includes the in-memory state and does not include disabled or offline regions. If you do not want to use the HMaster UI, you can determine the number of regions by counting the number of subdirectories of the /hbase/<table>/ subdirectories in HDFS, or by running the <code>bin/hbase hbck</code> command. Each of these methods may return a slightly different number, depending on the status of each region.</p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.regions.count"><a class="anchor" href="#ops.capacity.regions.count"></a>133.2.2. Number of regions per RS - upper bound</h4> <div class="paragraph"> <p>In production scenarios, where you have a lot of data, you are normally concerned with the maximum number of regions you can have per server. <a href="#too_many_regions">too many regions</a> has technical discussion on the subject. Basically, the maximum number of regions is mostly determined by memstore memory usage. Each region has its own memstores; these grow up to a configurable size; usually in 128-256 MB range, see <a href="#hbase.hregion.memstore.flush.size">hbase.hregion.memstore.flush.size</a>. One memstore exists per column family (so there’s only one per region if there’s one CF in the table). The RS dedicates some fraction of total memory to its memstores (see <a href="#hbase.regionserver.global.memstore.size">hbase.regionserver.global.memstore.size</a>). If this memory is exceeded (too much memstore usage), it can cause undesirable consequences such as unresponsive server or compaction storms. A good starting point for the number of regions per RS (assuming one table) is:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">((RS memory) * (total memstore fraction)) / ((memstore size)*(<span class="error">#</span> column families))</code></pre> </div> </div> <div class="paragraph"> <p>This formula is pseudo-code. Here are two formulas using the actual tunable parameters, first for HBase 0.98+ and second for HBase 0.94.x.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">HBase 0.98.x</dt> </dl> </div> <div class="listingblock"> <div class="content"> <pre>((RS Xmx) * hbase.regionserver.global.memstore.size) / (hbase.hregion.memstore.flush.size * (# column families))</pre> </div> </div> <div class="dlist"> <dl> <dt class="hdlist1">HBase 0.94.x</dt> </dl> </div> <div class="listingblock"> <div class="content"> <pre>((RS Xmx) * hbase.regionserver.global.memstore.upperLimit) / (hbase.hregion.memstore.flush.size * (# column families))+</pre> </div> </div> <div class="paragraph"> <p>If a given RegionServer has 16 GB of RAM, with default settings, the formula works out to 16384*0.4/128 ~ 51 regions per RS is a starting point. The formula can be extended to multiple tables; if they all have the same configuration, just use the total number of families.</p> </div> <div class="paragraph"> <p>This number can be adjusted; the formula above assumes all your regions are filled at approximately the same rate. If only a fraction of your regions are going to be actively written to, you can divide the result by that fraction to get a larger region count. Then, even if all regions are written to, all region memstores are not filled evenly, and eventually jitter appears even if they are (due to limited number of concurrent flushes). Thus, one can have as many as 2-3 times more regions than the starting point; however, increased numbers carry increased risk.</p> </div> <div class="paragraph"> <p>For write-heavy workload, memstore fraction can be increased in configuration at the expense of block cache; this will also allow one to have more regions.</p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.regions.mincount"><a class="anchor" href="#ops.capacity.regions.mincount"></a>133.2.3. Number of regions per RS - lower bound</h4> <div class="paragraph"> <p>HBase scales by having regions across many servers. Thus if you have 2 regions for 16GB data, on a 20 node machine your data will be concentrated on just a few machines - nearly the entire cluster will be idle. This really can’t be stressed enough, since a common problem is loading 200MB data into HBase and then wondering why your awesome 10 node cluster isn’t doing anything.</p> </div> <div class="paragraph"> <p>On the other hand, if you have a very large amount of data, you may also want to go for a larger number of regions to avoid having regions that are too large.</p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.regions.size"><a class="anchor" href="#ops.capacity.regions.size"></a>133.2.4. Maximum region size</h4> <div class="paragraph"> <p>For large tables in production scenarios, maximum region size is mostly limited by compactions - very large compactions, esp. major, can degrade cluster performance. Currently, the recommended maximum region size is 10-20Gb, and 5-10Gb is optimal. For older 0.90.x codebase, the upper-bound of regionsize is about 4Gb, with a default of 256Mb.</p> </div> <div class="paragraph"> <p>The size at which the region is split into two is generally configured via <a href="#hbase.hregion.max.filesize">hbase.hregion.max.filesize</a>; for details, see <a href="#arch.region.splits">arch.region.splits</a>.</p> </div> <div class="paragraph"> <p>If you cannot estimate the size of your tables well, when starting off, it’s probably best to stick to the default region size, perhaps going smaller for hot tables (or manually split hot regions to spread the load over the cluster), or go with larger region sizes if your cell sizes tend to be largish (100k and up).</p> </div> <div class="paragraph"> <p>In HBase 0.98, experimental stripe compactions feature was added that would allow for larger regions, especially for log data. See <a href="#ops.stripe">ops.stripe</a>.</p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.regions.total"><a class="anchor" href="#ops.capacity.regions.total"></a>133.2.5. Total data size per region server</h4> <div class="paragraph"> <p>According to above numbers for region size and number of regions per region server, in an optimistic estimate 10 GB x 100 regions per RS will give up to 1TB served per region server, which is in line with some of the reported multi-PB use cases. However, it is important to think about the data vs cache size ratio at the RS level. With 1TB of data per server and 10 GB block cache, only 1% of the data will be cached, which may barely cover all block indices.</p> </div> </div> </div> <div class="sect2"> <h3 id="ops.capacity.config"><a class="anchor" href="#ops.capacity.config"></a>133.3. Initial configuration and tuning</h3> <div class="paragraph"> <p>First, see <a href="#important_configurations">important configurations</a>. Note that some configurations, more than others, depend on specific scenarios. Pay special attention to:</p> </div> <div class="ulist"> <ul> <li> <p><a href="#hbase.regionserver.handler.count">hbase.regionserver.handler.count</a> - request handler thread count, vital for high-throughput workloads.</p> </li> <li> <p><a href="#config.wals">config.wals</a> - the blocking number of WAL files depends on your memstore configuration and should be set accordingly to prevent potential blocking when doing high volume of writes.</p> </li> </ul> </div> <div class="paragraph"> <p>Then, there are some considerations when setting up your cluster and tables.</p> </div> <div class="sect3"> <h4 id="ops.capacity.config.compactions"><a class="anchor" href="#ops.capacity.config.compactions"></a>133.3.1. Compactions</h4> <div class="paragraph"> <p>Depending on read/write volume and latency requirements, optimal compaction settings may be different. See <a href="#compaction">compaction</a> for some details.</p> </div> <div class="paragraph"> <p>When provisioning for large data sizes, however, it’s good to keep in mind that compactions can affect write throughput. Thus, for write-intensive workloads, you may opt for less frequent compactions and more store files per regions. Minimum number of files for compactions (<code>hbase.hstore.compaction.min</code>) can be set to higher value; <a href="#hbase.hstore.blockingstorefiles">hbase.hstore.blockingStoreFiles</a> should also be increased, as more files might accumulate in such case. You may also consider manually managing compactions: <a href="#managed.compactions">managed.compactions</a></p> </div> </div> <div class="sect3"> <h4 id="ops.capacity.config.presplit"><a class="anchor" href="#ops.capacity.config.presplit"></a>133.3.2. Pre-splitting the table</h4> <div class="paragraph"> <p>Based on the target number of the regions per RS (see <a href="#ops.capacity.regions.count">ops.capacity.regions.count</a>) and number of RSes, one can pre-split the table at creation time. This would both avoid some costly splitting as the table starts to fill up, and ensure that the table starts out already distributed across many servers.</p> </div> <div class="paragraph"> <p>If the table is expected to grow large enough to justify that, at least one region per RS should be created. It is not recommended to split immediately into the full target number of regions (e.g. 50 * number of RSes), but a low intermediate value can be chosen. For multiple tables, it is recommended to be conservative with presplitting (e.g. pre-split 1 region per RS at most), especially if you don’t know how much each table will grow. If you split too much, you may end up with too many regions, with some tables having too many small regions.</p> </div> <div class="paragraph"> <p>For pre-splitting howto, see <a href="#manual_region_splitting_decisions">manual region splitting decisions</a> and <a href="#precreate.regions">precreate.regions</a>.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="table.rename"><a class="anchor" href="#table.rename"></a>134. Table Rename</h2> <div class="sectionbody"> <div class="paragraph"> <p>In versions 0.90.x of hbase and earlier, we had a simple script that would rename the hdfs table directory and then do an edit of the hbase:meta table replacing all mentions of the old table name with the new. The script was called <code>./bin/rename_table.rb</code>. The script was deprecated and removed mostly because it was unmaintained and the operation performed by the script was brutal.</p> </div> <div class="paragraph"> <p>As of hbase 0.94.x, you can use the snapshot facility renaming a table. Here is how you would do it using the hbase shell:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase shell> disable 'tableName' hbase shell> snapshot 'tableName', 'tableSnapshot' hbase shell> clone_snapshot 'tableSnapshot', 'newTableName' hbase shell> delete_snapshot 'tableSnapshot' hbase shell> drop 'tableName'</pre> </div> </div> <div class="paragraph"> <p>or in code it would be as follows:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="type">void</span> rename(Admin admin, <span class="predefined-type">String</span> oldTableName, <span class="predefined-type">String</span> newTableName) { <span class="predefined-type">String</span> snapshotName = randomName(); admin.disableTable(oldTableName); admin.snapshot(snapshotName, oldTableName); admin.cloneSnapshot(snapshotName, newTableName); admin.deleteSnapshot(snapshotName); admin.deleteTable(oldTableName); }</code></pre> </div> </div> </div> </div> <h1 id="developer" class="sect0"><a class="anchor" href="#developer"></a>Building and Developing Apache HBase</h1> <div class="openblock partintro"> <div class="content"> This chapter contains information and guidelines for building and releasing HBase code and documentation. Being familiar with these guidelines will help the HBase committers to use your contributions more easily. </div> </div> <div class="sect1"> <h2 id="getting.involved"><a class="anchor" href="#getting.involved"></a>135. Getting Involved</h2> <div class="sectionbody"> <div class="paragraph"> <p>Apache HBase gets better only when people contribute! If you are looking to contribute to Apache HBase, look for <a href="https://issues.apache.org/jira/issues/?jql=project%20%3D%20HBASE%20AND%20labels%20in%20(beginner)%20AND%20status%20in%20(Open%2C%20%22In%20Progress%22%2C%20Reopened)">issues in JIRA tagged with the label 'beginner'</a>. These are issues HBase contributors have deemed worthy but not of immediate priority and a good way to ramp on HBase internals. See <a href="http://search-hadoop.com/m/DHED43re96">What label is used for issues that are good on ramps for new contributors?</a> from the dev mailing list for background.</p> </div> <div class="paragraph"> <p>Before you get started submitting code to HBase, please refer to <a href="#developing">developing</a>.</p> </div> <div class="paragraph"> <p>As Apache HBase is an Apache Software Foundation project, see <a href="#asf">asf</a> for more information about how the ASF functions.</p> </div> <div class="sect2"> <h3 id="mailing.list"><a class="anchor" href="#mailing.list"></a>135.1. Mailing Lists</h3> <div class="paragraph"> <p>Sign up for the dev-list and the user-list. See the <a href="https://hbase.apache.org/mail-lists.html">mailing lists</a> page. Posing questions - and helping to answer other people’s questions - is encouraged! There are varying levels of experience on both lists so patience and politeness are encouraged (and please stay on topic.)</p> </div> </div> <div class="sect2"> <h3 id="slack"><a class="anchor" href="#slack"></a>135.2. Slack</h3> <div class="paragraph"> <p>The Apache HBase project has its own link: <a href="http://apache-hbase.slack.com">Slack Channel</a> for real-time questions and discussion. Mail <a href="mailto:dev@hbase.apache.org">dev@hbase.apache.org</a> to request an invite.</p> </div> </div> <div class="sect2"> <h3 id="irc"><a class="anchor" href="#irc"></a>135.3. Internet Relay Chat (IRC)</h3> <div class="paragraph"> <p>(NOTE: Our IRC channel seems to have been deprecated in favor of the above Slack channel)</p> </div> <div class="paragraph"> <p>For real-time questions and discussions, use the <code>#hbase</code> IRC channel on the <a href="https://freenode.net/">FreeNode</a> IRC network. FreeNode offers a web-based client, but most people prefer a native client, and several clients are available for each operating system.</p> </div> </div> <div class="sect2"> <h3 id="_jira"><a class="anchor" href="#_jira"></a>135.4. Jira</h3> <div class="paragraph"> <p>Check for existing issues in <a href="https://issues.apache.org/jira/projects/HBASE/issues">Jira</a>. If it’s either a new feature request, enhancement, or a bug, file a ticket.</p> </div> <div class="paragraph"> <p>We track multiple types of work in JIRA:</p> </div> <div class="ulist"> <ul> <li> <p>Bug: Something is broken in HBase itself.</p> </li> <li> <p>Test: A test is needed, or a test is broken.</p> </li> <li> <p>New feature: You have an idea for new functionality. It’s often best to bring these up on the mailing lists first, and then write up a design specification that you add to the feature request JIRA.</p> </li> <li> <p>Improvement: A feature exists, but could be tweaked or augmented. It’s often best to bring these up on the mailing lists first and have a discussion, then summarize or link to the discussion if others seem interested in the improvement.</p> </li> <li> <p>Wish: This is like a new feature, but for something you may not have the background to flesh out yourself.</p> </li> </ul> </div> <div class="paragraph"> <p>Bugs and tests have the highest priority and should be actionable.</p> </div> <div class="sect3"> <h4 id="_guidelines_for_reporting_effective_issues"><a class="anchor" href="#_guidelines_for_reporting_effective_issues"></a>135.4.1. Guidelines for reporting effective issues</h4> <div class="ulist"> <ul> <li> <p><strong>Search for duplicates</strong>: Your issue may have already been reported. Have a look, realizing that someone else might have worded the summary differently.</p> <div class="paragraph"> <p>Also search the mailing lists, which may have information about your problem and how to work around it. Don’t file an issue for something that has already been discussed and resolved on a mailing list, unless you strongly disagree with the resolution <strong>and</strong> are willing to help take the issue forward.</p> </div> <div class="ulist"> <ul> <li> <p><strong>Discuss in public</strong>: Use the mailing lists to discuss what you’ve discovered and see if there is something you’ve missed. Avoid using back channels, so that you benefit from the experience and expertise of the project as a whole.</p> </li> <li> <p><strong>Don’t file on behalf of others</strong>: You might not have all the context, and you don’t have as much motivation to see it through as the person who is actually experiencing the bug. It’s more helpful in the long term to encourage others to file their own issues. Point them to this material and offer to help out the first time or two.</p> </li> <li> <p><strong>Write a good summary</strong>: A good summary includes information about the problem, the impact on the user or developer, and the area of the code.</p> <div class="ulist"> <ul> <li> <p>Good: <code>Address new license dependencies from hadoop3-alpha4</code></p> </li> <li> <p>Room for improvement: <code>Canary is broken</code></p> <div class="paragraph"> <p>If you write a bad title, someone else will rewrite it for you. This is time they could have spent working on the issue instead.</p> </div> </li> </ul> </div> </li> <li> <p><strong>Give context in the description</strong>: It can be good to think of this in multiple parts:</p> <div class="ulist"> <ul> <li> <p>What happens or doesn’t happen?</p> </li> <li> <p>How does it impact you?</p> </li> <li> <p>How can someone else reproduce it?</p> </li> <li> <p>What would "fixed" look like?</p> <div class="paragraph"> <p>You don’t need to know the answers for all of these, but give as much information as you can. If you can provide technical information, such as a Git commit SHA that you think might have caused the issue or a build failure on builds.apache.org where you think the issue first showed up, share that info.</p> </div> </li> </ul> </div> </li> <li> <p><strong>Fill in all relevant fields</strong>: These fields help us filter, categorize, and find things.</p> </li> <li> <p><strong>One bug, one issue, one patch</strong>: To help with back-porting, don’t split issues or fixes among multiple bugs.</p> </li> <li> <p><strong>Add value if you can</strong>: Filing issues is great, even if you don’t know how to fix them. But providing as much information as possible, being willing to triage and answer questions, and being willing to test potential fixes is even better! We want to fix your issue as quickly as you want it to be fixed.</p> </li> <li> <p><strong>Don’t be upset if we don’t fix it</strong>: Time and resources are finite. In some cases, we may not be able to (or might choose not to) fix an issue, especially if it is an edge case or there is a workaround. Even if it doesn’t get fixed, the JIRA is a public record of it, and will help others out if they run into a similar issue in the future.</p> </li> </ul> </div> </li> </ul> </div> </div> <div class="sect3"> <h4 id="_working_on_an_issue"><a class="anchor" href="#_working_on_an_issue"></a>135.4.2. Working on an issue</h4> <div class="paragraph"> <p>To check for existing issues which you can tackle as a beginner, search for <a href="https://issues.apache.org/jira/issues/?jql=project%20%3D%20HBASE%20AND%20labels%20in%20(beginner)%20AND%20status%20in%20(Open%2C%20%22In%20Progress%22%2C%20Reopened)">issues in JIRA tagged with the label 'beginner'</a>.</p> </div> <div class="ulist"> <div class="title">JIRA Priorites</div> <ul> <li> <p><strong>Blocker</strong>: Should only be used if the issue WILL cause data loss or cluster instability reliably.</p> </li> <li> <p><strong>Critical</strong>: The issue described can cause data loss or cluster instability in some cases.</p> </li> <li> <p><strong>Major</strong>: Important but not tragic issues, like updates to the client API that will add a lot of much-needed functionality or significant bugs that need to be fixed but that don’t cause data loss.</p> </li> <li> <p><strong>Minor</strong>: Useful enhancements and annoying but not damaging bugs.</p> </li> <li> <p><strong>Trivial</strong>: Useful enhancements but generally cosmetic.</p> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 49. Code Blocks in Jira Comments</div> <div class="content"> <div class="paragraph"> <p>A commonly used macro in Jira is {code}. Everything inside the tags is preformatted, as in this example.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">{code} code snippet {code}</code></pre> </div> </div> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="repos"><a class="anchor" href="#repos"></a>136. Apache HBase Repositories</h2> <div class="sectionbody"> <div class="paragraph"> <p>There are two different repositories for Apache HBase: Subversion (SVN) and Git. GIT is our repository of record for all but the Apache HBase website. We used to be on SVN. We migrated. See <a href="https://issues.apache.org/jira/browse/INFRA-7768">Migrate Apache HBase SVN Repos to Git</a>. See <a href="https://hbase.apache.org/source-repository.html">Source Code Management</a> page for contributor and committer links or search for HBase on the <a href="https://gitbox.apache.org/repos/asf/">Apache Git</a> page.</p> </div> </div> </div> <div class="sect1"> <h2 id="_ides"><a class="anchor" href="#_ides"></a>137. IDEs</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="eclipse"><a class="anchor" href="#eclipse"></a>137.1. Eclipse</h3> <div class="sect3"> <h4 id="eclipse.code.formatting"><a class="anchor" href="#eclipse.code.formatting"></a>137.1.1. Code Formatting</h4> <div class="paragraph"> <p>Under the <em>dev-support/</em> folder, you will find <em>hbase_eclipse_formatter.xml</em>. We encourage you to have this formatter in place in eclipse when editing HBase code.</p> </div> <div class="paragraph"> <p>Go to <code>Preferences→Java→Code Style→Formatter→Import</code> to load the xml file. Go to <code>Preferences→Java→Editor→Save Actions</code>, and make sure 'Format source code' and 'Format edited lines' is selected.</p> </div> <div class="paragraph"> <p>In addition to the automatic formatting, make sure you follow the style guidelines explained in <a href="#common.patch.feedback">common.patch.feedback</a>.</p> </div> </div> <div class="sect3"> <h4 id="eclipse.git.plugin"><a class="anchor" href="#eclipse.git.plugin"></a>137.1.2. Eclipse Git Plugin</h4> <div class="paragraph"> <p>If you cloned the project via git, download and install the Git plugin (EGit). Attach to your local git repo (via the <span class="label">Git Repositories</span> window) and you’ll be able to see file revision history, generate patches, etc.</p> </div> </div> <div class="sect3"> <h4 id="eclipse.maven.setup"><a class="anchor" href="#eclipse.maven.setup"></a>137.1.3. HBase Project Setup in Eclipse using <code>m2eclipse</code></h4> <div class="paragraph"> <p>The easiest way is to use the m2eclipse plugin for Eclipse. Eclipse Indigo or newer includes m2eclipse, or you can download it from <a href="http://www.eclipse.org/m2e/" class="bare">http://www.eclipse.org/m2e/</a>. It provides Maven integration for Eclipse, and even lets you use the direct Maven commands from within Eclipse to compile and test your project.</p> </div> <div class="paragraph"> <p>To import the project, click and select the HBase root directory. <code>m2eclipse</code> locates all the hbase modules for you.</p> </div> <div class="paragraph"> <p>If you install m2eclipse and import HBase in your workspace, do the following to fix your eclipse Build Path.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Remove <em>target</em> folder</p> </li> <li> <p>Add <em>target/generated-jamon</em> and <em>target/generated-sources/java</em> folders.</p> </li> <li> <p>Remove from your Build Path the exclusions on the <em>src/main/resources</em> and <em>src/test/resources</em> to avoid error message in the console, such as the following:</p> <div class="listingblock"> <div class="content"> <pre>Failed to execute goal org.apache.maven.plugins:maven-antrun-plugin:1.6:run (default) on project hbase: 'An Ant BuildException has occurred: Replace: source file .../target/classes/hbase-default.xml doesn't exist</pre> </div> </div> <div class="paragraph"> <p>This will also reduce the eclipse build cycles and make your life easier when developing.</p> </div> </li> </ol> </div> </div> <div class="sect3"> <h4 id="eclipse.commandline"><a class="anchor" href="#eclipse.commandline"></a>137.1.4. HBase Project Setup in Eclipse Using the Command Line</h4> <div class="paragraph"> <p>Instead of using <code>m2eclipse</code>, you can generate the Eclipse files from the command line.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>First, run the following command, which builds HBase. You only need to do this once.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean install -DskipTests</code></pre> </div> </div> </li> <li> <p>Close Eclipse, and execute the following command from the terminal, in your local HBase project directory, to generate new <em>.project</em> and <em>.classpath</em> files.</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn eclipse:eclipse</code></pre> </div> </div> </li> <li> <p>Reopen Eclipse and import the <em>.project</em> file in the HBase directory to a workspace.</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="eclipse.maven.class"><a class="anchor" href="#eclipse.maven.class"></a>137.1.5. Maven Classpath Variable</h4> <div class="paragraph"> <p>The <code>$M2_REPO</code> classpath variable needs to be set up for the project. This needs to be set to your local Maven repository, which is usually <em>~/.m2/repository</em></p> </div> <div class="paragraph"> <p>If this classpath variable is not configured, you will see compile errors in Eclipse like this:</p> </div> <div class="listingblock"> <div class="content"> <pre>Description Resource Path Location Type The project cannot be built until build path errors are resolved hbase Unknown Java Problem Unbound classpath variable: 'M2_REPO/asm/asm/3.1/asm-3.1.jar' in project 'hbase' hbase Build path Build Path Problem Unbound classpath variable: 'M2_REPO/com/google/guava/guava/r09/guava-r09.jar' in project 'hbase' hbase Build path Build Path Problem Unbound classpath variable: 'M2_REPO/com/google/protobuf/protobuf-java/2.3.0/protobuf-java-2.3.0.jar' in project 'hbase' hbase Build path Build Path Problem Unbound classpath variable:</pre> </div> </div> </div> <div class="sect3"> <h4 id="eclipse.issues"><a class="anchor" href="#eclipse.issues"></a>137.1.6. Eclipse Known Issues</h4> <div class="paragraph"> <p>Eclipse will currently complain about <em>Bytes.java</em>. It is not possible to turn these errors off.</p> </div> <div class="listingblock"> <div class="content"> <pre>Description Resource Path Location Type Access restriction: The method arrayBaseOffset(Class) from the type Unsafe is not accessible due to restriction on required library /System/Library/Java/JavaVirtualMachines/1.6.0.jdk/Contents/Classes/classes.jar Bytes.java /hbase/src/main/java/org/apache/hadoop/hbase/util line 1061 Java Problem Access restriction: The method arrayIndexScale(Class) from the type Unsafe is not accessible due to restriction on required library /System/Library/Java/JavaVirtualMachines/1.6.0.jdk/Contents/Classes/classes.jar Bytes.java /hbase/src/main/java/org/apache/hadoop/hbase/util line 1064 Java Problem Access restriction: The method getLong(Object, long) from the type Unsafe is not accessible due to restriction on required library /System/Library/Java/JavaVirtualMachines/1.6.0.jdk/Contents/Classes/classes.jar Bytes.java /hbase/src/main/java/org/apache/hadoop/hbase/util line 1111 Java Problem</pre> </div> </div> </div> <div class="sect3"> <h4 id="eclipse.more"><a class="anchor" href="#eclipse.more"></a>137.1.7. Eclipse - More Information</h4> <div class="paragraph"> <p>For additional information on setting up Eclipse for HBase development on Windows, see <a href="http://michaelmorello.blogspot.com/2011/09/hbase-subversion-eclipse-windows.html">Michael Morello’s blog</a> on the topic.</p> </div> </div> </div> <div class="sect2"> <h3 id="_intellij_idea"><a class="anchor" href="#_intellij_idea"></a>137.2. IntelliJ IDEA</h3> <div class="paragraph"> <p>You can set up IntelliJ IDEA for similar functionality as Eclipse. Follow these steps.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Select</p> </li> <li> <p>You do not need to select a profile. Be sure <span class="label">Maven project required</span> is selected, and click <b class="button">Next</b>.</p> </li> <li> <p>Select the location for the JDK.</p> </li> </ol> </div> <div class="paragraph"> <div class="title">Using the HBase Formatter in IntelliJ IDEA</div> <p>Using the Eclipse Code Formatter plugin for IntelliJ IDEA, you can import the HBase code formatter described in <a href="#eclipse.code.formatting">eclipse.code.formatting</a>.</p> </div> </div> <div class="sect2"> <h3 id="_other_ides"><a class="anchor" href="#_other_ides"></a>137.3. Other IDEs</h3> <div class="paragraph"> <p>It would be useful to mirror the <a href="#eclipse">eclipse</a> set-up instructions for other IDEs. If you would like to assist, please have a look at <a href="https://issues.apache.org/jira/browse/HBASE-11704">HBASE-11704</a>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="build"><a class="anchor" href="#build"></a>138. Building Apache HBase</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="build.basic"><a class="anchor" href="#build.basic"></a>138.1. Basic Compile</h3> <div class="paragraph"> <p>HBase is compiled using Maven. You must use at least Maven 3.0.4. To check your Maven version, run the command mvn -version.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">JDK Version Requirements</div> <div class="paragraph"> <p>Starting with HBase 1.0 you must use Java 7 or later to build from source code. See <a href="#java">java</a> for more complete information about supported JDK versions.</p> </div> </td> </tr> </table> </div> <div class="sect3"> <h4 id="maven.build.commands"><a class="anchor" href="#maven.build.commands"></a>138.1.1. Maven Build Commands</h4> <div class="paragraph"> <p>All commands are executed from the local HBase project directory.</p> </div> <div class="sect4"> <h5 id="_package"><a class="anchor" href="#_package"></a>Package</h5> <div class="paragraph"> <p>The simplest command to compile HBase from its java source code is to use the <code>package</code> target, which builds JARs with the compiled files.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn package -DskipTests</code></pre> </div> </div> <div class="paragraph"> <p>Or, to clean up before compiling:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean package -DskipTests</code></pre> </div> </div> <div class="paragraph"> <p>With Eclipse set up as explained above in <a href="#eclipse">eclipse</a>, you can also use the <span class="menu">Build</span> command in Eclipse. To create the full installable HBase package takes a little bit more work, so read on.</p> </div> </div> <div class="sect4"> <h5 id="maven.build.commands.compile"><a class="anchor" href="#maven.build.commands.compile"></a>Compile</h5> <div class="paragraph"> <p>The <code>compile</code> target does not create the JARs with the compiled files.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn compile</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean compile</code></pre> </div> </div> </div> <div class="sect4"> <h5 id="_install"><a class="anchor" href="#_install"></a>Install</h5> <div class="paragraph"> <p>To install the JARs in your <em>~/.m2/</em> directory, use the <code>install</code> target.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn install</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean install</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean install -DskipTests</code></pre> </div> </div> </div> </div> <div class="sect3"> <h4 id="maven.build.commands.unitall"><a class="anchor" href="#maven.build.commands.unitall"></a>138.1.2. Running all or individual Unit Tests</h4> <div class="paragraph"> <p>See the <a href="#hbase.unittests.cmds">hbase.unittests.cmds</a> section in <a href="#hbase.unittests">hbase.unittests</a></p> </div> </div> <div class="sect3"> <h4 id="maven.build.hadoop"><a class="anchor" href="#maven.build.hadoop"></a>138.1.3. Building against various hadoop versions.</h4> <div class="paragraph"> <p>HBase supports building against Apache Hadoop versions: 2.y and 3.y (early release artifacts). By default we build against Hadoop 2.x.</p> </div> <div class="paragraph"> <p>To build against a specific release from the Hadoop 2.y line, set e.g. <code>-Dhadoop-two.version=2.7.4</code>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn -Dhadoop-two.version=2.7.4 ...</code></pre> </div> </div> <div class="paragraph"> <p>To change the major release line of Hadoop we build against, add a hadoop.profile property when you invoke mvn:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn -Dhadoop.profile=3.0 ...</code></pre> </div> </div> <div class="paragraph"> <p>The above will build against whatever explicit hadoop 3.y version we have in our <em>pom.xml</em> as our '3.0' version. Tests may not all pass so you may need to pass <code>-DskipTests</code> unless you are inclined to fix the failing tests.</p> </div> <div class="paragraph"> <p>To pick a particular Hadoop 3.y release, you’d set e.g. <code>-Dhadoop-three.version=3.0.0-alpha1</code>.</p> </div> </div> <div class="sect3"> <h4 id="build.protobuf"><a class="anchor" href="#build.protobuf"></a>138.1.4. Build Protobuf</h4> <div class="paragraph"> <p>You may need to change the protobuf definitions that reside in the <em>hbase-protocol</em> module or other modules.</p> </div> <div class="paragraph"> <p>The protobuf files are located in <em>hbase-protocol/src/main/protobuf</em>. For the change to be effective, you will need to regenerate the classes. You can use maven profile <code>compile-protobuf</code> to do this.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn compile -Pcompile-protobuf</code></pre> </div> </div> <div class="paragraph"> <p>You may also want to define <code>protoc.path</code> for the protoc binary, using the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn compile -Pcompile-protobuf -Dprotoc.path=/opt/local/bin/protoc</code></pre> </div> </div> <div class="paragraph"> <p>Read the <em>hbase-protocol/README.txt</em> for more details.</p> </div> </div> <div class="sect3"> <h4 id="build.thrift"><a class="anchor" href="#build.thrift"></a>138.1.5. Build Thrift</h4> <div class="paragraph"> <p>You may need to change the thrift definitions that reside in the <em>hbase-thrift</em> module or other modules.</p> </div> <div class="paragraph"> <p>The thrift files are located in <em>hbase-thrift/src/main/resources</em>. For the change to be effective, you will need to regenerate the classes. You can use maven profile <code>compile-thrift</code> to do this.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn compile -Pcompile-thrift</code></pre> </div> </div> <div class="paragraph"> <p>You may also want to define <code>thrift.path</code> for the thrift binary, using the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> mvn compile -Pcompile-thrift -Dthrift.path=/opt/local/bin/thrift</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="_build_a_tarball"><a class="anchor" href="#_build_a_tarball"></a>138.1.6. Build a Tarball</h4> <div class="paragraph"> <p>You can build a tarball without going through the release process described in <a href="#releasing">releasing</a>, by running the following command:</p> </div> <div class="listingblock"> <div class="content"> <pre>mvn -DskipTests clean install && mvn -DskipTests package assembly:single</pre> </div> </div> <div class="paragraph"> <p>The distribution tarball is built in <em>hbase-assembly/target/hbase-<version>-bin.tar.gz</em>.</p> </div> <div class="paragraph"> <p>You can install or deploy the tarball by having the assembly:single goal before install or deploy in the maven command:</p> </div> <div class="listingblock"> <div class="content"> <pre>mvn -DskipTests package assembly:single install</pre> </div> </div> <div class="listingblock"> <div class="content"> <pre>mvn -DskipTests package assembly:single deploy</pre> </div> </div> </div> <div class="sect3"> <h4 id="build.gotchas"><a class="anchor" href="#build.gotchas"></a>138.1.7. Build Gotchas</h4> <div class="sect4"> <h5 id="_failure_to_find_dependencies_with_protocol_version_error"><a class="anchor" href="#_failure_to_find_dependencies_with_protocol_version_error"></a>Failure to find dependencies with "protocol_version" error</h5> <div class="paragraph"> <p>Many maven repositories, most notably Maven Central, now require TLSv1.2 for HTTPS connections. On older JDK7 instances you may need to manually add -Dhttps.protocols=TLSv1.2 to your Maven command line invocation.</p> </div> </div> <div class="sect4"> <h5 id="_maven_site_failure"><a class="anchor" href="#_maven_site_failure"></a>Maven Site failure</h5> <div class="paragraph"> <p>If you see <code>Unable to find resource 'VM_global_library.vm'</code>, ignore it. It’s not an error. It is <a href="https://issues.apache.org/jira/browse/MSITE-286">officially ugly</a> though.</p> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="releasing"><a class="anchor" href="#releasing"></a>139. Releasing Apache HBase</h2> <div class="sectionbody"> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Building against HBase 1.x</div> <div class="paragraph"> <p>HBase 1.x requires Java 7 to build. See <a href="#java">java</a> for Java requirements per HBase release.</p> </div> </td> </tr> </table> </div> <div id="maven.settings.xml" class="exampleblock"> <div class="title">Example 50. Example <em>~/.m2/settings.xml</em> File</div> <div class="content"> <div class="paragraph"> <p>Publishing to maven requires you sign the artifacts you want to upload. For the build to sign them for you, you a properly configured <em>settings.xml</em> in your local repository under <em>.m2</em>, such as the following.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><settings</span> <span class="attribute-name">xmlns</span>=<span class="string"><span class="delimiter">"</span><span class="content">http://maven.apache.org/SETTINGS/1.0.0</span><span class="delimiter">"</span></span> <span class="attribute-name">xmlns:xsi</span>=<span class="string"><span class="delimiter">"</span><span class="content">http://www.w3.org/2001/XMLSchema-instance</span><span class="delimiter">"</span></span> <span class="attribute-name">xsi:schemaLocation</span>=<span class="string"><span class="delimiter">"</span><span class="content">http://maven.apache.org/SETTINGS/1.0.0</span> <span class="content">http://maven.apache.org/xsd/settings-1.0.0.xsd</span><span class="delimiter">"</span></span><span class="tag">></span> <span class="tag"><servers></span> <span class="error"><</span>!- To publish a snapshot of some part of Maven --<span class="error">></span> <span class="tag"><server></span> <span class="tag"><id></span>apache.snapshots.https<span class="tag"></id></span> <span class="tag"><username></span>YOUR_APACHE_ID <span class="tag"></username></span> <span class="tag"><password></span>YOUR_APACHE_PASSWORD <span class="tag"></password></span> <span class="tag"></server></span> <span class="comment"></span> <span class="comment"></span> <span class="tag"><server></span> <span class="tag"><id></span>apache.releases.https<span class="tag"></id></span> <span class="tag"><username></span>YOUR_APACHE_ID <span class="tag"></username></span> <span class="tag"><password></span>YOUR_APACHE_PASSWORD <span class="tag"></password></span> <span class="tag"></server></span> <span class="tag"></servers></span> <span class="tag"><profiles></span> <span class="tag"><profile></span> <span class="tag"><id></span>apache-release<span class="tag"></id></span> <span class="tag"><properties></span> <span class="tag"><gpg.keyname></span>YOUR_KEYNAME<span class="tag"></gpg.keyname></span> <span class="comment"></span> <span class="tag"><gpg.passphrase></span>YOUR_KEY_PASSWORD <span class="tag"></gpg.passphrase></span> <span class="tag"></properties></span> <span class="tag"></profile></span> <span class="tag"></profiles></span> <span class="tag"></settings></span></code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="maven.release"><a class="anchor" href="#maven.release"></a>139.1. Making a Release Candidate</h3> <div class="paragraph"> <p>Only committers may make releases of hbase artifacts.</p> </div> <div class="paragraph"> <div class="title">Before You Begin</div> <p>Make sure your environment is properly set up. Maven and Git are the main tooling used in the below. You’ll need a properly configured <em>settings.xml</em> file in your local <em>~/.m2</em> maven repository with logins for apache repos (See <a href="#maven.settings.xml">Example <em>~/.m2/settings.xml</em> File</a>). You will also need to have a published signing key. Browse the Hadoop <a href="http://wiki.apache.org/hadoop/HowToRelease">How To Release</a> wiki page on how to release. It is a model for most of the instructions below. It often has more detail on particular steps, for example, on adding your code signing key to the project KEYS file up in Apache or on how to update JIRA in preparation for release.</p> </div> <div class="paragraph"> <p>Before you make a release candidate, do a practice run by deploying a SNAPSHOT. Check to be sure recent builds have been passing for the branch from where you are going to take your release. You should also have tried recent branch tips out on a cluster under load, perhaps by running the <code>hbase-it</code> integration test suite for a few hours to 'burn in' the near-candidate bits.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Specifying the Heap Space for Maven</div> <div class="paragraph"> <p>You may run into OutOfMemoryErrors building, particularly building the site and documentation. Up the heap for Maven by setting the <code>MAVEN_OPTS</code> variable. You can prefix the variable to the Maven command, as in the following example:</p> </div> <div class="listingblock"> <div class="content"> <pre>MAVEN_OPTS="-Xmx4g -XX:MaxPermSize=256m" mvn package</pre> </div> </div> <div class="paragraph"> <p>You could also set this in an environment variable or alias in your shell.</p> </div> </td> </tr> </table> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="paragraph"> <p>The script <em>dev-support/make_rc.sh</em> automates many of the below steps. It will checkout a tag, clean the checkout, build src and bin tarballs, and deploy the built jars to repository.apache.org. It does NOT do the modification of the <em>CHANGES.txt</em> for the release, the checking of the produced artifacts to ensure they are 'good' — e.g. extracting the produced tarballs, verifying that they look right, then starting HBase and checking that everything is running correctly — or the signing and pushing of the tarballs to <a href="https://people.apache.org">people.apache.org</a>. Take a look. Modify/improve as you see fit.</p> </div> </td> </tr> </table> </div> <div class="olist arabic"> <div class="title">Procedure: Release Procedure</div> <ol class="arabic"> <li> <p>Update the <em>CHANGES.txt</em> file and the POM files.</p> <div class="paragraph"> <p>Update <em>CHANGES.txt</em> with the changes since the last release. Make sure the URL to the JIRA points to the proper location which lists fixes for this release. Adjust the version in all the POM files appropriately. If you are making a release candidate, you must remove the <code>-SNAPSHOT</code> label from all versions in all pom.xml files. If you are running this receipe to publish a snapshot, you must keep the <code>-SNAPSHOT</code> suffix on the hbase version. The <a href="http://www.mojohaus.org/versions-maven-plugin/">Versions Maven Plugin</a> can be of use here. To set a version in all the many poms of the hbase multi-module project, use a command like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ mvn clean org.codehaus.mojo:versions-maven-plugin:2.5:set -DnewVersion=1.5.0</code></pre> </div> </div> <div class="paragraph"> <p>Make sure all versions in poms are changed! Checkin the <em>CHANGES.txt</em> and any maven version changes.</p> </div> </li> <li> <p>Update the documentation.</p> <div class="paragraph"> <p>Update the documentation under <em>src/main/asciidoc</em>. This usually involves copying the latest from master branch and making version-particular adjustments to suit this release candidate version.</p> </div> </li> <li> <p>Clean the checkout dir</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ mvn clean $ git clean -f -x -d</code></pre> </div> </div> </li> <li> <p>Run Apache-Rat Check licenses are good</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ mvn apache-rat</code></pre> </div> </div> <div class="paragraph"> <p>If the above fails, check the rat log.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ grep 'Rat check' patchprocess/mvn_apache_rat.log</code></pre> </div> </div> </li> <li> <p>Create a release tag. Presuming you have run basic tests, the rat check, passes and all is looking good, now is the time to tag the release candidate (You always remove the tag if you need to redo). To tag, do what follows substituting in the version appropriate to your build. All tags should be signed tags; i.e. pass the <em>-s</em> option (See <a href="http://https://git-scm.com/book/id/v2/Git-Tools-Signing-Your-Work">Signing Your Work</a> for how to set up your git environment for signing).</p> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ git tag -s 1.5.0-RC0 -m "Tagging the 1.5.0 first Releae Candidate (Candidates start at zero)"</code></pre> </div> </div> </li> </ol> </div> <div class="paragraph"> <p>Or, if you are making a release, tags should have a <em>rel/</em> prefix to ensure they are preserved in the Apache repo as in:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">+$ git tag -s rel/1.5.0 -m "Tagging the 1.5.0 Release"</code></pre> </div> </div> <div class="paragraph"> <p>Push the (specific) tag (only) so others have access.</p> </div> <div class="paragraph"> <p>+</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ git push origin 1.5.0-RC0</code></pre> </div> </div> <div class="paragraph"> <p>+ For how to delete tags, see <a href="http://www.manikrathee.com/how-to-delete-a-tag-in-git.html">How to Delete a Tag</a>. Covers deleting tags that have not yet been pushed to the remote Apache repo as well as delete of tags pushed to Apache.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Build the source tarball.</p> <div class="paragraph"> <p>Now, build the source tarball. Lets presume we are building the source tarball for the tag <em>1.5.0-RC0</em> into <em>/tmp/hbase-1.5.0-RC0/</em> (This step requires that the mvn and git clean steps described above have just been done).</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ git archive --format=tar.gz --output="/tmp/hbase-1.5.0-RC0/hbase-1.5.0-src.tar.gz" --prefix="hbase-1.5.0/" $git_tag</code></pre> </div> </div> </li> </ol> </div> <div class="paragraph"> <p>Above we generate the hbase-1.5.0-src.tar.gz tarball into the <em>/tmp/hbase-1.5.0-RC0</em> build output directory (We don’t want the <em>RC0</em> in the name or prefix. These bits are currently a release candidate but if the VOTE passes, they will become the release so we do not taint the artifact names with <em>RCX</em>).</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Build the binary tarball. Next, build the binary tarball. Add the <code>-Prelease</code> profile when building. It runs the license apache-rat check among other rules that help ensure all is wholesome. Do it in two steps.</p> </li> </ol> </div> <div class="paragraph"> <p>First install into the local repository</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ mvn clean install -DskipTests -Prelease</code></pre> </div> </div> <div class="paragraph"> <p>Next, generate documentation and assemble the tarball. Be warned, this next step can take a good while, a couple of hours generating site documentation.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ mvn install -DskipTests site assembly:single -Prelease</code></pre> </div> </div> <div class="paragraph"> <p>+ Otherwise, the build complains that hbase modules are not in the maven repository when you try to do it all in one step, especially on a fresh repository. It seems that you need the install goal in both steps.</p> </div> <div class="paragraph"> <p>+ Extract the generated tarball — you’ll find it under <em>hbase-assembly/target</em> and check it out. Look at the documentation, see if it runs, etc. If good, copy the tarball beside the source tarball in the build output directory.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Deploy to the Maven Repository.</p> <div class="paragraph"> <p>Next, deploy HBase to the Apache Maven repository. Add the apache-release` profile when running the <code>mvn deploy</code> command. This profile comes from the Apache parent pom referenced by our pom files. It does signing of your artifacts published to Maven, as long as the <em>settings.xml</em> is configured correctly, as described in <a href="#maven.settings.xml">Example <em>~/.m2/settings.xml</em> File</a>. This step depends on the local repository having been populate by the just-previous bin tarball build.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ mvn deploy -DskipTests -Papache-release -Prelease</code></pre> </div> </div> <div class="paragraph"> <p>This command copies all artifacts up to a temporary staging Apache mvn repository in an 'open' state. More work needs to be done on these maven artifacts to make them generally available.</p> </div> <div class="paragraph"> <p>We do not release HBase tarball to the Apache Maven repository. To avoid deploying the tarball, do not include the <code>assembly:single</code> goal in your <code>mvn deploy</code> command. Check the deployed artifacts as described in the next section.</p> </div> </li> </ol> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">make_rc.sh</div> <div class="paragraph"> <p>If you run the <em>dev-support/make_rc.sh</em> script, this is as far as it takes you. To finish the release, take up the script from here on out.</p> </div> </td> </tr> </table> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Make the Release Candidate available.</p> <div class="paragraph"> <p>The artifacts are in the maven repository in the staging area in the 'open' state. While in this 'open' state you can check out what you’ve published to make sure all is good. To do this, log in to Apache’s Nexus at <a href="https://repository.apache.org">repository.apache.org</a> using your Apache ID. Find your artifacts in the staging repository. Click on 'Staging Repositories' and look for a new one ending in "hbase" with a status of 'Open', select it. Use the tree view to expand the list of repository contents and inspect if the artifacts you expect are present. Check the POMs. As long as the staging repo is open you can re-upload if something is missing or built incorrectly.</p> </div> <div class="paragraph"> <p>If something is seriously wrong and you would like to back out the upload, you can use the 'Drop' button to drop and delete the staging repository. Sometimes the upload fails in the middle. This is another reason you might have to 'Drop' the upload from the staging repository.</p> </div> <div class="paragraph"> <p>If it checks out, close the repo using the 'Close' button. The repository must be closed before a public URL to it becomes available. It may take a few minutes for the repository to close. Once complete you’ll see a public URL to the repository in the Nexus UI. You may also receive an email with the URL. Provide the URL to the temporary staging repository in the email that announces the release candidate. (Folks will need to add this repo URL to their local poms or to their local <em>settings.xml</em> file to pull the published release candidate artifacts.)</p> </div> <div class="paragraph"> <p>When the release vote concludes successfully, return here and click the 'Release' button to release the artifacts to central. The release process will automatically drop and delete the staging repository.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">hbase-downstreamer</div> <div class="paragraph"> <p>See the <a href="https://github.com/saintstack/hbase-downstreamer">hbase-downstreamer</a> test for a simple example of a project that is downstream of HBase an depends on it. Check it out and run its simple test to make sure maven artifacts are properly deployed to the maven repository. Be sure to edit the pom to point to the proper staging repository. Make sure you are pulling from the repository when tests run and that you are not getting from your local repository, by either passing the <code>-U</code> flag or deleting your local repo content and check maven is pulling from remote out of the staging repository.</p> </div> </td> </tr> </table> </div> </li> </ol> </div> <div class="paragraph"> <p>See <a href="https://www.apache.org/dev/publishing-maven-artifacts.html">Publishing Maven Artifacts</a> for some pointers on this maven staging process.</p> </div> <div class="paragraph"> <p>If the HBase version ends in <code>-SNAPSHOT</code>, the artifacts go elsewhere. They are put into the Apache snapshots repository directly and are immediately available. Making a SNAPSHOT release, this is what you want to happen.</p> </div> <div class="paragraph"> <p>At this stage, you have two tarballs in your 'build output directory' and a set of artifacts in a staging area of the maven repository, in the 'closed' state. Next sign, fingerprint and then 'stage' your release candiate build output directory via svnpubsub by committing your directory to <a href="https://dist.apache.org/repos/dist/dev/hbase/">The 'dev' distribution directory</a> (See comments on <a href="https://issues.apache.org/jira/browse/HBASE-10554">HBASE-10554 Please delete old releases from mirroring system</a> but in essence it is an svn checkout of <a href="https://dist.apache.org/repos/dist/dev/hbase — releases" class="bare">https://dist.apache.org/repos/dist/dev/hbase — releases</a> are at <a href="https://dist.apache.org/repos/dist/release/hbase" class="bare">https://dist.apache.org/repos/dist/release/hbase</a>). In the <em>version directory</em> run the following commands:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ for i in *.tar.gz; do echo $i; gpg --print-md MD5 $i > $i.md5 ; done $ for i in *.tar.gz; do echo $i; gpg --print-md SHA512 $i > $i.sha ; done $ for i in *.tar.gz; do echo $i; gpg --armor --output $i.asc --detach-sig $i ; done $ cd .. # Presuming our 'build output directory' is named 1.5.0RC0, copy it to the svn checkout of the dist dev dir # in this case named hbase.dist.dev.svn $ cd /Users/stack/checkouts/hbase.dist.dev.svn $ svn info Path: . Working Copy Root Path: /Users/stack/checkouts/hbase.dist.dev.svn URL: https://dist.apache.org/repos/dist/dev/hbase Repository Root: https://dist.apache.org/repos/dist Repository UUID: 0d268c88-bc11-4956-87df-91683dc98e59 Revision: 15087 Node Kind: directory Schedule: normal Last Changed Author: ndimiduk Last Changed Rev: 15045 Last Changed Date: 2016-08-28 11:13:36 -0700 (Sun, 28 Aug 2016) $ mv 1.5.0RC0 /Users/stack/checkouts/hbase.dist.dev.svn $ svn add 1.5.0RC0 $ svn commit ...</code></pre> </div> </div> <div class="paragraph"> <p>+ Ensure it actually gets published by checking <a href="https://dist.apache.org/repos/dist/dev/hbase/">https://dist.apache.org/repos/dist/dev/hbase/</a>.</p> </div> <div class="paragraph"> <p>Announce the release candidate on the mailing list and call a vote.</p> </div> </div> <div class="sect2"> <h3 id="maven.snapshot"><a class="anchor" href="#maven.snapshot"></a>139.2. Publishing a SNAPSHOT to maven</h3> <div class="paragraph"> <p>Make sure your <em>settings.xml</em> is set up properly (see <a href="#maven.settings.xml">Example <em>~/.m2/settings.xml</em> File</a>). Make sure the hbase version includes <code>-SNAPSHOT</code> as a suffix. Following is an example of publishing SNAPSHOTS of a release that had an hbase version of 1.5.0 in its poms.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> $ mvn clean install -DskipTests javadoc:aggregate site assembly:single -Prelease $ mvn -DskipTests deploy -Papache-release</code></pre> </div> </div> <div class="paragraph"> <p>The <em>make_rc.sh</em> script mentioned above (see <a href="#maven.release">maven.release</a>) can help you publish <code>SNAPSHOTS</code>. Make sure your <code>hbase.version</code> has a <code>-SNAPSHOT</code> suffix before running the script. It will put a snapshot up into the apache snapshot repository for you.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbase.rc.voting"><a class="anchor" href="#hbase.rc.voting"></a>140. Voting on Release Candidates</h2> <div class="sectionbody"> <div class="paragraph"> <p>Everyone is encouraged to try and vote on HBase release candidates. Only the votes of PMC members are binding. PMC members, please read this WIP doc on policy voting for a release candidate, <a href="https://github.com/rectang/asfrelease/blob/master/release.md">Release Policy</a>.</p> </div> <div class="quoteblock"> <blockquote> <div class="paragraph"> <p>Before casting +1 binding votes, individuals are required to download the signed source code package onto their own hardware, compile it as provided, and test the resulting executable on their own platform, along with also validating cryptographic signatures and verifying that the package meets the requirements of the ASF policy on releases.</p> </div> </blockquote> </div> <div class="paragraph"> <p>Regards the latter, run <code>mvn apache-rat:check</code> to verify all files are suitably licensed. See <a href="http://search-hadoop.com/m/DHED4dhFaU">HBase, mail # dev - On recent discussion clarifying ASF release policy</a> for how we arrived at this process.</p> </div> </div> </div> <div class="sect1"> <h2 id="documentation"><a class="anchor" href="#documentation"></a>141. Generating the HBase Reference Guide</h2> <div class="sectionbody"> <div class="paragraph"> <p>The manual is marked up using Asciidoc. We then use the <a href="http://asciidoctor.org/docs/asciidoctor-maven-plugin/">Asciidoctor maven plugin</a> to transform the markup to html. This plugin is run when you specify the site goal as in when you run mvn site. See <a href="#appendix_contributing_to_documentation">appendix contributing to documentation</a> for more information on building the documentation.</p> </div> </div> </div> <div class="sect1"> <h2 id="hbase.org"><a class="anchor" href="#hbase.org"></a>142. Updating <a href="https://hbase.apache.org">hbase.apache.org</a></h2> <div class="sectionbody"> <div class="sect2"> <h3 id="hbase.org.site.contributing"><a class="anchor" href="#hbase.org.site.contributing"></a>142.1. Contributing to hbase.apache.org</h3> <div class="paragraph"> <p>See <a href="#appendix_contributing_to_documentation">appendix contributing to documentation</a> for more information on contributing to the documentation or website.</p> </div> </div> <div class="sect2"> <h3 id="hbase.org.site.publishing"><a class="anchor" href="#hbase.org.site.publishing"></a>142.2. Publishing <a href="https://hbase.apache.org">hbase.apache.org</a></h3> <div class="paragraph"> <p>See <a href="#website_publish">[website_publish]</a> for instructions on publishing the website and documentation.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbase.tests"><a class="anchor" href="#hbase.tests"></a>143. Tests</h2> <div class="sectionbody"> <div class="paragraph"> <p>Developers, at a minimum, should familiarize themselves with the unit test detail; unit tests in HBase have a character not usually seen in other projects.</p> </div> <div class="paragraph"> <p>This information is about unit tests for HBase itself. For developing unit tests for your HBase applications, see <a href="#unit.tests">unit.tests</a>.</p> </div> <div class="sect2"> <h3 id="hbase.moduletests"><a class="anchor" href="#hbase.moduletests"></a>143.1. Apache HBase Modules</h3> <div class="paragraph"> <p>As of 0.96, Apache HBase is split into multiple modules. This creates "interesting" rules for how and where tests are written. If you are writing code for <code>hbase-server</code>, see <a href="#hbase.unittests">hbase.unittests</a> for how to write your tests. These tests can spin up a minicluster and will need to be categorized. For any other module, for example <code>hbase-common</code>, the tests must be strict unit tests and just test the class under test - no use of the HBaseTestingUtility or minicluster is allowed (or even possible given the dependency tree).</p> </div> <div class="sect3"> <h4 id="hbase.moduletest.shell"><a class="anchor" href="#hbase.moduletest.shell"></a>143.1.1. Testing the HBase Shell</h4> <div class="paragraph"> <p>The HBase shell and its tests are predominantly written in jruby. In order to make these tests run as a part of the standard build, there is a single JUnit test, <code>TestShell</code>, that takes care of loading the jruby implemented tests and running them. You can run all of these tests from the top level with:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> mvn clean test -Dtest=TestShell</code></pre> </div> </div> <div class="paragraph"> <p>Alternatively, you may limit the shell tests that run using the system variable <code>shell.test</code>. This value should specify the ruby literal equivalent of a particular test case by name. For example, the tests that cover the shell commands for altering tables are contained in the test case <code>AdminAlterTableTest</code> and you can run them with:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> mvn clean test -Dtest=TestShell -Dshell.test=/AdminAlterTableTest/</code></pre> </div> </div> <div class="paragraph"> <p>You may also use a <a href="http://docs.ruby-doc.com/docs/ProgrammingRuby/html/language.html#UJ">Ruby Regular Expression literal</a> (in the <code>/pattern/</code> style) to select a set of test cases. You can run all of the HBase admin related tests, including both the normal administration and the security administration, with the command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> mvn clean test -Dtest=TestShell -Dshell.test=/.*Admin.*Test/</code></pre> </div> </div> <div class="paragraph"> <p>In the event of a test failure, you can see details by examining the XML version of the surefire report results</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne"> vim hbase-shell/target/surefire-reports/TEST-org.apache.hadoop.hbase.client.TestShell.xml</code></pre> </div> </div> </div> <div class="sect3"> <h4 id="hbase.moduletest.run"><a class="anchor" href="#hbase.moduletest.run"></a>143.1.2. Running Tests in other Modules</h4> <div class="paragraph"> <p>If the module you are developing in has no other dependencies on other HBase modules, then you can cd into that module and just run:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn test</code></pre> </div> </div> <div class="paragraph"> <p>which will just run the tests IN THAT MODULE. If there are other dependencies on other modules, then you will have run the command from the ROOT HBASE DIRECTORY. This will run the tests in the other modules, unless you specify to skip the tests in that module. For instance, to skip the tests in the hbase-server module, you would run:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean test -PskipServerTests</code></pre> </div> </div> <div class="paragraph"> <p>from the top level directory to run all the tests in modules other than hbase-server. Note that you can specify to skip tests in multiple modules as well as just for a single module. For example, to skip the tests in <code>hbase-server</code> and <code>hbase-common</code>, you would run:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean test -PskipServerTests -PskipCommonTests</code></pre> </div> </div> <div class="paragraph"> <p>Also, keep in mind that if you are running tests in the <code>hbase-server</code> module you will need to apply the maven profiles discussed in <a href="#hbase.unittests.cmds">hbase.unittests.cmds</a> to get the tests to run properly.</p> </div> </div> </div> <div class="sect2"> <h3 id="hbase.unittests"><a class="anchor" href="#hbase.unittests"></a>143.2. Unit Tests</h3> <div class="paragraph"> <p>Apache HBase test cases are subdivided into four categories: small, medium, large, and integration with corresponding JUnit <a href="https://github.com/junit-team/junit4/wiki/Categories">categories</a>: <code>SmallTests</code>, <code>MediumTests</code>, <code>LargeTests</code>, <code>IntegrationTests</code>. JUnit categories are denoted using java annotations and look like this in your unit test code.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... <span class="annotation">@Category</span>(SmallTests.class) <span class="directive">public</span> <span class="type">class</span> <span class="class">TestHRegionInfo</span> { <span class="annotation">@Test</span> <span class="directive">public</span> <span class="type">void</span> testCreateHRegionInfoName() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="comment">// ...</span> } }</code></pre> </div> </div> <div class="paragraph"> <p>The above example shows how to mark a test case as belonging to the <code>small</code> category. All test cases in HBase should have a categorization.</p> </div> <div class="paragraph"> <p>The first three categories, <code>small</code>, <code>medium</code>, and <code>large</code>, are for test cases which run when you type <code>$ mvn test</code>. In other words, these three categorizations are for HBase unit tests. The <code>integration</code> category is not for unit tests, but for integration tests. These are run when you invoke <code>$ mvn verify</code>. Integration tests are described in <a href="#integration.tests">integration.tests</a>.</p> </div> <div class="paragraph"> <p>HBase uses a patched maven surefire plugin and maven profiles to implement its unit test characterizations.</p> </div> <div class="paragraph"> <p>Keep reading to figure which annotation of the set small, medium, and large to put on your new HBase test case.</p> </div> <div class="dlist"> <div class="title">Categorizing Tests</div> <dl> <dt class="hdlist1">Small Tests </dt> <dd> <p><em>Small</em> test cases are executed in a shared JVM and individual test cases should run in 15 seconds or less; i.e. a <a href="https://en.wikipedia.org/wiki/JUnit">junit test fixture</a>, a java object made up of test methods, should finish in under 15 seconds. These test cases can not use mini cluster. These are run as part of patch pre-commit.</p> </dd> <dt class="hdlist1">Medium Tests </dt> <dd> <p><em>Medium</em> test cases are executed in separate JVM and individual test case should run in 50 seconds or less. Together, they should take less than 30 minutes, and are quite stable in their results. These test cases can use a mini cluster. These are run as part of patch pre-commit.</p> </dd> <dt class="hdlist1">Large Tests </dt> <dd> <p><em>Large</em> test cases are everything else. They are typically large-scale tests, regression tests for specific bugs, timeout tests, performance tests. They are executed before a commit on the pre-integration machines. They can be run on the developer machine as well.</p> </dd> <dt class="hdlist1">Integration Tests </dt> <dd> <p><em>Integration</em> tests are system level tests. See <a href="#integration.tests">integration.tests</a> for more info.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="hbase.unittests.cmds"><a class="anchor" href="#hbase.unittests.cmds"></a>143.3. Running tests</h3> <div class="sect3"> <h4 id="hbase.unittests.cmds.test"><a class="anchor" href="#hbase.unittests.cmds.test"></a>143.3.1. Default: small and medium category tests</h4> <div class="paragraph"> <p>Running <code>mvn test</code> will execute all small tests in a single JVM (no fork) and then medium tests in a separate JVM for each test instance. Medium tests are NOT executed if there is an error in a small test. Large tests are NOT executed. There is one report for small tests, and one report for medium tests if they are executed.</p> </div> </div> <div class="sect3"> <h4 id="hbase.unittests.cmds.test.runalltests"><a class="anchor" href="#hbase.unittests.cmds.test.runalltests"></a>143.3.2. Running all tests</h4> <div class="paragraph"> <p>Running <code>mvn test -P runAllTests</code> will execute small tests in a single JVM then medium and large tests in a separate JVM for each test. Medium and large tests are NOT executed if there is an error in a small test. Large tests are NOT executed if there is an error in a small or medium test. There is one report for small tests, and one report for medium and large tests if they are executed.</p> </div> </div> <div class="sect3"> <h4 id="hbase.unittests.cmds.test.localtests.mytest"><a class="anchor" href="#hbase.unittests.cmds.test.localtests.mytest"></a>143.3.3. Running a single test or all tests in a package</h4> <div class="paragraph"> <p>To run an individual test, e.g. <code>MyTest</code>, rum <code>mvn test -Dtest=MyTest</code> You can also pass multiple, individual tests as a comma-delimited list:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">mvn test -Dtest=MyTest1,MyTest2,MyTest3</code></pre> </div> </div> <div class="paragraph"> <p>You can also pass a package, which will run all tests under the package:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">mvn test '-Dtest=org.apache.hadoop.hbase.client.*'</code></pre> </div> </div> <div class="paragraph"> <p>When <code>-Dtest</code> is specified, the <code>localTests</code> profile will be used. It will use the official release of maven surefire, rather than our custom surefire plugin, and the old connector (The HBase build uses a patched version of the maven surefire plugin). Each junit test is executed in a separate JVM (A fork per test class). There is no parallelization when tests are running in this mode. You will see a new message at the end of the -report: <code>"[INFO] Tests are skipped"</code>. It’s harmless. However, you need to make sure the sum of <code>Tests run:</code> in the <code>Results:</code> section of test reports matching the number of tests you specified because no error will be reported when a non-existent test case is specified.</p> </div> </div> <div class="sect3"> <h4 id="hbase.unittests.cmds.test.profiles"><a class="anchor" href="#hbase.unittests.cmds.test.profiles"></a>143.3.4. Other test invocation permutations</h4> <div class="paragraph"> <p>Running <code>mvn test -P runSmallTests</code> will execute "small" tests only, using a single JVM.</p> </div> <div class="paragraph"> <p>Running <code>mvn test -P runMediumTests</code> will execute "medium" tests only, launching a new JVM for each test-class.</p> </div> <div class="paragraph"> <p>Running <code>mvn test -P runLargeTests</code> will execute "large" tests only, launching a new JVM for each test-class.</p> </div> <div class="paragraph"> <p>For convenience, you can run <code>mvn test -P runDevTests</code> to execute both small and medium tests, using a single JVM.</p> </div> </div> <div class="sect3"> <h4 id="hbase.unittests.test.faster"><a class="anchor" href="#hbase.unittests.test.faster"></a>143.3.5. Running tests faster</h4> <div class="paragraph"> <p>By default, <code>$ mvn test -P runAllTests</code> runs 5 tests in parallel. It can be increased on a developer’s machine. Allowing that you can have 2 tests in parallel per core, and you need about 2GB of memory per test (at the extreme), if you have an 8 core, 24GB box, you can have 16 tests in parallel. but the memory available limits it to 12 (24/2), To run all tests with 12 tests in parallel, do this: mvn test -P runAllTests -Dsurefire.secondPartForkCount=12. If using a version earlier than 2.0, do: +mvn test -P runAllTests -Dsurefire.secondPartThreadCount=12 +. To increase the speed, you can as well use a ramdisk. You will need 2GB of memory to run all tests. You will also need to delete the files between two test run. The typical way to configure a ramdisk on Linux is:</p> </div> <div class="listingblock"> <div class="content"> <pre>$ sudo mkdir /ram2G sudo mount -t tmpfs -o size=2048M tmpfs /ram2G</pre> </div> </div> <div class="paragraph"> <p>You can then use it to run all HBase tests on 2.0 with the command:</p> </div> <div class="listingblock"> <div class="content"> <pre>mvn test -P runAllTests -Dsurefire.secondPartForkCount=12 -Dtest.build.data.basedirectory=/ram2G</pre> </div> </div> <div class="paragraph"> <p>On earlier versions, use:</p> </div> <div class="listingblock"> <div class="content"> <pre>mvn test -P runAllTests -Dsurefire.secondPartThreadCount=12 -Dtest.build.data.basedirectory=/ram2G</pre> </div> </div> </div> <div class="sect3"> <h4 id="hbase.unittests.cmds.test.hbasetests"><a class="anchor" href="#hbase.unittests.cmds.test.hbasetests"></a>143.3.6. hbasetests.sh</h4> <div class="paragraph"> <p>It’s also possible to use the script hbasetests.sh. This script runs the medium and large tests in parallel with two maven instances, and provides a single report. This script does not use the hbase version of surefire so no parallelization is being done other than the two maven instances the script sets up. It must be executed from the directory which contains the <em>pom.xml</em>.</p> </div> <div class="paragraph"> <p>For example running ./dev-support/hbasetests.sh will execute small and medium tests. Running ./dev-support/hbasetests.sh runAllTests will execute all tests. Running ./dev-support/hbasetests.sh replayFailed will rerun the failed tests a second time, in a separate jvm and without parallelisation.</p> </div> </div> <div class="sect3"> <h4 id="hbase.unittests.resource.checker"><a class="anchor" href="#hbase.unittests.resource.checker"></a>143.3.7. Test Resource Checker</h4> <div class="paragraph"> <p>A custom Maven SureFire plugin listener checks a number of resources before and after each HBase unit test runs and logs its findings at the end of the test output files which can be found in <em>target/surefire-reports</em> per Maven module (Tests write test reports named for the test class into this directory. Check the <em>*-out.txt</em> files). The resources counted are the number of threads, the number of file descriptors, etc. If the number has increased, it adds a <em>LEAK?</em> comment in the logs. As you can have an HBase instance running in the background, some threads can be deleted/created without any specific action in the test. However, if the test does not work as expected, or if the test should not impact these resources, it’s worth checking these log lines <span class="computeroutput">...hbase.ResourceChecker(157): before...</span> and <span class="computeroutput">...hbase.ResourceChecker(157): after...</span>. For example:</p> </div> <div class="listingblock"> <div class="content"> <pre>2012-09-26 09:22:15,315 INFO [pool-1-thread-1] hbase.ResourceChecker(157): after: regionserver.TestColumnSeeking#testReseeking Thread=65 (was 65), OpenFileDescriptor=107 (was 107), MaxFileDescriptor=10240 (was 10240), ConnectionCount=1 (was 1)</pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="hbase.tests.writing"><a class="anchor" href="#hbase.tests.writing"></a>143.4. Writing Tests</h3> <div class="sect3"> <h4 id="hbase.tests.rules"><a class="anchor" href="#hbase.tests.rules"></a>143.4.1. General rules</h4> <div class="ulist"> <ul> <li> <p>As much as possible, tests should be written as category small tests.</p> </li> <li> <p>All tests must be written to support parallel execution on the same machine, hence they should not use shared resources as fixed ports or fixed file names.</p> </li> <li> <p>Tests should not overlog. More than 100 lines/second makes the logs complex to read and use i/o that are hence not available for the other tests.</p> </li> <li> <p>Tests can be written with <code>HBaseTestingUtility</code>. This class offers helper functions to create a temp directory and do the cleanup, or to start a cluster.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="hbase.tests.categories"><a class="anchor" href="#hbase.tests.categories"></a>143.4.2. Categories and execution time</h4> <div class="ulist"> <ul> <li> <p>All tests must be categorized, if not they could be skipped.</p> </li> <li> <p>All tests should be written to be as fast as possible.</p> </li> <li> <p>See <a href="#hbase.unittests">hbase.unittests</a> for test case categories and corresponding timeouts. This should ensure a good parallelization for people using it, and ease the analysis when the test fails.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="hbase.tests.sleeps"><a class="anchor" href="#hbase.tests.sleeps"></a>143.4.3. Sleeps in tests</h4> <div class="paragraph"> <p>Whenever possible, tests should not use <span class="method">Thread.sleep</span>, but rather waiting for the real event they need. This is faster and clearer for the reader. Tests should not do a <span class="method">Thread.sleep</span> without testing an ending condition. This allows understanding what the test is waiting for. Moreover, the test will work whatever the machine performance is. Sleep should be minimal to be as fast as possible. Waiting for a variable should be done in a 40ms sleep loop. Waiting for a socket operation should be done in a 200 ms sleep loop.</p> </div> </div> <div class="sect3"> <h4 id="hbase.tests.cluster"><a class="anchor" href="#hbase.tests.cluster"></a>143.4.4. Tests using a cluster</h4> <div class="paragraph"> <p>Tests using a HRegion do not have to start a cluster: A region can use the local file system. Start/stopping a cluster cost around 10 seconds. They should not be started per test method but per test class. Started cluster must be shutdown using <span class="method">HBaseTestingUtility#shutdownMiniCluster</span>, which cleans the directories. As most as possible, tests should use the default settings for the cluster. When they don’t, they should document it. This will allow to share the cluster later.</p> </div> </div> <div class="sect3"> <h4 id="hbase.tests.example.code"><a class="anchor" href="#hbase.tests.example.code"></a>143.4.5. Tests Skeleton Code</h4> <div class="paragraph"> <p>Here is a test skeleton code with Categorization and a Category-based timeout rule to copy and paste and use as basis for test contribution.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="comment">/** * Describe what this testcase tests. Talk about resources initialized in @BeforeClass (before * any test is run) and before each test is run, etc. */</span> <span class="comment">// Specify the category as explained in <<hbase.unittests,hbase.unittests>>.</span> <span class="annotation">@Category</span>(SmallTests.class) <span class="directive">public</span> <span class="type">class</span> <span class="class">TestExample</span> { <span class="comment">// Replace the TestExample.class in the below with the name of your test fixture class.</span> <span class="directive">private</span> <span class="directive">static</span> <span class="directive">final</span> Log LOG = LogFactory.getLog(TestExample.class); <span class="comment">// Handy test rule that allows you subsequently get the name of the current method. See</span> <span class="comment">// down in 'testExampleFoo()' where we use it to log current test's name.</span> <span class="annotation">@Rule</span> <span class="directive">public</span> TestName testName = <span class="keyword">new</span> TestName(); <span class="comment">// The below rule does two things. It decides the timeout based on the category</span> <span class="comment">// (small/medium/large) of the testcase. This @Rule requires that the full testcase runs</span> <span class="comment">// within this timeout irrespective of individual test methods' times. The second</span> <span class="comment">// feature is we'll dump in the log when the test is done a count of threads still</span> <span class="comment">// running.</span> <span class="annotation">@Rule</span> <span class="directive">public</span> <span class="directive">static</span> TestRule timeout = CategoryBasedTimeout.builder(). withTimeout(<span class="local-variable">this</span>.getClass()).withLookingForStuckThread(<span class="predefined-constant">true</span>).build(); <span class="annotation">@Before</span> <span class="directive">public</span> <span class="type">void</span> setUp() <span class="directive">throws</span> <span class="exception">Exception</span> { } <span class="annotation">@After</span> <span class="directive">public</span> <span class="type">void</span> tearDown() <span class="directive">throws</span> <span class="exception">Exception</span> { } <span class="annotation">@Test</span> <span class="directive">public</span> <span class="type">void</span> testExampleFoo() { LOG.info(<span class="string"><span class="delimiter">"</span><span class="content">Running test </span><span class="delimiter">"</span></span> + testName.getMethodName()); } }</code></pre> </div> </div> </div> </div> <div class="sect2"> <h3 id="integration.tests"><a class="anchor" href="#integration.tests"></a>143.5. Integration Tests</h3> <div class="paragraph"> <p>HBase integration/system tests are tests that are beyond HBase unit tests. They are generally long-lasting, sizeable (the test can be asked to 1M rows or 1B rows), targetable (they can take configuration that will point them at the ready-made cluster they are to run against; integration tests do not include cluster start/stop code), and verifying success, integration tests rely on public APIs only; they do not attempt to examine server internals asserting success/fail. Integration tests are what you would run when you need to more elaborate proofing of a release candidate beyond what unit tests can do. They are not generally run on the Apache Continuous Integration build server, however, some sites opt to run integration tests as a part of their continuous testing on an actual cluster.</p> </div> <div class="paragraph"> <p>Integration tests currently live under the <em>src/test</em> directory in the hbase-it submodule and will match the regex: <em>*<strong>/IntegrationTest</strong>.java</em>. All integration tests are also annotated with <code>@Category(IntegrationTests.class)</code>.</p> </div> <div class="paragraph"> <p>Integration tests can be run in two modes: using a mini cluster, or against an actual distributed cluster. Maven failsafe is used to run the tests using the mini cluster. IntegrationTestsDriver class is used for executing the tests against a distributed cluster. Integration tests SHOULD NOT assume that they are running against a mini cluster, and SHOULD NOT use private API’s to access cluster state. To interact with the distributed or mini cluster uniformly, <code>IntegrationTestingUtility</code>, and <code>HBaseCluster</code> classes, and public client API’s can be used.</p> </div> <div class="paragraph"> <p>On a distributed cluster, integration tests that use ChaosMonkey or otherwise manipulate services thru cluster manager (e.g. restart regionservers) use SSH to do it. To run these, test process should be able to run commands on remote end, so ssh should be configured accordingly (for example, if HBase runs under hbase user in your cluster, you can set up passwordless ssh for that user and run the test also under it). To facilitate that, <code>hbase.it.clustermanager.ssh.user</code>, <code>hbase.it.clustermanager.ssh.opts</code> and <code>hbase.it.clustermanager.ssh.cmd</code> configuration settings can be used. "User" is the remote user that cluster manager should use to perform ssh commands. "Opts" contains additional options that are passed to SSH (for example, "-i /tmp/my-key"). Finally, if you have some custom environment setup, "cmd" is the override format for the entire tunnel (ssh) command. The default string is {<code>/usr/bin/ssh %1$s %2$s%3$s%4$s "%5$s"</code>} and is a good starting point. This is a standard Java format string with 5 arguments that is used to execute the remote command. The argument 1 (%1$s) is SSH options set the via opts setting or via environment variable, 2 is SSH user name, 3 is "@" if username is set or "" otherwise, 4 is the target host name, and 5 is the logical command to execute (that may include single quotes, so don’t use them). For example, if you run the tests under non-hbase user and want to ssh as that user and change to hbase on remote machine, you can use:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">/usr/bin/ssh %1$s %2$s%3$s%4$s "su hbase - -c \"%5$s\""</code></pre> </div> </div> <div class="paragraph"> <p>That way, to kill RS (for example) integration tests may run:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">{/usr/bin/ssh some-hostname "su hbase - -c \"ps aux | ... | kill ...\""}</code></pre> </div> </div> <div class="paragraph"> <p>The command is logged in the test logs, so you can verify it is correct for your environment.</p> </div> <div class="paragraph"> <p>To disable the running of Integration Tests, pass the following profile on the command line <code>-PskipIntegrationTests</code>. For example,</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="error">$</span> mvn clean install test -Dtest=TestZooKeeper -PskipIntegrationTests</code></pre> </div> </div> <div class="sect3"> <h4 id="maven.build.commands.integration.tests.mini"><a class="anchor" href="#maven.build.commands.integration.tests.mini"></a>143.5.1. Running integration tests against mini cluster</h4> <div class="paragraph"> <p>HBase 0.92 added a <code>verify</code> maven target. Invoking it, for example by doing <code>mvn verify</code>, will run all the phases up to and including the verify phase via the maven <a href="https://maven.apache.org/plugins/maven-failsafe-plugin/">failsafe plugin</a>, running all the above mentioned HBase unit tests as well as tests that are in the HBase integration test group. After you have completed mvn install -DskipTests You can run just the integration tests by invoking:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">cd hbase-it mvn verify</code></pre> </div> </div> <div class="paragraph"> <p>If you just want to run the integration tests in top-level, you need to run two commands. First: mvn failsafe:integration-test This actually runs ALL the integration tests.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> This command will always output <code>BUILD SUCCESS</code> even if there are test failures. </td> </tr> </table> </div> <div class="paragraph"> <p>At this point, you could grep the output by hand looking for failed tests. However, maven will do this for us; just use: mvn failsafe:verify The above command basically looks at all the test results (so don’t remove the 'target' directory) for test failures and reports the results.</p> </div> <div class="sect4"> <h5 id="maven.build.commands.integration.tests2"><a class="anchor" href="#maven.build.commands.integration.tests2"></a>Running a subset of Integration tests</h5> <div class="paragraph"> <p>This is very similar to how you specify running a subset of unit tests (see above), but use the property <code>it.test</code> instead of <code>test</code>. To just run <code>IntegrationTestClassXYZ.java</code>, use: mvn failsafe:integration-test -Dit.test=IntegrationTestClassXYZ The next thing you might want to do is run groups of integration tests, say all integration tests that are named IntegrationTestClassX*.java: mvn failsafe:integration-test -Dit.test=*ClassX* This runs everything that is an integration test that matches <strong>ClassX</strong>. This means anything matching: "<strong>*/IntegrationTest*ClassX</strong>". You can also run multiple groups of integration tests using comma-delimited lists (similar to unit tests). Using a list of matches still supports full regex matching for each of the groups. This would look something like: mvn failsafe:integration-test -Dit.test=*ClassX*, *ClassY</p> </div> </div> </div> <div class="sect3"> <h4 id="maven.build.commands.integration.tests.distributed"><a class="anchor" href="#maven.build.commands.integration.tests.distributed"></a>143.5.2. Running integration tests against distributed cluster</h4> <div class="paragraph"> <p>If you have an already-setup HBase cluster, you can launch the integration tests by invoking the class <code>IntegrationTestsDriver</code>. You may have to run test-compile first. The configuration will be picked by the bin/hbase script.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn test-compile</code></pre> </div> </div> <div class="paragraph"> <p>Then launch the tests with:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">bin/hbase [--config config_dir] org.apache.hadoop.hbase.IntegrationTestsDriver</code></pre> </div> </div> <div class="paragraph"> <p>Pass <code>-h</code> to get usage on this sweet tool. Running the IntegrationTestsDriver without any argument will launch tests found under <code>hbase-it/src/test</code>, having <code>@Category(IntegrationTests.class)</code> annotation, and a name starting with <code>IntegrationTests</code>. See the usage, by passing -h, to see how to filter test classes. You can pass a regex which is checked against the full class name; so, part of class name can be used. IntegrationTestsDriver uses Junit to run the tests. Currently there is no support for running integration tests against a distributed cluster using maven (see <a href="https://issues.apache.org/jira/browse/HBASE-6201">HBASE-6201</a>).</p> </div> <div class="paragraph"> <p>The tests interact with the distributed cluster by using the methods in the <code>DistributedHBaseCluster</code> (implementing <code>HBaseCluster</code>) class, which in turn uses a pluggable <code>ClusterManager</code>. Concrete implementations provide actual functionality for carrying out deployment-specific and environment-dependent tasks (SSH, etc). The default <code>ClusterManager</code> is <code>HBaseClusterManager</code>, which uses SSH to remotely execute start/stop/kill/signal commands, and assumes some posix commands (ps, etc). Also assumes the user running the test has enough "power" to start/stop servers on the remote machines. By default, it picks up <code>HBASE_SSH_OPTS</code>, <code>HBASE_HOME</code>, <code>HBASE_CONF_DIR</code> from the env, and uses <code>bin/hbase-daemon.sh</code> to carry out the actions. Currently tarball deployments, deployments which uses <em>hbase-daemons.sh</em>, and <a href="https://incubator.apache.org/ambari/">Apache Ambari</a> deployments are supported. <em>/etc/init.d/</em> scripts are not supported for now, but it can be easily added. For other deployment options, a ClusterManager can be implemented and plugged in.</p> </div> </div> <div class="sect3"> <h4 id="maven.build.commands.integration.tests.destructive"><a class="anchor" href="#maven.build.commands.integration.tests.destructive"></a>143.5.3. Destructive integration / system tests (ChaosMonkey)</h4> <div class="paragraph"> <p>HBase 0.96 introduced a tool named <code>ChaosMonkey</code>, modeled after <a href="https://netflix.github.io/chaosmonkey/">same-named tool by Netflix’s Chaos Monkey tool</a>. ChaosMonkey simulates real-world faults in a running cluster by killing or disconnecting random servers, or injecting other failures into the environment. You can use ChaosMonkey as a stand-alone tool to run a policy while other tests are running. In some environments, ChaosMonkey is always running, in order to constantly check that high availability and fault tolerance are working as expected.</p> </div> <div class="paragraph"> <p>ChaosMonkey defines <strong>Actions</strong> and <strong>Policies</strong>.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Actions</dt> <dd> <p>Actions are predefined sequences of events, such as the following:</p> <div class="ulist"> <ul> <li> <p>Restart active master (sleep 5 sec)</p> </li> <li> <p>Restart random regionserver (sleep 5 sec)</p> </li> <li> <p>Restart random regionserver (sleep 60 sec)</p> </li> <li> <p>Restart META regionserver (sleep 5 sec)</p> </li> <li> <p>Restart ROOT regionserver (sleep 5 sec)</p> </li> <li> <p>Batch restart of 50% of regionservers (sleep 5 sec)</p> </li> <li> <p>Rolling restart of 100% of regionservers (sleep 5 sec)</p> </li> </ul> </div> </dd> <dt class="hdlist1">Policies</dt> <dd> <p>A policy is a strategy for executing one or more actions. The default policy executes a random action every minute based on predefined action weights. A given policy will be executed until ChaosMonkey is interrupted.</p> </dd> </dl> </div> <div class="paragraph"> <p>Most ChaosMonkey actions are configured to have reasonable defaults, so you can run ChaosMonkey against an existing cluster without any additional configuration. The following example runs ChaosMonkey with the default configuration:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bash">$ bin/hbase org.apache.hadoop.hbase.util.ChaosMonkey 12/11/19 23:21:57 INFO util.ChaosMonkey: Using ChaosMonkey Policy: class org.apache.hadoop.hbase.util.ChaosMonkey$PeriodicRandomActionPolicy, period:60000 12/11/19 23:21:57 INFO util.ChaosMonkey: Sleeping for 26953 to add jitter 12/11/19 23:22:24 INFO util.ChaosMonkey: Performing action: Restart active master 12/11/19 23:22:24 INFO util.ChaosMonkey: Killing master:master.example.com,60000,1353367210440 12/11/19 23:22:24 INFO hbase.HBaseCluster: Aborting Master: master.example.com,60000,1353367210440 12/11/19 23:22:24 INFO hbase.ClusterManager: Executing remote command: ps aux | grep master | grep -v grep | tr -s ' ' | cut -d ' ' -f2 | xargs kill -s SIGKILL , hostname:master.example.com 12/11/19 23:22:25 INFO hbase.ClusterManager: Executed remote command, exit code:0 , output: 12/11/19 23:22:25 INFO hbase.HBaseCluster: Waiting service:master to stop: master.example.com,60000,1353367210440 12/11/19 23:22:25 INFO hbase.ClusterManager: Executing remote command: ps aux | grep master | grep -v grep | tr -s ' ' | cut -d ' ' -f2 , hostname:master.example.com 12/11/19 23:22:25 INFO hbase.ClusterManager: Executed remote command, exit code:0 , output: 12/11/19 23:22:25 INFO util.ChaosMonkey: Killed master server:master.example.com,60000,1353367210440 12/11/19 23:22:25 INFO util.ChaosMonkey: Sleeping for:5000 12/11/19 23:22:30 INFO util.ChaosMonkey: Starting master:master.example.com 12/11/19 23:22:30 INFO hbase.HBaseCluster: Starting Master on: master.example.com 12/11/19 23:22:30 INFO hbase.ClusterManager: Executing remote command: /homes/enis/code/hbase-0.94/bin/../bin/hbase-daemon.sh --config /homes/enis/code/hbase-0.94/bin/../conf start master , hostname:master.example.com 12/11/19 23:22:31 INFO hbase.ClusterManager: Executed remote command, exit code:0 , output:starting master, logging to /homes/enis/code/hbase-0.94/bin/../logs/hbase-enis-master-master.example.com.out .... 12/11/19 23:22:33 INFO util.ChaosMonkey: Started master: master.example.com,60000,1353367210440 12/11/19 23:22:33 INFO util.ChaosMonkey: Sleeping for:51321 12/11/19 23:23:24 INFO util.ChaosMonkey: Performing action: Restart random region server 12/11/19 23:23:24 INFO util.ChaosMonkey: Killing region server:rs3.example.com,60020,1353367027826 12/11/19 23:23:24 INFO hbase.HBaseCluster: Aborting RS: rs3.example.com,60020,1353367027826 12/11/19 23:23:24 INFO hbase.ClusterManager: Executing remote command: ps aux | grep regionserver | grep -v grep | tr -s ' ' | cut -d ' ' -f2 | xargs kill -s SIGKILL , hostname:rs3.example.com 12/11/19 23:23:25 INFO hbase.ClusterManager: Executed remote command, exit code:0 , output: 12/11/19 23:23:25 INFO hbase.HBaseCluster: Waiting service:regionserver to stop: rs3.example.com,60020,1353367027826 12/11/19 23:23:25 INFO hbase.ClusterManager: Executing remote command: ps aux | grep regionserver | grep -v grep | tr -s ' ' | cut -d ' ' -f2 , hostname:rs3.example.com 12/11/19 23:23:25 INFO hbase.ClusterManager: Executed remote command, exit code:0 , output: 12/11/19 23:23:25 INFO util.ChaosMonkey: Killed region server:rs3.example.com,60020,1353367027826. Reported num of rs:6 12/11/19 23:23:25 INFO util.ChaosMonkey: Sleeping for:60000 12/11/19 23:24:25 INFO util.ChaosMonkey: Starting region server:rs3.example.com 12/11/19 23:24:25 INFO hbase.HBaseCluster: Starting RS on: rs3.example.com 12/11/19 23:24:25 INFO hbase.ClusterManager: Executing remote command: /homes/enis/code/hbase-0.94/bin/../bin/hbase-daemon.sh --config /homes/enis/code/hbase-0.94/bin/../conf start regionserver , hostname:rs3.example.com 12/11/19 23:24:26 INFO hbase.ClusterManager: Executed remote command, exit code:0 , output:starting regionserver, logging to /homes/enis/code/hbase-0.94/bin/../logs/hbase-enis-regionserver-rs3.example.com.out 12/11/19 23:24:27 INFO util.ChaosMonkey: Started region server:rs3.example.com,60020,1353367027826. Reported num of rs:6</code></pre> </div> </div> <div class="paragraph"> <p>The output indicates that ChaosMonkey started the default <code>PeriodicRandomActionPolicy</code> policy, which is configured with all the available actions. It chose to run <code>RestartActiveMaster</code> and <code>RestartRandomRs</code> actions.</p> </div> </div> <div class="sect3"> <h4 id="_available_policies"><a class="anchor" href="#_available_policies"></a>143.5.4. Available Policies</h4> <div class="paragraph"> <p>HBase ships with several ChaosMonkey policies, available in the <code>hbase/hbase-it/src/test/java/org/apache/hadoop/hbase/chaos/policies/</code> directory.</p> </div> </div> <div class="sect3"> <h4 id="chaos.monkey.properties"><a class="anchor" href="#chaos.monkey.properties"></a>143.5.5. Configuring Individual ChaosMonkey Actions</h4> <div class="paragraph"> <p>Since HBase version 1.0.0 (<a href="https://issues.apache.org/jira/browse/HBASE-11348">HBASE-11348</a>), ChaosMonkey integration tests can be configured per test run. Create a Java properties file in the HBase classpath and pass it to ChaosMonkey using the <code>-monkeyProps</code> configuration flag. Configurable properties, along with their default values if applicable, are listed in the <code>org.apache.hadoop.hbase.chaos.factories.MonkeyConstants</code> class. For properties that have defaults, you can override them by including them in your properties file.</p> </div> <div class="paragraph"> <p>The following example uses a properties file called <a href="#monkey.properties">monkey.properties</a>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ bin/hbase org.apache.hadoop.hbase.IntegrationTestIngest -m slowDeterministic -monkeyProps monkey.properties</code></pre> </div> </div> <div class="paragraph"> <p>The above command will start the integration tests and chaos monkey passing the properties file <em>monkey.properties</em>. Here is an example chaos monkey file:</p> </div> <div id="monkey.properties" class="listingblock"> <div class="title">Example ChaosMonkey Properties File</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">sdm.action1.period=<span class="integer">120000</span> sdm.action2.period=<span class="integer">40000</span> move.regions.sleep.time=<span class="integer">80000</span> move.regions.max.time=<span class="integer">1000000</span> move.regions.sleep.time=<span class="integer">80000</span> batch.restart.rs.ratio=<span class="float">0.4f</span></code></pre> </div> </div> <div class="paragraph"> <p>HBase 1.0.2 and newer adds the ability to restart HBase’s underlying ZooKeeper quorum or HDFS nodes. To use these actions, you need to configure some new properties, which have no reasonable defaults because they are deployment-specific, in your ChaosMonkey properties file, which may be <code>hbase-site.xml</code> or a different properties file.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.it.clustermanager.hadoop.home<span class="tag"></name></span> <span class="tag"><value></span>$HADOOP_HOME<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.it.clustermanager.zookeeper.home<span class="tag"></name></span> <span class="tag"><value></span>$ZOOKEEPER_HOME<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.it.clustermanager.hbase.user<span class="tag"></name></span> <span class="tag"><value></span>hbase<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.it.clustermanager.hadoop.hdfs.user<span class="tag"></name></span> <span class="tag"><value></span>hdfs<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.it.clustermanager.zookeeper.user<span class="tag"></name></span> <span class="tag"><value></span>zookeeper<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="developing"><a class="anchor" href="#developing"></a>144. Developer Guidelines</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_branches"><a class="anchor" href="#_branches"></a>144.1. Branches</h3> <div class="paragraph"> <p>We use Git for source code management and latest development happens on <code>master</code> branch. There are branches for past major/minor/maintenance releases and important features and bug fixes are often back-ported to them.</p> </div> </div> <div class="sect2"> <h3 id="_release_managers"><a class="anchor" href="#_release_managers"></a>144.2. Release Managers</h3> <div class="paragraph"> <p>Each maintained release branch has a release manager, who volunteers to coordinate new features and bug fixes are backported to that release. The release managers are <a href="https://hbase.apache.org/team-list.html">committers</a>. If you would like your feature or bug fix to be included in a given release, communicate with that release manager. If this list goes out of date or you can’t reach the listed person, reach out to someone else on the list.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> End-of-life releases are not included in this list. </td> </tr> </table> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 10. Release Managers</caption> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Release</th> <th class="tableblock halign-left valign-top">Release Manager</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">1.1</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Nick Dimiduk</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">1.2</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Sean Busbey</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">1.3</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Mikhail Antonov</p></td> </tr> </tbody> </table> </div> <div class="sect2"> <h3 id="_policy_for_fix_version_in_jira"><a class="anchor" href="#_policy_for_fix_version_in_jira"></a>144.3. Policy for Fix Version in JIRA</h3> <div class="paragraph"> <p>To determine if a given fix is in a given release purely from the release numbers following rules are defined:</p> </div> <div class="paragraph"> <p>Fix version of X.Y.Z ⇒ fixed in all releases X.Y.Z' (where Z' = Z).<br> Fix version of X.Y.0 ⇒ fixed in all releases X.Y'.* (where Y' = Y).<br> Fix version of X.0.0 ⇒ fixed in all releases X'.*.* (where X' = X).<br></p> </div> <div class="paragraph"> <p>By this policy, fix version of 1.3.0 implies 1.4.0, but 1.3.2 does not imply 1.4.0 as we could not tell purely from the numbers which release came first.</p> </div> </div> <div class="sect2"> <h3 id="code.standards"><a class="anchor" href="#code.standards"></a>144.4. Code Standards</h3> <div class="sect3"> <h4 id="_interface_classifications"><a class="anchor" href="#_interface_classifications"></a>144.4.1. Interface Classifications</h4> <div class="paragraph"> <p>Interfaces are classified both by audience and by stability level. These labels appear at the head of a class. The conventions followed by HBase are inherited by its parent project, Hadoop.</p> </div> <div class="paragraph"> <p>The following interface classifications are commonly used:</p> </div> <div class="dlist"> <div class="title">InterfaceAudience</div> <dl> <dt class="hdlist1"><code>@InterfaceAudience.Public</code></dt> <dd> <p>APIs for users and HBase applications. These APIs will be deprecated through major versions of HBase.</p> </dd> <dt class="hdlist1"><code>@InterfaceAudience.Private</code></dt> <dd> <p>APIs for HBase internals developers. No guarantees on compatibility or availability in future versions. Private interfaces do not need an <code>@InterfaceStability</code> classification.</p> </dd> <dt class="hdlist1"><code>@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.COPROC)</code></dt> <dd> <p>APIs for HBase coprocessor writers.</p> </dd> <dt class="hdlist1">No <code>@InterfaceAudience</code> Classification</dt> <dd> <p>Packages without an <code>@InterfaceAudience</code> label are considered private. Mark your new packages if publicly accessible.</p> </dd> </dl> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Excluding Non-Public Interfaces from API Documentation</div> <div class="paragraph"> <p>Only interfaces classified <code>@InterfaceAudience.Public</code> should be included in API documentation (Javadoc). Committers must add new package excludes <code>ExcludePackageNames</code> section of the <em>pom.xml</em> for new packages which do not contain public classes.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <div class="title">@InterfaceStability</div> <p><code>@InterfaceStability</code> is important for packages marked <code>@InterfaceAudience.Public</code>.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1"><code>@InterfaceStability.Stable</code></dt> <dd> <p>Public packages marked as stable cannot be changed without a deprecation path or a very good reason.</p> </dd> <dt class="hdlist1"><code>@InterfaceStability.Unstable</code></dt> <dd> <p>Public packages marked as unstable can be changed without a deprecation path.</p> </dd> <dt class="hdlist1"><code>@InterfaceStability.Evolving</code></dt> <dd> <p>Public packages marked as evolving may be changed, but it is discouraged.</p> </dd> <dt class="hdlist1">No <code>@InterfaceStability</code> Label</dt> <dd> <p>Public classes with no <code>@InterfaceStability</code> label are discouraged, and should be considered implicitly unstable.</p> </dd> </dl> </div> <div class="paragraph"> <p>If you are unclear about how to mark packages, ask on the development list.</p> </div> </div> <div class="sect3"> <h4 id="common.patch.feedback"><a class="anchor" href="#common.patch.feedback"></a>144.4.2. Code Formatting Conventions</h4> <div class="paragraph"> <p>Please adhere to the following guidelines so that your patches can be reviewed more quickly. These guidelines have been developed based upon common feedback on patches from new contributors.</p> </div> <div class="paragraph"> <p>See the <a href="http://www.oracle.com/technetwork/java/index-135089.html">Code Conventions for the Java Programming Language</a> for more information on coding conventions in Java. See <a href="#eclipse.code.formatting">eclipse.code.formatting</a> to setup Eclipse to check for some of these guidelines automatically.</p> </div> <div class="sect4"> <h5 id="common.patch.feedback.space.invaders"><a class="anchor" href="#common.patch.feedback.space.invaders"></a>Space Invaders</h5> <div class="paragraph"> <p>Do not use extra spaces around brackets. Use the second style, rather than the first.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="keyword">if</span> ( foo.equals( bar ) ) { <span class="comment">// don't do this</span></code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="keyword">if</span> (foo.equals(bar)) {</code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">foo = barArray[ i ]; <span class="comment">// don't do this</span></code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">foo = barArray[i];</code></pre> </div> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.autogen"><a class="anchor" href="#common.patch.feedback.autogen"></a>Auto Generated Code</h5> <div class="paragraph"> <p>Auto-generated code in Eclipse often uses bad variable names such as <code>arg0</code>. Use more informative variable names. Use code like the second example here.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> <span class="directive">public</span> <span class="type">void</span> readFields(<span class="predefined-type">DataInput</span> arg0) <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="comment">// don't do this</span> foo = arg0.readUTF(); <span class="comment">// don't do this</span></code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> <span class="directive">public</span> <span class="type">void</span> readFields(<span class="predefined-type">DataInput</span> di) <span class="directive">throws</span> <span class="exception">IOException</span> { foo = di.readUTF();</code></pre> </div> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.longlines"><a class="anchor" href="#common.patch.feedback.longlines"></a>Long Lines</h5> <div class="paragraph"> <p>Keep lines less than 100 characters. You can configure your IDE to do this automatically.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Bar bar = foo.veryLongMethodWithManyArguments(argument1, argument2, argument3, argument4, argument5, argument6, argument7, argument8, argument9); <span class="comment">// don't do this</span></code></pre> </div> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Bar bar = foo.veryLongMethodWithManyArguments( argument1, argument2, argument3,argument4, argument5, argument6, argument7, argument8, argument9);</code></pre> </div> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.trailingspaces"><a class="anchor" href="#common.patch.feedback.trailingspaces"></a>Trailing Spaces</h5> <div class="paragraph"> <p>Be sure there is a line break after the end of your code, and avoid lines with nothing but whitespace. This makes diffs more meaningful. You can configure your IDE to help with this.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Bar bar = foo.getBar(); <--- imagine there is an extra space(s) after the semicolon.</code></pre> </div> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.javadoc"><a class="anchor" href="#common.patch.feedback.javadoc"></a>API Documentation (Javadoc)</h5> <div class="paragraph"> <p>Don’t forget Javadoc!</p> </div> <div class="paragraph"> <p>Javadoc warnings are checked during precommit. If the precommit tool gives you a '-1', please fix the javadoc issue. Your patch won’t be committed if it adds such warnings.</p> </div> <div class="paragraph"> <p>Also, no <code>@author</code> tags - that’s a rule.</p> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.findbugs"><a class="anchor" href="#common.patch.feedback.findbugs"></a>Findbugs</h5> <div class="paragraph"> <p><code>Findbugs</code> is used to detect common bugs pattern. It is checked during the precommit build. If errors are found, please fix them. You can run findbugs locally with <code>mvn findbugs:findbugs</code>, which will generate the <code>findbugs</code> files locally. Sometimes, you may have to write code smarter than <code>findbugs</code>. You can annotate your code to tell <code>findbugs</code> you know what you’re doing, by annotating your class with the following annotation:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="annotation">@edu</span>.umd.cs.findbugs.annotations.SuppressWarnings( value=<span class="string"><span class="delimiter">"</span><span class="content">HE_EQUALS_USE_HASHCODE</span><span class="delimiter">"</span></span>, justification=<span class="string"><span class="delimiter">"</span><span class="content">I know what I'm doing</span><span class="delimiter">"</span></span>)</code></pre> </div> </div> <div class="paragraph"> <p>It is important to use the Apache-licensed version of the annotations. That generally means using annotations in the <code>edu.umd.cs.findbugs.annotations</code> package so that we can rely on the cleanroom reimplementation rather than annotations in the <code>javax.annotations</code> package.</p> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.javadoc.defaults"><a class="anchor" href="#common.patch.feedback.javadoc.defaults"></a>Javadoc - Useless Defaults</h5> <div class="paragraph"> <p>Don’t just leave javadoc tags the way IDE generates them, or fill redundant information in them.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> <span class="comment">/** * @param table <---- don't leave them empty! * @param region An HRegion object. <---- don't fill redundant information! * @return Foo Object foo just created. <---- Not useful information * @throws SomeException <---- Not useful. Function declarations already tell that! * @throws BarException when something went wrong <---- really? */</span> <span class="directive">public</span> Foo createFoo(Bar bar);</code></pre> </div> </div> <div class="paragraph"> <p>Either add something descriptive to the tags, or just remove them. The preference is to add something descriptive and useful.</p> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.onething"><a class="anchor" href="#common.patch.feedback.onething"></a>One Thing At A Time, Folks</h5> <div class="paragraph"> <p>If you submit a patch for one thing, don’t do auto-reformatting or unrelated reformatting of code on a completely different area of code.</p> </div> <div class="paragraph"> <p>Likewise, don’t add unrelated cleanup or refactorings outside the scope of your Jira.</p> </div> </div> <div class="sect4"> <h5 id="common.patch.feedback.tests"><a class="anchor" href="#common.patch.feedback.tests"></a>Ambigious Unit Tests</h5> <div class="paragraph"> <p>Make sure that you’re clear about what you are testing in your unit tests and why.</p> </div> </div> </div> <div class="sect3"> <h4 id="_garbage_collection_conserving_guidelines"><a class="anchor" href="#_garbage_collection_conserving_guidelines"></a>144.4.3. Garbage-Collection Conserving Guidelines</h4> <div class="paragraph"> <p>The following guidelines were borrowed from <a href="http://engineering.linkedin.com/performance/linkedin-feed-faster-less-jvm-garbage" class="bare">http://engineering.linkedin.com/performance/linkedin-feed-faster-less-jvm-garbage</a>. Keep them in mind to keep preventable garbage collection to a minimum. Have a look at the blog post for some great examples of how to refactor your code according to these guidelines.</p> </div> <div class="ulist"> <ul> <li> <p>Be careful with Iterators</p> </li> <li> <p>Estimate the size of a collection when initializing</p> </li> <li> <p>Defer expression evaluation</p> </li> <li> <p>Compile the regex patterns in advance</p> </li> <li> <p>Cache it if you can</p> </li> <li> <p>String Interns are useful but dangerous</p> </li> </ul> </div> </div> </div> <div class="sect2"> <h3 id="design.invariants"><a class="anchor" href="#design.invariants"></a>144.5. Invariants</h3> <div class="paragraph"> <p>We don’t have many but what we have we list below. All are subject to challenge of course but until then, please hold to the rules of the road.</p> </div> <div class="sect3"> <h4 id="design.invariants.zk.data"><a class="anchor" href="#design.invariants.zk.data"></a>144.5.1. No permanent state in ZooKeeper</h4> <div class="paragraph"> <p>ZooKeeper state should transient (treat it like memory). If ZooKeeper state is deleted, hbase should be able to recover and essentially be in the same state.</p> </div> <div class="ulist"> <ul> <li> <p>.Exceptions: There are currently a few exceptions that we need to fix around whether a table is enabled or disabled.</p> </li> <li> <p>Replication data is currently stored only in ZooKeeper. Deleting ZooKeeper data related to replication may cause replication to be disabled. Do not delete the replication tree, <em>/hbase/replication/</em>.</p> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> Replication may be disrupted and data loss may occur if you delete the replication tree (<em>/hbase/replication/</em>) from ZooKeeper. Follow progress on this issue at <a href="https://issues.apache.org/jira/browse/HBASE-10295">HBASE-10295</a>. </td> </tr> </table> </div> </li> </ul> </div> </div> </div> <div class="sect2"> <h3 id="run.insitu"><a class="anchor" href="#run.insitu"></a>144.6. Running In-Situ</h3> <div class="paragraph"> <p>If you are developing Apache HBase, frequently it is useful to test your changes against a more-real cluster than what you find in unit tests. In this case, HBase can be run directly from the source in local-mode. All you need to do is run:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">${HBASE_HOME}/bin/start-hbase.sh</code></pre> </div> </div> <div class="paragraph"> <p>This will spin up a full local-cluster, just as if you had packaged up HBase and installed it on your machine.</p> </div> <div class="paragraph"> <p>Keep in mind that you will need to have installed HBase into your local maven repository for the in-situ cluster to work properly. That is, you will need to run:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">mvn clean install -DskipTests</code></pre> </div> </div> <div class="paragraph"> <p>to ensure that maven can find the correct classpath and dependencies. Generally, the above command is just a good thing to try running first, if maven is acting oddly.</p> </div> </div> <div class="sect2"> <h3 id="add.metrics"><a class="anchor" href="#add.metrics"></a>144.7. Adding Metrics</h3> <div class="paragraph"> <p>After adding a new feature a developer might want to add metrics. HBase exposes metrics using the Hadoop Metrics 2 system, so adding a new metric involves exposing that metric to the hadoop system. Unfortunately the API of metrics2 changed from hadoop 1 to hadoop 2. In order to get around this a set of interfaces and implementations have to be loaded at runtime. To get an in-depth look at the reasoning and structure of these classes you can read the blog post located <a href="https://blogs.apache.org/hbase/entry/migration_to_the_new_metrics">here</a>. To add a metric to an existing MBean follow the short guide below:</p> </div> <div class="sect3"> <h4 id="_add_metric_name_and_function_to_hadoop_compat_interface"><a class="anchor" href="#_add_metric_name_and_function_to_hadoop_compat_interface"></a>144.7.1. Add Metric name and Function to Hadoop Compat Interface.</h4> <div class="paragraph"> <p>Inside of the source interface the corresponds to where the metrics are generated (eg MetricsMasterSource for things coming from HMaster) create new static strings for metric name and description. Then add a new method that will be called to add new reading.</p> </div> </div> <div class="sect3"> <h4 id="_add_the_implementation_to_both_hadoop_1_and_hadoop_2_compat_modules"><a class="anchor" href="#_add_the_implementation_to_both_hadoop_1_and_hadoop_2_compat_modules"></a>144.7.2. Add the Implementation to Both Hadoop 1 and Hadoop 2 Compat modules.</h4> <div class="paragraph"> <p>Inside of the implementation of the source (eg. MetricsMasterSourceImpl in the above example) create a new histogram, counter, gauge, or stat in the init method. Then in the method that was added to the interface wire up the parameter passed in to the histogram.</p> </div> <div class="paragraph"> <p>Now add tests that make sure the data is correctly exported to the metrics 2 system. For this the MetricsAssertHelper is provided.</p> </div> </div> </div> <div class="sect2"> <h3 id="git.best.practices"><a class="anchor" href="#git.best.practices"></a>144.8. Git Best Practices</h3> <div class="dlist"> <dl> <dt class="hdlist1">Avoid git merges.</dt> <dd> <p>Use <code>git pull --rebase</code> or <code>git fetch</code> followed by <code>git rebase</code>.</p> </dd> <dt class="hdlist1">Do not use <code>git push --force</code>.</dt> <dd> <p>If the push does not work, fix the problem or ask for help.</p> </dd> </dl> </div> <div class="paragraph"> <p>Please contribute to this document if you think of other Git best practices.</p> </div> <div class="sect3"> <h4 id="__code_rebase_all_git_branches_sh_code"><a class="anchor" href="#__code_rebase_all_git_branches_sh_code"></a>144.8.1. <code>rebase_all_git_branches.sh</code></h4> <div class="paragraph"> <p>The <em>dev-support/rebase_all_git_branches.sh</em> script is provided to help keep your Git repository clean. Use the <code>-h</code> parameter to get usage instructions. The script automatically refreshes your tracking branches, attempts an automatic rebase of each local branch against its remote branch, and gives you the option to delete any branch which represents a closed <code>HBASE-</code> JIRA. The script has one optional configuration option, the location of your Git directory. You can set a default by editing the script. Otherwise, you can pass the git directory manually by using the <code>-d</code> parameter, followed by an absolute or relative directory name, or even '.' for the current working directory. The script checks the directory for sub-directory called <em>.git/</em>, before proceeding.</p> </div> </div> </div> <div class="sect2"> <h3 id="submitting.patches"><a class="anchor" href="#submitting.patches"></a>144.9. Submitting Patches</h3> <div class="paragraph"> <p>If you are new to submitting patches to open source or new to submitting patches to Apache, start by reading the <a href="https://commons.apache.org/patches.html">On Contributing Patches</a> page from <a href="https://commons.apache.org/">Apache Commons Project</a>. It provides a nice overview that applies equally to the Apache HBase Project. <a href="https://accumulo.apache.org/git.html">Accumulo doc on how to contribute and develop</a> is also good read to understand development workflow.</p> </div> <div class="sect3"> <h4 id="submitting.patches.create"><a class="anchor" href="#submitting.patches.create"></a>144.9.1. Create Patch</h4> <div class="paragraph"> <p>Make sure you review <a href="#common.patch.feedback">common.patch.feedback</a> for code style. If your patch was generated incorrectly or your code does not adhere to the code formatting guidelines, you may be asked to redo some work.</p> </div> <div class="listingblock"> <div class="title">Using submit-patch.py (recommended)</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ dev-support/submit-patch.py -jid HBASE-xxxxx</code></pre> </div> </div> <div class="paragraph"> <p>Use this script to create patches, upload to jira and optionally create/update reviews on Review Board. Patch name is automatically formatted as <em>(JIRA).(branch name).(patch number).patch</em> to follow Yetus' naming rules. Use <code>-h</code> flag to know detailed usage information. Most useful options are:</p> </div> <div class="ulist"> <ul> <li> <p><code>-b BRANCH, --branch BRANCH</code> : Specify base branch for generating the diff. If not specified, tracking branch is used. If there is no tracking branch, error will be thrown.</p> </li> <li> <p><code>-jid JIRA_ID, --jira-id JIRA_ID</code> : If used, deduces next patch version from attachments in the jira and uploads the new patch. Script will ask for jira username/password for authentication. If not set, patch is named <branch>.patch.</p> </li> </ul> </div> <div class="paragraph"> <p>By default, it’ll also create/update review board. To skip that action, use <code>-srb</code> option. It uses 'Issue Links' in the jira to figure out if a review request already exists. If no review request is present, then creates a new one and populates all required fields using jira summary, patch description, etc. Also adds this review’s link to the jira.</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Save authentication credentials (optional)</dt> <dd> <p>Since attaching patches on JIRA and creating/changing review request on ReviewBoard requires valid user authentication, the script will prompt you for username and password. To avoid the hassle every time, set up <code>~/.apache-creds</code> with login details and encrypt it by following the steps in footer of script’s help message.</p> </dd> <dt class="hdlist1">Python dependencies</dt> <dd> <p>To install required python dependencies, execute <code>pip install -r dev-support/python-requirements.txt</code> from the master branch.</p> </dd> </dl> </div> <div class="olist arabic"> <div class="title">Manually</div> <ol class="arabic"> <li> <p>Use <code>git rebase -i</code> first, to combine (squash) smaller commits into a single larger one.</p> </li> <li> <p>Create patch using IDE or Git commands. <code>git format-patch</code> is preferred since it preserves patch author’s name and commit message. Also, it handles binary files by default, whereas <code>git diff</code> ignores them unless you use the <code>--binary</code> option.</p> </li> <li> <p>Patch name should be as follows to adhere to Yetus' naming convention:<br> <code>(JIRA).(branch name).(patch number).patch</code><br> For eg. HBASE-11625.master.001.patch, HBASE-XXXXX.branch-1.2.0005.patch, etc.</p> </li> <li> <p>Attach the patch to the JIRA using <code>More→Attach Files</code> then click on <b class="button">Submit Patch</b> button, which’ll trigger Hudson job to check patch for validity.</p> </li> <li> <p>If your patch is longer than a single screen, also create a review on Review Board and add the link to JIRA. See <a href="#reviewboard">reviewboard</a>.</p> </li> </ol> </div> <div class="ulist"> <div class="title">Few general guidelines</div> <ul> <li> <p>Always patch against the master branch first, even if you want to patch in another branch. HBase committers always apply patches first to the master branch, and backport if necessary.</p> </li> <li> <p>Submit one single patch for a fix. If necessary, squash local commits to merge local commits into a single one first. See this <a href="http://stackoverflow.com/questions/5308816/how-to-use-git-merge-squash">Stack Overflow question</a> for more information about squashing commits.</p> </li> <li> <p>Please understand that not every patch may get committed, and that feedback will likely be provided on the patch.</p> </li> <li> <p>If you need to revise your patch, leave the previous patch file(s) attached to the JIRA, and upload a new one with incremented patch number.<br> Click on <b class="button">Cancel Patch</b> and then on <b class="button">Submit Patch</b> to trigger the presubmit run.</p> </li> </ul> </div> </div> <div class="sect3"> <h4 id="submitting.patches.tests"><a class="anchor" href="#submitting.patches.tests"></a>144.9.2. Unit Tests</h4> <div class="paragraph"> <p>Always add and/or update relevant unit tests when making the changes. Make sure that new/changed unit tests pass locally before submitting the patch because it is faster than waiting for presubmit result which runs full test suite. This will save your own time and effort. Use <a href="#mockito">mockito</a> to make mocks which are very useful for testing failure scenarios by injecting appropriate failures.</p> </div> <div class="paragraph"> <p>If you are creating a new unit test class, notice how other unit test classes have classification/sizing annotations before class name and a static methods for setup/teardown of testing environment. Be sure to include annotations in any new unit test files. See <a href="#hbase.tests">hbase.tests</a> for more information on tests.</p> </div> </div> <div class="sect3"> <h4 id="_integration_tests"><a class="anchor" href="#_integration_tests"></a>144.9.3. Integration Tests</h4> <div class="paragraph"> <p>Significant new features should provide an integration test in addition to unit tests, suitable for exercising the new feature at different points in its configuration space.</p> </div> </div> <div class="sect3"> <h4 id="reviewboard"><a class="anchor" href="#reviewboard"></a>144.9.4. ReviewBoard</h4> <div class="paragraph"> <p>Patches larger than one screen, or patches that will be tricky to review, should go through <a href="https://reviews.apache.org">ReviewBoard</a>.</p> </div> <div class="olist arabic"> <div class="title">Procedure: Use ReviewBoard</div> <ol class="arabic"> <li> <p>Register for an account if you don’t already have one. It does not use the credentials from <a href="https://issues.apache.org">issues.apache.org</a>. Log in.</p> </li> <li> <p>Click <span class="label">New Review Request</span>.</p> </li> <li> <p>Choose the <code>hbase-git</code> repository. Click Choose File to select the diff and optionally a parent diff. Click <b class="button">Create Review Request</b>.</p> </li> <li> <p>Fill in the fields as required. At the minimum, fill in the <span class="label">Summary</span> and choose <code>hbase</code> as the <span class="label">Review Group</span>. If you fill in the <span class="label">Bugs</span> field, the review board links back to the relevant JIRA. The more fields you fill in, the better. Click <b class="button">Publish</b> to make your review request public. An email will be sent to everyone in the <code>hbase</code> group, to review the patch.</p> </li> <li> <p>Back in your JIRA, click , and paste in the URL of your ReviewBoard request. This attaches the ReviewBoard to the JIRA, for easy access.</p> </li> <li> <p>To cancel the request, click .</p> </li> </ol> </div> <div class="paragraph"> <p>For more information on how to use ReviewBoard, see <a href="http://www.reviewboard.org/docs/manual/1.5/">the ReviewBoard documentation</a>.</p> </div> </div> <div class="sect3"> <h4 id="_guide_for_hbase_committers"><a class="anchor" href="#_guide_for_hbase_committers"></a>144.9.5. Guide for HBase Committers</h4> <div class="sect4"> <h5 id="_new_committers"><a class="anchor" href="#_new_committers"></a>New committers</h5> <div class="paragraph"> <p>New committers are encouraged to first read Apache’s generic committer documentation:</p> </div> <div class="ulist"> <ul> <li> <p><a href="https://www.apache.org/dev/new-committers-guide.html">Apache New Committer Guide</a></p> </li> <li> <p><a href="https://www.apache.org/dev/committers.html">Apache Committer FAQ</a></p> </li> </ul> </div> </div> <div class="sect4"> <h5 id="_review"><a class="anchor" href="#_review"></a>Review</h5> <div class="paragraph"> <p>HBase committers should, as often as possible, attempt to review patches submitted by others. Ideally every submitted patch will get reviewed by a committer <em>within a few days</em>. If a committer reviews a patch they have not authored, and believe it to be of sufficient quality, then they can commit the patch, otherwise the patch should be cancelled with a clear explanation for why it was rejected.</p> </div> <div class="paragraph"> <p>The list of submitted patches is in the <a href="https://issues.apache.org/jira/secure/IssueNavigator.jspa?mode=hide&requestId=12312392">HBase Review Queue</a>, which is ordered by time of last modification. Committers should scan the list from top to bottom, looking for patches that they feel qualified to review and possibly commit.</p> </div> <div class="paragraph"> <p>For non-trivial changes, it is required to get another committer to review your own patches before commit. Use the <b class="button">Submit Patch</b> button in JIRA, just like other contributors, and then wait for a <code>+1</code> response from another committer before committing.</p> </div> </div> <div class="sect4"> <h5 id="_reject"><a class="anchor" href="#_reject"></a>Reject</h5> <div class="paragraph"> <p>Patches which do not adhere to the guidelines in <a href="https://hbase.apache.org/book.html#developer">HowToContribute</a> and to the <a href="https://wiki.apache.org/hadoop/CodeReviewChecklist">code review checklist</a> should be rejected. Committers should always be polite to contributors and try to instruct and encourage them to contribute better patches. If a committer wishes to improve an unacceptable patch, then it should first be rejected, and a new patch should be attached by the committer for review.</p> </div> </div> <div class="sect4"> <h5 id="committing.patches"><a class="anchor" href="#committing.patches"></a>Commit</h5> <div class="paragraph"> <p>Committers commit patches to the Apache HBase GIT repository.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">Before you commit!!!!</div> <div class="paragraph"> <p>Make sure your local configuration is correct, especially your identity and email. Examine the output of the $ git config --list command and be sure it is correct. See this GitHub article, <a href="https://help.github.com/articles/set-up-git">Set Up Git</a> if you need pointers.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>When you commit a patch, please:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Include the Jira issue id in the commit message along with a short description of the change. Try to add something more than just the Jira title so that someone looking at git log doesn’t have to go to Jira to discern what the change is about. Be sure to get the issue ID right, as this causes Jira to link to the change in Git (use the issue’s "All" tab to see these).</p> </li> <li> <p>Commit the patch to a new branch based off master or other intended branch. It’s a good idea to call this branch by the JIRA ID. Then check out the relevant target branch where you want to commit, make sure your local branch has all remote changes, by doing a git pull --rebase or another similar command, cherry-pick the change into each relevant branch (such as master), and do git push <remote-server> <remote-branch>.</p> <div class="admonitionblock warning"> <table> <tr> <td class="icon"> <i class="fa icon-warning" title="Warning"></i> </td> <td class="content"> If you do not have all remote changes, the push will fail. If the push fails for any reason, fix the problem or ask for help. Do not do a git push --force. </td> </tr> </table> </div> <div class="paragraph"> <p>Before you can commit a patch, you need to determine how the patch was created. The instructions and preferences around the way to create patches have changed, and there will be a transition period.</p> </div> <div class="ulist"> <div class="title">Determine How a Patch Was Created</div> <ul> <li> <p>If the first few lines of the patch look like the headers of an email, with a From, Date, and Subject, it was created using git format-patch. This is the preferred way, because you can reuse the submitter’s commit message. If the commit message is not appropriate, you can still use the commit, then run <code>git commit --amend</code> and reword as appropriate.</p> </li> <li> <p>If the first line of the patch looks similar to the following, it was created using git diff without <code>--no-prefix</code>. This is acceptable too. Notice the <code>a</code> and <code>b</code> in front of the file names. This is the indication that the patch was not created with <code>--no-prefix</code>.</p> <div class="listingblock"> <div class="content"> <pre>diff --git a/src/main/asciidoc/_chapters/developer.adoc b/src/main/asciidoc/_chapters/developer.adoc</pre> </div> </div> </li> <li> <p>If the first line of the patch looks similar to the following (without the <code>a</code> and <code>b</code>), the patch was created with git diff --no-prefix and you need to add <code>-p0</code> to the git apply command below.</p> <div class="listingblock"> <div class="content"> <pre>diff --git src/main/asciidoc/_chapters/developer.adoc src/main/asciidoc/_chapters/developer.adoc</pre> </div> </div> </li> </ul> </div> <div class="exampleblock"> <div class="title">Example 51. Example of committing a Patch</div> <div class="content"> <div class="paragraph"> <p>One thing you will notice with these examples is that there are a lot of git pull commands. The only command that actually writes anything to the remote repository is git push, and you need to make absolutely sure you have the correct versions of everything and don’t have any conflicts before pushing. The extra git pull commands are usually redundant, but better safe than sorry.</p> </div> <div class="paragraph"> <p>The first example shows how to apply a patch that was generated with git format-patch and apply it to the <code>master</code> and <code>branch-1</code> branches.</p> </div> <div class="paragraph"> <p>The directive to use git format-patch rather than git diff, and not to use <code>--no-prefix</code>, is a new one. See the second example for how to apply a patch created with git diff, and educate the person who created the patch.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ git checkout -b HBASE-XXXX $ git am ~/Downloads/HBASE-XXXX-v2.patch --signoff # If you are committing someone else's patch. $ git checkout master $ git pull --rebase $ git cherry-pick <sha-from-commit> # Resolve conflicts if necessary or ask the submitter to do it $ git pull --rebase # Better safe than sorry $ git push origin master # Backport to branch-1 $ git checkout branch-1 $ git pull --rebase $ git cherry-pick <sha-from-commit> # Resolve conflicts if necessary $ git pull --rebase # Better safe than sorry $ git push origin branch-1 $ git branch -D HBASE-XXXX</pre> </div> </div> <div class="paragraph"> <p>This example shows how to commit a patch that was created using git diff without <code>--no-prefix</code>. If the patch was created with <code>--no-prefix</code>, add <code>-p0</code> to the git apply command.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ git apply ~/Downloads/HBASE-XXXX-v2.patch $ git commit -m "HBASE-XXXX Really Good Code Fix (Joe Schmo)" --author=<contributor> -a # This and next command is needed for patches created with 'git diff' $ git commit --amend --signoff $ git checkout master $ git pull --rebase $ git cherry-pick <sha-from-commit> # Resolve conflicts if necessary or ask the submitter to do it $ git pull --rebase # Better safe than sorry $ git push origin master # Backport to branch-1 $ git checkout branch-1 $ git pull --rebase $ git cherry-pick <sha-from-commit> # Resolve conflicts if necessary or ask the submitter to do it $ git pull --rebase # Better safe than sorry $ git push origin branch-1 $ git branch -D HBASE-XXXX</pre> </div> </div> </div> </div> </li> <li> <p>Resolve the issue as fixed, thanking the contributor. Always set the "Fix Version" at this point, but please only set a single fix version for each branch where the change was committed, the earliest release in that branch in which the change will appear.</p> </li> </ol> </div> <div class="sect5"> <h6 id="_commit_message_format"><a class="anchor" href="#_commit_message_format"></a>Commit Message Format</h6> <div class="paragraph"> <p>The commit message should contain the JIRA ID and a description of what the patch does. The preferred commit message format is:</p> </div> <div class="listingblock"> <div class="content"> <pre><jira-id> <jira-title> (<contributor-name-if-not-commit-author>)</pre> </div> </div> <div class="listingblock"> <div class="content"> <pre>HBASE-12345 Fix All The Things (jane@example.com)</pre> </div> </div> <div class="paragraph"> <p>If the contributor used git format-patch to generate the patch, their commit message is in their patch and you can use that, but be sure the JIRA ID is at the front of the commit message, even if the contributor left it out.</p> </div> </div> <div class="sect5"> <h6 id="committer.amending.author"><a class="anchor" href="#committer.amending.author"></a>Add Amending-Author when a conflict cherrypick backporting</h6> <div class="paragraph"> <p>We’ve established the practice of committing to master and then cherry picking back to branches whenever possible, unless</p> </div> <div class="ulist"> <ul> <li> <p>it’s breaking compat: In which case, if it can go in minor releases, backport to branch-1 and branch-2.</p> </li> <li> <p>it’s a new feature: No for maintenance releases, For minor releases, discuss and arrive at consensus.</p> </li> </ul> </div> <div class="paragraph"> <p>When there is a minor conflict we can fix it up and just proceed with the commit. The resulting commit retains the original author. When the amending author is different from the original committer, add notice of this at the end of the commit message as: <code>Amending-Author: Author <committer&apache></code> See discussion at <a href="http://search-hadoop.com/m/DHED4wHGYS">HBase, mail # dev - [DISCUSSION</a> Best practice when amending commits cherry picked from master to branch].</p> </div> </div> <div class="sect5"> <h6 id="committer.tests"><a class="anchor" href="#committer.tests"></a>Committers are responsible for making sure commits do not break the build or tests</h6> <div class="paragraph"> <p>If a committer commits a patch, it is their responsibility to make sure it passes the test suite. It is helpful if contributors keep an eye out that their patch does not break the hbase build and/or tests, but ultimately, a contributor cannot be expected to be aware of all the particular vagaries and interconnections that occur in a project like HBase. A committer should.</p> </div> </div> <div class="sect5"> <h6 id="git.patch.flow"><a class="anchor" href="#git.patch.flow"></a>Patching Etiquette</h6> <div class="paragraph"> <p>In the thread <a href="http://search-hadoop.com/m/DHED4EiwOz">HBase, mail # dev - ANNOUNCEMENT: Git Migration In Progress (WAS ⇒ Re: Git Migration)</a>, it was agreed on the following patch flow</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Develop and commit the patch against master first.</p> </li> <li> <p>Try to cherry-pick the patch when backporting if possible.</p> </li> <li> <p>If this does not work, manually commit the patch to the branch.</p> </li> </ol> </div> </div> <div class="sect5"> <h6 id="_merge_commits"><a class="anchor" href="#_merge_commits"></a>Merge Commits</h6> <div class="paragraph"> <p>Avoid merge commits, as they create problems in the git history.</p> </div> </div> <div class="sect5"> <h6 id="_committing_documentation"><a class="anchor" href="#_committing_documentation"></a>Committing Documentation</h6> <div class="paragraph"> <p>See <a href="#appendix_contributing_to_documentation">appendix contributing to documentation</a>.</p> </div> </div> </div> </div> <div class="sect3"> <h4 id="_dialog"><a class="anchor" href="#_dialog"></a>144.9.6. Dialog</h4> <div class="paragraph"> <p>Committers should hang out in the #hbase room on irc.freenode.net for real-time discussions. However any substantive discussion (as with any off-list project-related discussion) should be re-iterated in Jira or on the developer list.</p> </div> </div> <div class="sect3"> <h4 id="_do_not_edit_jira_comments"><a class="anchor" href="#_do_not_edit_jira_comments"></a>144.9.7. Do not edit JIRA comments</h4> <div class="paragraph"> <p>Misspellings and/or bad grammar is preferable to the disruption a JIRA comment edit causes: See the discussion at <a href="http://search-hadoop.com/?q=%5BReopened%5D+%28HBASE-451%29+Remove+HTableDescriptor+from+HRegionInfo&fc_project=HBase">Re:(HBASE-451) Remove HTableDescriptor from HRegionInfo</a></p> </div> </div> </div> <div class="sect2"> <h3 id="hbase.archetypes.development"><a class="anchor" href="#hbase.archetypes.development"></a>144.10. Development of HBase-related Maven archetypes</h3> <div class="paragraph"> <p>The development of HBase-related Maven archetypes was begun with <a href="https://issues.apache.org/jira/browse/HBASE-14876">HBASE-14876</a>. For an overview of the hbase-archetypes infrastructure and instructions for developing new HBase-related Maven archetypes, please see <code>hbase/hbase-archetypes/README.md</code>.</p> </div> </div> </div> </div> <h1 id="unit.tests" class="sect0"><a class="anchor" href="#unit.tests"></a>Unit Testing HBase Applications</h1> <div class="openblock partintro"> <div class="content"> This chapter discusses unit testing your HBase application using JUnit, Mockito, MRUnit, and HBaseTestingUtility. Much of the information comes from <a href="http://blog.cloudera.com/blog/2013/09/how-to-test-hbase-applications-using-popular-tools/">a community blog post about testing HBase applications</a>. For information on unit tests for HBase itself, see <a href="#hbase.tests">hbase.tests</a>. </div> </div> <div class="sect1"> <h2 id="_junit"><a class="anchor" href="#_junit"></a>145. JUnit</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase uses <a href="http://junit.org">JUnit</a> 4 for unit tests</p> </div> <div class="paragraph"> <p>This example will add unit tests to the following example class:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">class</span> <span class="class">MyHBaseDAO</span> { <span class="directive">public</span> <span class="directive">static</span> <span class="type">void</span> insertRecord(Table.getTable(table), HBaseTestObj obj) <span class="directive">throws</span> <span class="exception">Exception</span> { Put put = createPut(obj); table.put(put); } <span class="directive">private</span> <span class="directive">static</span> Put createPut(HBaseTestObj obj) { Put put = <span class="keyword">new</span> Put(Bytes.toBytes(obj.getRowKey())); put.add(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>), Bytes.toBytes(obj.getData1())); put.add(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-2</span><span class="delimiter">"</span></span>), Bytes.toBytes(obj.getData2())); <span class="keyword">return</span> put; } }</code></pre> </div> </div> <div class="paragraph"> <p>The first step is to add JUnit dependencies to your Maven POM file:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>junit<span class="tag"></groupId></span> <span class="tag"><artifactId></span>junit<span class="tag"></artifactId></span> <span class="tag"><version></span>4.11<span class="tag"></version></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span></code></pre> </div> </div> <div class="paragraph"> <p>Next, add some unit tests to your code. Tests are annotated with <code>@Test</code>. Here, the unit tests are in bold.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">class</span> <span class="class">TestMyHbaseDAOData</span> { <span class="annotation">@Test</span> <span class="directive">public</span> <span class="type">void</span> testCreatePut() <span class="directive">throws</span> <span class="exception">Exception</span> { HBaseTestObj obj = <span class="keyword">new</span> HBaseTestObj(); obj.setRowKey(<span class="string"><span class="delimiter">"</span><span class="content">ROWKEY-1</span><span class="delimiter">"</span></span>); obj.setData1(<span class="string"><span class="delimiter">"</span><span class="content">DATA-1</span><span class="delimiter">"</span></span>); obj.setData2(<span class="string"><span class="delimiter">"</span><span class="content">DATA-2</span><span class="delimiter">"</span></span>); Put put = MyHBaseDAO.createPut(obj); assertEquals(obj.getRowKey(), Bytes.toString(put.getRow())); assertEquals(obj.getData1(), Bytes.toString(put.get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>)).get(<span class="integer">0</span>).getValue())); assertEquals(obj.getData2(), Bytes.toString(put.get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-2</span><span class="delimiter">"</span></span>)).get(<span class="integer">0</span>).getValue())); } }</code></pre> </div> </div> <div class="paragraph"> <p>These tests ensure that your <code>createPut</code> method creates, populates, and returns a <code>Put</code> object with expected values. Of course, JUnit can do much more than this. For an introduction to JUnit, see link:https://github.com/junit-team/junit/wiki/Getting-started.</p> </div> </div> </div> <div class="sect1"> <h2 id="_mockito"><a class="anchor" href="#_mockito"></a>146. Mockito</h2> <div class="sectionbody"> <div class="paragraph"> <p>Mockito is a mocking framework. It goes further than JUnit by allowing you to test the interactions between objects without having to replicate the entire environment. You can read more about Mockito at its project site, link:https://code.google.com/p/mockito/.</p> </div> <div class="paragraph"> <p>You can use Mockito to do unit testing on smaller units. For instance, you can mock a <code>org.apache.hadoop.hbase.Server</code> instance or a <code>org.apache.hadoop.hbase.master.MasterServices</code> interface reference rather than a full-blown <code>org.apache.hadoop.hbase.master.HMaster</code>.</p> </div> <div class="paragraph"> <p>This example builds upon the example code in <a href="#unit.tests">unit.tests</a>, to test the <code>insertRecord</code> method.</p> </div> <div class="paragraph"> <p>First, add a dependency for Mockito to your Maven POM file.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>org.mockito<span class="tag"></groupId></span> <span class="tag"><artifactId></span>mockito-all<span class="tag"></artifactId></span> <span class="tag"><version></span>1.9.5<span class="tag"></version></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span></code></pre> </div> </div> <div class="paragraph"> <p>Next, add a <code>@RunWith</code> annotation to your test class, to direct it to use Mockito.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="annotation">@RunWith</span>(MockitoJUnitRunner.class) <span class="directive">public</span> <span class="type">class</span> <span class="class">TestMyHBaseDAO</span>{ <span class="annotation">@Mock</span> <span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); <span class="annotation">@Mock</span> <span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(config); <span class="annotation">@Mock</span> <span class="directive">private</span> Table table; <span class="annotation">@Captor</span> <span class="directive">private</span> ArgumentCaptor putCaptor; <span class="annotation">@Test</span> <span class="directive">public</span> <span class="type">void</span> testInsertRecord() <span class="directive">throws</span> <span class="exception">Exception</span> { <span class="comment">//return mock table when getTable is called</span> when(connection.getTable(TableName.valueOf(<span class="string"><span class="delimiter">"</span><span class="content">tablename</span><span class="delimiter">"</span></span>)).thenReturn(table); <span class="comment">//create test object and make a call to the DAO that needs testing</span> HBaseTestObj obj = <span class="keyword">new</span> HBaseTestObj(); obj.setRowKey(<span class="string"><span class="delimiter">"</span><span class="content">ROWKEY-1</span><span class="delimiter">"</span></span>); obj.setData1(<span class="string"><span class="delimiter">"</span><span class="content">DATA-1</span><span class="delimiter">"</span></span>); obj.setData2(<span class="string"><span class="delimiter">"</span><span class="content">DATA-2</span><span class="delimiter">"</span></span>); MyHBaseDAO.insertRecord(table, obj); verify(table).put(putCaptor.capture()); Put put = putCaptor.getValue(); assertEquals(Bytes.toString(put.getRow()), obj.getRowKey()); <span class="keyword">assert</span>(put.has(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>))); <span class="keyword">assert</span>(put.has(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-2</span><span class="delimiter">"</span></span>))); assertEquals(Bytes.toString(put.get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>),Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>)).get(<span class="integer">0</span>).getValue()), <span class="string"><span class="delimiter">"</span><span class="content">DATA-1</span><span class="delimiter">"</span></span>); assertEquals(Bytes.toString(put.get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>),Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CQ-2</span><span class="delimiter">"</span></span>)).get(<span class="integer">0</span>).getValue()), <span class="string"><span class="delimiter">"</span><span class="content">DATA-2</span><span class="delimiter">"</span></span>); } }</code></pre> </div> </div> <div class="paragraph"> <p>This code populates <code>HBaseTestObj</code> with <code>ROWKEY-1'', </code>DATA-1'', ``DATA-2'' as values. It then inserts the record into the mocked table. The Put that the DAO would have inserted is captured, and values are tested to verify that they are what you expected them to be.</p> </div> <div class="paragraph"> <p>The key here is to manage Connection and Table instance creation outside the DAO. This allows you to mock them cleanly and test Puts as shown above. Similarly, you can now expand into other operations such as Get, Scan, or Delete.</p> </div> </div> </div> <div class="sect1"> <h2 id="_mrunit"><a class="anchor" href="#_mrunit"></a>147. MRUnit</h2> <div class="sectionbody"> <div class="paragraph"> <p><a href="http://mrunit.apache.org/">Apache MRUnit</a> is a library that allows you to unit-test MapReduce jobs. You can use it to test HBase jobs in the same way as other MapReduce jobs.</p> </div> <div class="paragraph"> <p>Given a MapReduce job that writes to an HBase table called <code>MyTest</code>, which has one column family called <code>CF</code>, the reducer of such a job could look like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">class</span> <span class="class">MyReducer</span> <span class="directive">extends</span> TableReducer<Text, Text, ImmutableBytesWritable> { <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="directive">static</span> <span class="directive">final</span> <span class="type">byte</span><span class="type">[]</span> QUALIFIER = <span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>.getBytes(); <span class="directive">public</span> <span class="type">void</span> reduce(Text key, <span class="predefined-type">Iterable</span><Text> values, <span class="predefined-type">Context</span> context) <span class="directive">throws</span> <span class="exception">IOException</span>, <span class="exception">InterruptedException</span> { <span class="comment">//bunch of processing to extract data to be inserted, in our case, lets say we are simply</span> <span class="comment">//appending all the records we receive from the mapper for this particular</span> <span class="comment">//key and insert one record into HBase</span> <span class="predefined-type">StringBuffer</span> data = <span class="keyword">new</span> <span class="predefined-type">StringBuffer</span>(); Put put = <span class="keyword">new</span> Put(Bytes.toBytes(key.toString())); <span class="keyword">for</span> (Text val : values) { data = data.append(val); } put.add(CF, QUALIFIER, Bytes.toBytes(data.toString())); <span class="comment">//write to HBase</span> context.write(<span class="keyword">new</span> ImmutableBytesWritable(Bytes.toBytes(key.toString())), put); } }</code></pre> </div> </div> <div class="paragraph"> <p>To test this code, the first step is to add a dependency to MRUnit to your Maven POM file.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.mrunit<span class="tag"></groupId></span> <span class="tag"><artifactId></span>mrunit<span class="tag"></artifactId></span> <span class="tag"><version></span>1.0.0 <span class="tag"></version></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span></code></pre> </div> </div> <div class="paragraph"> <p>Next, use the ReducerDriver provided by MRUnit, in your Reducer job.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">class</span> <span class="class">MyReducerTest</span> { ReduceDriver<Text, Text, ImmutableBytesWritable, Writable> reduceDriver; <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>.getBytes(); <span class="type">byte</span><span class="type">[]</span> QUALIFIER = <span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>.getBytes(); <span class="annotation">@Before</span> <span class="directive">public</span> <span class="type">void</span> setUp() { MyReducer reducer = <span class="keyword">new</span> MyReducer(); reduceDriver = ReduceDriver.newReduceDriver(reducer); } <span class="annotation">@Test</span> <span class="directive">public</span> <span class="type">void</span> testHBaseInsert() <span class="directive">throws</span> <span class="exception">IOException</span> { <span class="predefined-type">String</span> strKey = <span class="string"><span class="delimiter">"</span><span class="content">RowKey-1</span><span class="delimiter">"</span></span>, strValue = <span class="string"><span class="delimiter">"</span><span class="content">DATA</span><span class="delimiter">"</span></span>, strValue1 = <span class="string"><span class="delimiter">"</span><span class="content">DATA1</span><span class="delimiter">"</span></span>, strValue2 = <span class="string"><span class="delimiter">"</span><span class="content">DATA2</span><span class="delimiter">"</span></span>; <span class="predefined-type">List</span><Text> list = <span class="keyword">new</span> <span class="predefined-type">ArrayList</span><Text>(); list.add(<span class="keyword">new</span> Text(strValue)); list.add(<span class="keyword">new</span> Text(strValue1)); list.add(<span class="keyword">new</span> Text(strValue2)); <span class="comment">//since in our case all that the reducer is doing is appending the records that the mapper</span> <span class="comment">//sends it, we should get the following back</span> <span class="predefined-type">String</span> expectedOutput = strValue + strValue1 + strValue2; <span class="comment">//Setup Input, mimic what mapper would have passed</span> <span class="comment">//to the reducer and run test</span> reduceDriver.withInput(<span class="keyword">new</span> Text(strKey), list); <span class="comment">//run the reducer and get its output</span> <span class="predefined-type">List</span><Pair<ImmutableBytesWritable, Writable>> result = reduceDriver.run(); <span class="comment">//extract key from result and verify</span> assertEquals(Bytes.toString(result.get(<span class="integer">0</span>).getFirst().get()), strKey); <span class="comment">//extract value for CF/QUALIFIER and verify</span> Put a = (Put)result.get(<span class="integer">0</span>).getSecond(); <span class="predefined-type">String</span> c = Bytes.toString(a.get(CF, QUALIFIER).get(<span class="integer">0</span>).getValue()); assertEquals(expectedOutput,c ); } }</code></pre> </div> </div> <div class="paragraph"> <p>Your MRUnit test verifies that the output is as expected, the Put that is inserted into HBase has the correct value, and the ColumnFamily and ColumnQualifier have the correct values.</p> </div> <div class="paragraph"> <p>MRUnit includes a MapperDriver to test mapping jobs, and you can use MRUnit to test other operations, including reading from HBase, processing data, or writing to HDFS,</p> </div> </div> </div> <div class="sect1"> <h2 id="_integration_testing_with_a_hbase_mini_cluster"><a class="anchor" href="#_integration_testing_with_a_hbase_mini_cluster"></a>148. Integration Testing with a HBase Mini-Cluster</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase ships with HBaseTestingUtility, which makes it easy to write integration tests using a <em class="firstterm">mini-cluster</em>. The first step is to add some dependencies to your Maven POM file. Check the versions to be sure they are appropriate.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hadoop<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hadoop-common<span class="tag"></artifactId></span> <span class="tag"><version></span>2.0.0<span class="tag"></version></span> <span class="tag"><type></span>test-jar<span class="tag"></type></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span> <span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hbase<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hbase<span class="tag"></artifactId></span> <span class="tag"><version></span>0.98.3<span class="tag"></version></span> <span class="tag"><type></span>test-jar<span class="tag"></type></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span> <span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hadoop<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hadoop-hdfs<span class="tag"></artifactId></span> <span class="tag"><version></span>2.0.0<span class="tag"></version></span> <span class="tag"><type></span>test-jar<span class="tag"></type></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span> <span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hadoop<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hadoop-hdfs<span class="tag"></artifactId></span> <span class="tag"><version></span>2.0.0<span class="tag"></version></span> <span class="tag"><scope></span>test<span class="tag"></scope></span> <span class="tag"></dependency></span></code></pre> </div> </div> <div class="paragraph"> <p>This code represents an integration test for the MyDAO insert shown in <a href="#unit.tests">unit.tests</a>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">class</span> <span class="class">MyHBaseIntegrationTest</span> { <span class="directive">private</span> <span class="directive">static</span> HBaseTestingUtility utility; <span class="type">byte</span><span class="type">[]</span> CF = <span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>.getBytes(); <span class="type">byte</span><span class="type">[]</span> QUALIFIER = <span class="string"><span class="delimiter">"</span><span class="content">CQ-1</span><span class="delimiter">"</span></span>.getBytes(); <span class="annotation">@Before</span> <span class="directive">public</span> <span class="type">void</span> setup() <span class="directive">throws</span> <span class="exception">Exception</span> { utility = <span class="keyword">new</span> HBaseTestingUtility(); utility.startMiniCluster(); } <span class="annotation">@Test</span> <span class="directive">public</span> <span class="type">void</span> testInsert() <span class="directive">throws</span> <span class="exception">Exception</span> { HTableInterface table = utility.createTable(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">MyTest</span><span class="delimiter">"</span></span>), Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">CF</span><span class="delimiter">"</span></span>)); HBaseTestObj obj = <span class="keyword">new</span> HBaseTestObj(); obj.setRowKey(<span class="string"><span class="delimiter">"</span><span class="content">ROWKEY-1</span><span class="delimiter">"</span></span>); obj.setData1(<span class="string"><span class="delimiter">"</span><span class="content">DATA-1</span><span class="delimiter">"</span></span>); obj.setData2(<span class="string"><span class="delimiter">"</span><span class="content">DATA-2</span><span class="delimiter">"</span></span>); MyHBaseDAO.insertRecord(table, obj); Get get1 = <span class="keyword">new</span> Get(Bytes.toBytes(obj.getRowKey())); get1.addColumn(CF, CQ1); <span class="predefined-type">Result</span> result1 = table.get(get1); assertEquals(Bytes.toString(result1.getRow()), obj.getRowKey()); assertEquals(Bytes.toString(result1.value()), obj.getData1()); Get get2 = <span class="keyword">new</span> Get(Bytes.toBytes(obj.getRowKey())); get2.addColumn(CF, CQ2); <span class="predefined-type">Result</span> result2 = table.get(get2); assertEquals(Bytes.toString(result2.getRow()), obj.getRowKey()); assertEquals(Bytes.toString(result2.value()), obj.getData2()); } }</code></pre> </div> </div> <div class="paragraph"> <p>This code creates an HBase mini-cluster and starts it. Next, it creates a table called <code>MyTest</code> with one column family, <code>CF</code>. A record is inserted, a Get is performed from the same table, and the insertion is verified.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Starting the mini-cluster takes about 20-30 seconds, but that should be appropriate for integration testing. </td> </tr> </table> </div> <div class="paragraph"> <p>To use an HBase mini-cluster on Microsoft Windows, you need to use a Cygwin environment.</p> </div> <div class="paragraph"> <p>See the paper at <a href="http://blog.sematext.com/2010/08/30/hbase-case-study-using-hbasetestingutility-for-local-testing-development/">HBase Case-Study: Using HBaseTestingUtility for Local Testing and Development</a> (2010) for more information about HBaseTestingUtility.</p> </div> </div> </div> <h1 id="zookeeper" class="sect0"><a class="anchor" href="#zookeeper"></a>ZooKeeper</h1> <div class="openblock partintro"> <div class="content"> <div class="paragraph"> <p>A distributed Apache HBase installation depends on a running ZooKeeper cluster. All participating nodes and clients need to be able to access the running ZooKeeper ensemble. Apache HBase by default manages a ZooKeeper "cluster" for you. It will start and stop the ZooKeeper ensemble as part of the HBase start/stop process. You can also manage the ZooKeeper ensemble independent of HBase and just point HBase at the cluster it should use. To toggle HBase management of ZooKeeper, use the <code>HBASE_MANAGES_ZK</code> variable in <em>conf/hbase-env.sh</em>. This variable, which defaults to <code>true</code>, tells HBase whether to start/stop the ZooKeeper ensemble servers as part of HBase start/stop.</p> </div> <div class="paragraph"> <p>When HBase manages the ZooKeeper ensemble, you can specify ZooKeeper configuration using its native <em>zoo.cfg</em> file, or, the easier option is to just specify ZooKeeper options directly in <em>conf/hbase-site.xml</em>. A ZooKeeper configuration option can be set as a property in the HBase <em>hbase-site.xml</em> XML configuration file by prefacing the ZooKeeper option name with <code>hbase.zookeeper.property</code>. For example, the <code>clientPort</code> setting in ZooKeeper can be changed by setting the <code>hbase.zookeeper.property.clientPort</code> property. For all default values used by HBase, including ZooKeeper configuration, see <a href="#hbase_default_configurations">hbase default configurations</a>. Look for the <code>hbase.zookeeper.property</code> prefix. For the full list of ZooKeeper configurations, see ZooKeeper’s <em>zoo.cfg</em>. HBase does not ship with a <em>zoo.cfg</em> so you will need to browse the <em>conf</em> directory in an appropriate ZooKeeper download.</p> </div> <div class="paragraph"> <p>You must at least list the ensemble servers in <em>hbase-site.xml</em> using the <code>hbase.zookeeper.quorum</code> property. This property defaults to a single ensemble member at <code>localhost</code> which is not suitable for a fully distributed HBase. (It binds to the local machine only and remote clients will not be able to connect).</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="title">How many ZooKeepers should I run?</div> <div class="paragraph"> <p>You can run a ZooKeeper ensemble that comprises 1 node only but in production it is recommended that you run a ZooKeeper ensemble of 3, 5 or 7 machines; the more members an ensemble has, the more tolerant the ensemble is of host failures. Also, run an odd number of machines. In ZooKeeper, an even number of peers is supported, but it is normally not used because an even sized ensemble requires, proportionally, more peers to form a quorum than an odd sized ensemble requires. For example, an ensemble with 4 peers requires 3 to form a quorum, while an ensemble with 5 also requires 3 to form a quorum. Thus, an ensemble of 5 allows 2 peers to fail, and thus is more fault tolerant than the ensemble of 4, which allows only 1 down peer.</p> </div> <div class="paragraph"> <p>Give each ZooKeeper server around 1GB of RAM, and if possible, its own dedicated disk (A dedicated disk is the best thing you can do to ensure a performant ZooKeeper ensemble). For very heavily loaded clusters, run ZooKeeper servers on separate machines from RegionServers (DataNodes and TaskTrackers).</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>For example, to have HBase manage a ZooKeeper quorum on nodes <em>rs{1,2,3,4,5}.example.com</em>, bound to port 2222 (the default is 2181) ensure <code>HBASE_MANAGE_ZK</code> is commented out or set to <code>true</code> in <em>conf/hbase-env.sh</em> and then edit <em>conf/hbase-site.xml</em> and set <code>hbase.zookeeper.property.clientPort</code> and <code>hbase.zookeeper.quorum</code>. You should also set <code>hbase.zookeeper.property.dataDir</code> to other than the default as the default has ZooKeeper persist data under <em>/tmp</em> which is often cleared on system restart. In the example below we have ZooKeeper persist to <em>/user/local/zookeeper</em>.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"> <configuration> ... <property> <name>hbase.zookeeper.property.clientPort</name> <value><span class="integer">2222</span></value> <description>Property from ZooKeeper<span class="string"><span class="delimiter">'</span><span class="content">s config zoo.cfg. The port at which the clients will connect. </description> </property> <property> <name>hbase.zookeeper.quorum</name> <value>rs1.example.com,rs2.example.com,rs3.example.com,rs4.example.com,rs5.example.com</value> <description>Comma separated list of servers in the ZooKeeper Quorum. For example, "host1.mydomain.com,host2.mydomain.com,host3.mydomain.com". By default this is set to localhost for local and pseudo-distributed modes of operation. For a fully-distributed setup, this should be set to a full list of ZooKeeper quorum servers. If HBASE_MANAGES_ZK is set in hbase-env.sh this is the list of servers which we will start/stop ZooKeeper on. </description> </property> <property> <name>hbase.zookeeper.property.dataDir</name> <value>/usr/local/zookeeper</value> <description>Property from ZooKeeper</span><span class="delimiter">'</span></span>s config zoo.cfg. The directory where the snapshot is stored. </description> </property> ... </configuration></code></pre> </div> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> <div class="title">What verion of ZooKeeper should I use?</div> <div class="paragraph"> <p>The newer version, the better. For example, some folks have been bitten by <a href="https://issues.apache.org/jira/browse/ZOOKEEPER-1277">ZOOKEEPER-1277</a>. If running zookeeper 3.5+, you can ask hbase to make use of the new multi operation by enabling <a href="#hbase.zookeeper.usemulti">hbase.zookeeper.useMulti</a>" in your <em>hbase-site.xml</em>.</p> </div> </td> </tr> </table> </div> <div class="admonitionblock caution"> <table> <tr> <td class="icon"> <i class="fa icon-caution" title="Caution"></i> </td> <td class="content"> <div class="title">ZooKeeper Maintenance</div> <div class="paragraph"> <p>Be sure to set up the data dir cleaner described under <a href="http://zookeeper.apache.org/doc/r3.1.2/zookeeperAdmin.html#sc_maintenance">Zookeeper Maintenance</a> else you could have 'interesting' problems a couple of months in; i.e. zookeeper could start dropping sessions if it has to run through a directory of hundreds of thousands of logs which is wont to do around leader reelection time — a process rare but run on occasion whether because a machine is dropped or happens to hiccup.</p> </div> </td> </tr> </table> </div> </div> </div> <div class="sect1"> <h2 id="_using_existing_zookeeper_ensemble"><a class="anchor" href="#_using_existing_zookeeper_ensemble"></a>149. Using existing ZooKeeper ensemble</h2> <div class="sectionbody"> <div class="paragraph"> <p>To point HBase at an existing ZooKeeper cluster, one that is not managed by HBase, set <code>HBASE_MANAGES_ZK</code> in <em>conf/hbase-env.sh</em> to false</p> </div> <div class="listingblock"> <div class="content"> <pre> ... # Tell HBase whether it should manage its own instance of Zookeeper or not. export HBASE_MANAGES_ZK=false</pre> </div> </div> <div class="paragraph"> <p>Next set ensemble locations and client port, if non-standard, in <em>hbase-site.xml</em>, or add a suitably configured <em>zoo.cfg</em> to HBase’s <em>CLASSPATH</em>. HBase will prefer the configuration found in <em>zoo.cfg</em> over any settings in <em>hbase-site.xml</em>.</p> </div> <div class="paragraph"> <p>When HBase manages ZooKeeper, it will start/stop the ZooKeeper servers as a part of the regular start/stop scripts. If you would like to run ZooKeeper yourself, independent of HBase start/stop, you would do the following</p> </div> <div class="listingblock"> <div class="content"> <pre>${HBASE_HOME}/bin/hbase-daemons.sh {start,stop} zookeeper</pre> </div> </div> <div class="paragraph"> <p>Note that you can use HBase in this manner to spin up a ZooKeeper cluster, unrelated to HBase. Just make sure to set <code>HBASE_MANAGES_ZK</code> to <code>false</code> if you want it to stay up across HBase restarts so that when HBase shuts down, it doesn’t take ZooKeeper down with it.</p> </div> <div class="paragraph"> <p>For more information about running a distinct ZooKeeper cluster, see the ZooKeeper <a href="http://hadoop.apache.org/zookeeper/docs/current/zookeeperStarted.html">Getting Started Guide</a>. Additionally, see the <a href="http://wiki.apache.org/hadoop/ZooKeeper/FAQ#A7">ZooKeeper Wiki</a> or the <a href="http://zookeeper.apache.org/doc/r3.3.3/zookeeperAdmin.html#sc_zkMulitServerSetup">ZooKeeper documentation</a> for more information on ZooKeeper sizing.</p> </div> </div> </div> <div class="sect1"> <h2 id="zk.sasl.auth"><a class="anchor" href="#zk.sasl.auth"></a>150. SASL Authentication with ZooKeeper</h2> <div class="sectionbody"> <div class="paragraph"> <p>Newer releases of Apache HBase (>= 0.92) will support connecting to a ZooKeeper Quorum that supports SASL authentication (which is available in Zookeeper versions 3.4.0 or later).</p> </div> <div class="paragraph"> <p>This describes how to set up HBase to mutually authenticate with a ZooKeeper Quorum. ZooKeeper/HBase mutual authentication (<a href="https://issues.apache.org/jira/browse/HBASE-2418">HBASE-2418</a>) is required as part of a complete secure HBase configuration (<a href="https://issues.apache.org/jira/browse/HBASE-3025">HBASE-3025</a>). For simplicity of explication, this section ignores additional configuration required (Secure HDFS and Coprocessor configuration). It’s recommended to begin with an HBase-managed Zookeeper configuration (as opposed to a standalone Zookeeper quorum) for ease of learning.</p> </div> <div class="sect2"> <h3 id="_operating_system_prerequisites"><a class="anchor" href="#_operating_system_prerequisites"></a>150.1. Operating System Prerequisites</h3> <div class="paragraph"> <p>You need to have a working Kerberos KDC setup. For each <code>$HOST</code> that will run a ZooKeeper server, you should have a principle <code>zookeeper/$HOST</code>. For each such host, add a service key (using the <code>kadmin</code> or <code>kadmin.local</code> tool’s <code>ktadd</code> command) for <code>zookeeper/$HOST</code> and copy this file to <code>$HOST</code>, and make it readable only to the user that will run zookeeper on <code>$HOST</code>. Note the location of this file, which we will use below as <em>$PATH_TO_ZOOKEEPER_KEYTAB</em>.</p> </div> <div class="paragraph"> <p>Similarly, for each <code>$HOST</code> that will run an HBase server (master or regionserver), you should have a principle: <code>hbase/$HOST</code>. For each host, add a keytab file called <em>hbase.keytab</em> containing a service key for <code>hbase/$HOST</code>, copy this file to <code>$HOST</code>, and make it readable only to the user that will run an HBase service on <code>$HOST</code>. Note the location of this file, which we will use below as <em>$PATH_TO_HBASE_KEYTAB</em>.</p> </div> <div class="paragraph"> <p>Each user who will be an HBase client should also be given a Kerberos principal. This principal should usually have a password assigned to it (as opposed to, as with the HBase servers, a keytab file) which only this user knows. The client’s principal’s <code>maxrenewlife</code> should be set so that it can be renewed enough so that the user can complete their HBase client processes. For example, if a user runs a long-running HBase client process that takes at most 3 days, we might create this user’s principal within <code>kadmin</code> with: <code>addprinc -maxrenewlife 3days</code>. The Zookeeper client and server libraries manage their own ticket refreshment by running threads that wake up periodically to do the refreshment.</p> </div> <div class="paragraph"> <p>On each host that will run an HBase client (e.g. <code>hbase shell</code>), add the following file to the HBase home directory’s <em>conf</em> directory:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Client { com.sun.security.auth.module.Krb5LoginModule required useKeyTab=<span class="predefined-constant">false</span> useTicketCache=<span class="predefined-constant">true</span>; };</code></pre> </div> </div> <div class="paragraph"> <p>We’ll refer to this JAAS configuration file as <em>$CLIENT_CONF</em> below.</p> </div> </div> <div class="sect2"> <h3 id="_hbase_managed_zookeeper_configuration"><a class="anchor" href="#_hbase_managed_zookeeper_configuration"></a>150.2. HBase-managed Zookeeper Configuration</h3> <div class="paragraph"> <p>On each node that will run a zookeeper, a master, or a regionserver, create a <a href="http://docs.oracle.com/javase/1.4.2/docs/guide/security/jgss/tutorials/LoginConfigFile.html">JAAS</a> configuration file in the conf directory of the node’s <em>HBASE_HOME</em> directory that looks like the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Server { com.sun.security.auth.module.Krb5LoginModule required useKeyTab=<span class="predefined-constant">true</span> keyTab=<span class="string"><span class="delimiter">"</span><span class="content">$PATH_TO_ZOOKEEPER_KEYTAB</span><span class="delimiter">"</span></span> storeKey=<span class="predefined-constant">true</span> useTicketCache=<span class="predefined-constant">false</span> principal=<span class="string"><span class="delimiter">"</span><span class="content">zookeeper/$HOST</span><span class="delimiter">"</span></span>; }; Client { com.sun.security.auth.module.Krb5LoginModule required useKeyTab=<span class="predefined-constant">true</span> useTicketCache=<span class="predefined-constant">false</span> keyTab=<span class="string"><span class="delimiter">"</span><span class="content">$PATH_TO_HBASE_KEYTAB</span><span class="delimiter">"</span></span> principal=<span class="string"><span class="delimiter">"</span><span class="content">hbase/$HOST</span><span class="delimiter">"</span></span>; };</code></pre> </div> </div> <div class="paragraph"> <p>where the <em>$PATH_TO_HBASE_KEYTAB</em> and <em>$PATH_TO_ZOOKEEPER_KEYTAB</em> files are what you created above, and <code>$HOST</code> is the hostname for that node.</p> </div> <div class="paragraph"> <p>The <code>Server</code> section will be used by the Zookeeper quorum server, while the <code>Client</code> section will be used by the HBase master and regionservers. The path to this file should be substituted for the text <em>$HBASE_SERVER_CONF</em> in the <em>hbase-env.sh</em> listing below.</p> </div> <div class="paragraph"> <p>The path to this file should be substituted for the text <em>$CLIENT_CONF</em> in the <em>hbase-env.sh</em> listing below.</p> </div> <div class="paragraph"> <p>Modify your <em>hbase-env.sh</em> to include the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">export HBASE_OPTS="-Djava.security.auth.login.config=$CLIENT_CONF" export HBASE_MANAGES_ZK=true export HBASE_ZOOKEEPER_OPTS="-Djava.security.auth.login.config=$HBASE_SERVER_CONF" export HBASE_MASTER_OPTS="-Djava.security.auth.login.config=$HBASE_SERVER_CONF" export HBASE_REGIONSERVER_OPTS="-Djava.security.auth.login.config=$HBASE_SERVER_CONF"</code></pre> </div> </div> <div class="paragraph"> <p>where <em>$HBASE_SERVER_CONF</em> and <em>$CLIENT_CONF</em> are the full paths to the JAAS configuration files created above.</p> </div> <div class="paragraph"> <p>Modify your <em>hbase-site.xml</em> on each node that will run zookeeper, master or regionserver to contain:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><configuration> <property> <name>hbase.zookeeper.quorum</name> <value><span class="error">$</span>ZK_NODES</value> </property> <property> <name>hbase.cluster.distributed</name> <value><span class="predefined-constant">true</span></value> </property> <property> <name>hbase.zookeeper.property.authProvider<span class="float">.1</span></name> <value>org.apache.zookeeper.server.auth.SASLAuthenticationProvider</value> </property> <property> <name>hbase.zookeeper.property.kerberos.removeHostFromPrincipal</name> <value><span class="predefined-constant">true</span></value> </property> <property> <name>hbase.zookeeper.property.kerberos.removeRealmFromPrincipal</name> <value><span class="predefined-constant">true</span></value> </property> </configuration></code></pre> </div> </div> <div class="paragraph"> <p>where <code>$ZK_NODES</code> is the comma-separated list of hostnames of the Zookeeper Quorum hosts.</p> </div> <div class="paragraph"> <p>Start your hbase cluster by running one or more of the following set of commands on the appropriate hosts:</p> </div> <div class="listingblock"> <div class="content"> <pre>bin/hbase zookeeper start bin/hbase master start bin/hbase regionserver start</pre> </div> </div> </div> <div class="sect2"> <h3 id="_external_zookeeper_configuration"><a class="anchor" href="#_external_zookeeper_configuration"></a>150.3. External Zookeeper Configuration</h3> <div class="paragraph"> <p>Add a JAAS configuration file that looks like:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Client { com.sun.security.auth.module.Krb5LoginModule required useKeyTab=<span class="predefined-constant">true</span> useTicketCache=<span class="predefined-constant">false</span> keyTab=<span class="string"><span class="delimiter">"</span><span class="content">$PATH_TO_HBASE_KEYTAB</span><span class="delimiter">"</span></span> principal=<span class="string"><span class="delimiter">"</span><span class="content">hbase/$HOST</span><span class="delimiter">"</span></span>; };</code></pre> </div> </div> <div class="paragraph"> <p>where the <em>$PATH_TO_HBASE_KEYTAB</em> is the keytab created above for HBase services to run on this host, and <code>$HOST</code> is the hostname for that node. Put this in the HBase home’s configuration directory. We’ll refer to this file’s full pathname as <em>$HBASE_SERVER_CONF</em> below.</p> </div> <div class="paragraph"> <p>Modify your hbase-env.sh to include the following:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">export HBASE_OPTS="-Djava.security.auth.login.config=$CLIENT_CONF" export HBASE_MANAGES_ZK=false export HBASE_MASTER_OPTS="-Djava.security.auth.login.config=$HBASE_SERVER_CONF" export HBASE_REGIONSERVER_OPTS="-Djava.security.auth.login.config=$HBASE_SERVER_CONF"</code></pre> </div> </div> <div class="paragraph"> <p>Modify your <em>hbase-site.xml</em> on each node that will run a master or regionserver to contain:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><configuration></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zookeeper.quorum<span class="tag"></name></span> <span class="tag"><value></span>$ZK_NODES<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.cluster.distributed<span class="tag"></name></span> <span class="tag"><value></span>true<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"></configuration></span></code></pre> </div> </div> <div class="paragraph"> <p>where <code>$ZK_NODES</code> is the comma-separated list of hostnames of the Zookeeper Quorum hosts.</p> </div> <div class="paragraph"> <p>Add a <em>zoo.cfg</em> for each Zookeeper Quorum host containing:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">authProvider<span class="float">.1</span>=org.apache.zookeeper.server.auth.SASLAuthenticationProvider kerberos.removeHostFromPrincipal=<span class="predefined-constant">true</span> kerberos.removeRealmFromPrincipal=<span class="predefined-constant">true</span></code></pre> </div> </div> <div class="paragraph"> <p>Also on each of these hosts, create a JAAS configuration file containing:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">Server { com.sun.security.auth.module.Krb5LoginModule required useKeyTab=<span class="predefined-constant">true</span> keyTab=<span class="string"><span class="delimiter">"</span><span class="content">$PATH_TO_ZOOKEEPER_KEYTAB</span><span class="delimiter">"</span></span> storeKey=<span class="predefined-constant">true</span> useTicketCache=<span class="predefined-constant">false</span> principal=<span class="string"><span class="delimiter">"</span><span class="content">zookeeper/$HOST</span><span class="delimiter">"</span></span>; };</code></pre> </div> </div> <div class="paragraph"> <p>where <code>$HOST</code> is the hostname of each Quorum host. We will refer to the full pathname of this file as <em>$ZK_SERVER_CONF</em> below.</p> </div> <div class="paragraph"> <p>Start your Zookeepers on each Zookeeper Quorum host with:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">SERVER_JVMFLAGS="-Djava.security.auth.login.config=$ZK_SERVER_CONF" bin/zkServer start</code></pre> </div> </div> <div class="paragraph"> <p>Start your HBase cluster by running one or more of the following set of commands on the appropriate nodes:</p> </div> <div class="listingblock"> <div class="content"> <pre>bin/hbase master start bin/hbase regionserver start</pre> </div> </div> </div> <div class="sect2"> <h3 id="_zookeeper_server_authentication_log_output"><a class="anchor" href="#_zookeeper_server_authentication_log_output"></a>150.4. Zookeeper Server Authentication Log Output</h3> <div class="paragraph"> <p>If the configuration above is successful, you should see something similar to the following in your Zookeeper server logs:</p> </div> <div class="listingblock"> <div class="content"> <pre>11/12/05 22:43:39 INFO zookeeper.Login: successfully logged in. 11/12/05 22:43:39 INFO server.NIOServerCnxnFactory: binding to port 0.0.0.0/0.0.0.0:2181 11/12/05 22:43:39 INFO zookeeper.Login: TGT refresh thread started. 11/12/05 22:43:39 INFO zookeeper.Login: TGT valid starting at: Mon Dec 05 22:43:39 UTC 2011 11/12/05 22:43:39 INFO zookeeper.Login: TGT expires: Tue Dec 06 22:43:39 UTC 2011 11/12/05 22:43:39 INFO zookeeper.Login: TGT refresh sleeping until: Tue Dec 06 18:36:42 UTC 2011 .. 11/12/05 22:43:59 INFO auth.SaslServerCallbackHandler: Successfully authenticated client: authenticationID=hbase/ip-10-166-175-249.us-west-1.compute.internal@HADOOP.LOCALDOMAIN; authorizationID=hbase/ip-10-166-175-249.us-west-1.compute.internal@HADOOP.LOCALDOMAIN. 11/12/05 22:43:59 INFO auth.SaslServerCallbackHandler: Setting authorizedID: hbase 11/12/05 22:43:59 INFO server.ZooKeeperServer: adding SASL authorization for authorizationID: hbase</pre> </div> </div> </div> <div class="sect2"> <h3 id="_zookeeper_client_authentication_log_output"><a class="anchor" href="#_zookeeper_client_authentication_log_output"></a>150.5. Zookeeper Client Authentication Log Output</h3> <div class="paragraph"> <p>On the Zookeeper client side (HBase master or regionserver), you should see something similar to the following:</p> </div> <div class="listingblock"> <div class="content"> <pre>11/12/05 22:43:59 INFO zookeeper.ZooKeeper: Initiating client connection, connectString=ip-10-166-175-249.us-west-1.compute.internal:2181 sessionTimeout=180000 watcher=master:60000 11/12/05 22:43:59 INFO zookeeper.ClientCnxn: Opening socket connection to server /10.166.175.249:2181 11/12/05 22:43:59 INFO zookeeper.RecoverableZooKeeper: The identifier of this process is 14851@ip-10-166-175-249 11/12/05 22:43:59 INFO zookeeper.Login: successfully logged in. 11/12/05 22:43:59 INFO client.ZooKeeperSaslClient: Client will use GSSAPI as SASL mechanism. 11/12/05 22:43:59 INFO zookeeper.Login: TGT refresh thread started. 11/12/05 22:43:59 INFO zookeeper.ClientCnxn: Socket connection established to ip-10-166-175-249.us-west-1.compute.internal/10.166.175.249:2181, initiating session 11/12/05 22:43:59 INFO zookeeper.Login: TGT valid starting at: Mon Dec 05 22:43:59 UTC 2011 11/12/05 22:43:59 INFO zookeeper.Login: TGT expires: Tue Dec 06 22:43:59 UTC 2011 11/12/05 22:43:59 INFO zookeeper.Login: TGT refresh sleeping until: Tue Dec 06 18:30:37 UTC 2011 11/12/05 22:43:59 INFO zookeeper.ClientCnxn: Session establishment complete on server ip-10-166-175-249.us-west-1.compute.internal/10.166.175.249:2181, sessionid = 0x134106594320000, negotiated timeout = 180000</pre> </div> </div> </div> <div class="sect2"> <h3 id="_configuration_from_scratch"><a class="anchor" href="#_configuration_from_scratch"></a>150.6. Configuration from Scratch</h3> <div class="paragraph"> <p>This has been tested on the current standard Amazon Linux AMI. First setup KDC and principals as described above. Next checkout code and run a sanity check.</p> </div> <div class="listingblock"> <div class="content"> <pre>git clone https://gitbox.apache.org/repos/asf/hbase.git cd hbase mvn clean test -Dtest=TestZooKeeperACL</pre> </div> </div> <div class="paragraph"> <p>Then configure HBase as described above. Manually edit target/cached_classpath.txt (see below):</p> </div> <div class="listingblock"> <div class="content"> <pre>bin/hbase zookeeper & bin/hbase master & bin/hbase regionserver &</pre> </div> </div> </div> <div class="sect2"> <h3 id="_future_improvements"><a class="anchor" href="#_future_improvements"></a>150.7. Future improvements</h3> <div class="sect3"> <h4 id="_fix_target_cached_classpath_txt"><a class="anchor" href="#_fix_target_cached_classpath_txt"></a>150.7.1. Fix target/cached_classpath.txt</h4> <div class="paragraph"> <p>You must override the standard hadoop-core jar file from the <code>target/cached_classpath.txt</code> file with the version containing the HADOOP-7070 fix. You can use the following script to do this:</p> </div> <div class="listingblock"> <div class="content"> <pre>echo `find ~/.m2 -name "*hadoop-core*7070*SNAPSHOT.jar"` ':' `cat target/cached_classpath.txt` | sed 's/ //g' > target/tmp.txt mv target/tmp.txt target/cached_classpath.txt</pre> </div> </div> </div> <div class="sect3"> <h4 id="_set_jaas_configuration_programmatically"><a class="anchor" href="#_set_jaas_configuration_programmatically"></a>150.7.2. Set JAAS configuration programmatically</h4> <div class="paragraph"> <p>This would avoid the need for a separate Hadoop jar that fixes <a href="https://issues.apache.org/jira/browse/HADOOP-7070">HADOOP-7070</a>.</p> </div> </div> <div class="sect3"> <h4 id="_elimination_of_code_kerberos_removehostfromprincipal_code_and_kerberos_removerealmfromprincipal"><a class="anchor" href="#_elimination_of_code_kerberos_removehostfromprincipal_code_and_kerberos_removerealmfromprincipal"></a>150.7.3. Elimination of <code>kerberos.removeHostFromPrincipal</code> and`kerberos.removeRealmFromPrincipal`</h4> </div> </div> </div> </div> <h1 id="community" class="sect0"><a class="anchor" href="#community"></a>Community</h1> <div class="sect1"> <h2 id="_decisions"><a class="anchor" href="#_decisions"></a>151. Decisions</h2> <div class="sectionbody"> <div class="paragraph"> <div class="title">Feature Branches</div> <p>Feature Branches are easy to make. You do not have to be a committer to make one. Just request the name of your branch be added to JIRA up on the developer’s mailing list and a committer will add it for you. Thereafter you can file issues against your feature branch in Apache HBase JIRA. Your code you keep elsewhere — it should be public so it can be observed — and you can update dev mailing list on progress. When the feature is ready for commit, 3 +1s from committers will get your feature merged. See <a href="http://search-hadoop.com/m/asM982C5FkS1">HBase, mail # dev - Thoughts about large feature dev branches</a></p> </div> <div id="patchplusonepolicy" class="paragraph"> <div class="title">Patch +1 Policy</div> <p>The below policy is something we put in place 09/2012. It is a suggested policy rather than a hard requirement. We want to try it first to see if it works before we cast it in stone.</p> </div> <div class="paragraph"> <p>Apache HBase is made of <a href="https://issues.apache.org/jira/browse/HBASE#selectedTab=com.atlassian.jira.plugin.system.project%3Acomponents-panel">components</a>. Components have one or more <a href="#owner">OWNER</a>s. See the 'Description' field on the <a href="https://issues.apache.org/jira/browse/HBASE#selectedTab=com.atlassian.jira.plugin.system.project%3Acomponents-panel">components</a> JIRA page for who the current owners are by component.</p> </div> <div class="paragraph"> <p>Patches that fit within the scope of a single Apache HBase component require, at least, a +1 by one of the component’s owners before commit. If owners are absent — busy or otherwise — two +1s by non-owners will suffice.</p> </div> <div class="paragraph"> <p>Patches that span components need at least two +1s before they can be committed, preferably +1s by owners of components touched by the x-component patch (TODO: This needs tightening up but I think fine for first pass).</p> </div> <div class="paragraph"> <p>Any -1 on a patch by anyone vetos a patch; it cannot be committed until the justification for the -1 is addressed.</p> </div> <div id="hbase.fix.version.in.jira" class="paragraph"> <div class="title">How to set fix version in JIRA on issue resolve</div> <p>Here is how <a href="http://search-hadoop.com/m/azemIi5RCJ1">we agreed</a> to set versions in JIRA when we resolve an issue. If trunk is going to be 0.98.0 then:</p> </div> <div class="ulist"> <ul> <li> <p>Commit only to trunk: Mark with 0.98</p> </li> <li> <p>Commit to 0.95 and trunk : Mark with 0.98, and 0.95.x</p> </li> <li> <p>Commit to 0.94.x and 0.95, and trunk: Mark with 0.98, 0.95.x, and 0.94.x</p> </li> <li> <p>Commit to 89-fb: Mark with 89-fb.</p> </li> <li> <p>Commit site fixes: no version</p> </li> </ul> </div> <div id="hbase.when.to.close.jira" class="paragraph"> <div class="title">Policy on when to set a RESOLVED JIRA as CLOSED</div> <p>We <a href="http://search-hadoop.com/m/4cIKs1iwXMS1">agreed</a> that for issues that list multiple releases in their <em>Fix Version/s</em> field, CLOSE the issue on the release of any of the versions listed; subsequent change to the issue must happen in a new JIRA.</p> </div> <div id="no.permanent.state.in.zk" class="paragraph"> <div class="title">Only transient state in ZooKeeper!</div> <p>You should be able to kill the data in zookeeper and hbase should ride over it recreating the zk content as it goes. This is an old adage around these parts. We just made note of it now. We also are currently in violation of this basic tenet — replication at least keeps permanent state in zk — but we are working to undo this breaking of a golden rule.</p> </div> </div> </div> <div class="sect1"> <h2 id="community.roles"><a class="anchor" href="#community.roles"></a>152. Community Roles</h2> <div class="sectionbody"> <div id="owner" class="paragraph"> <div class="title">Component Owner/Lieutenant</div> <p>Component owners are listed in the description field on this Apache HBase JIRA <a href="https://issues.apache.org/jira/browse/HBASE#selectedTab=com.atlassian.jira.plugin.system.project%3Acomponents-panel">components</a> page. The owners are listed in the 'Description' field rather than in the 'Component Lead' field because the latter only allows us list one individual whereas it is encouraged that components have multiple owners.</p> </div> <div class="paragraph"> <p>Owners or component lieutenants are volunteers who are (usually, but not necessarily) expert in their component domain and may have an agenda on how they think their Apache HBase component should evolve.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Owners will try and review patches that land within their component’s scope.</p> </li> <li> <p>If applicable, if an owner has an agenda, they will publish their goals or the design toward which they are driving their component</p> </li> </ol> </div> <div class="paragraph"> <p>If you would like to be volunteer as a component owner, just write the dev list and we’ll sign you up. Owners do not need to be committers.</p> </div> </div> </div> <div class="sect1"> <h2 id="hbase.commit.msg.format"><a class="anchor" href="#hbase.commit.msg.format"></a>153. Commit Message format</h2> <div class="sectionbody"> <div class="paragraph"> <p>We <a href="http://search-hadoop.com/m/Gwxwl10cFHa1">agreed</a> to the following SVN commit message format:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">HBASE-xxxxx <title>. (<contributor>)</code></pre> </div> </div> <div class="paragraph"> <p>If the person making the commit is the contributor, leave off the '(<contributor>)' element.</p> </div> </div> </div> <h1 id="_appendix" class="sect0"><a class="anchor" href="#_appendix"></a>Appendix</h1> <div class="sect1"> <h2 id="appendix_contributing_to_documentation"><a class="anchor" href="#appendix_contributing_to_documentation"></a>Appendix A: Contributing to Documentation</h2> <div class="sectionbody"> <div class="paragraph"> <p>The Apache HBase project welcomes contributions to all aspects of the project, including the documentation.</p> </div> <div class="paragraph"> <p>In HBase, documentation includes the following areas, and probably some others:</p> </div> <div class="ulist"> <ul> <li> <p>The <a href="http://hbase.apache.org/book.html">HBase Reference Guide</a> (this book)</p> </li> <li> <p>The <a href="http://hbase.apache.org/">HBase website</a></p> </li> <li> <p>The <a href="http://wiki.apache.org/hadoop/Hbase">HBase Wiki</a></p> </li> <li> <p>API documentation</p> </li> <li> <p>Command-line utility output and help text</p> </li> <li> <p>Web UI strings, explicit help text, context-sensitive strings, and others</p> </li> <li> <p>Log messages</p> </li> <li> <p>Comments in source files, configuration files, and others</p> </li> <li> <p>Localization of any of the above into target languages other than English</p> </li> </ul> </div> <div class="paragraph"> <p>No matter which area you want to help out with, the first step is almost always to download (typically by cloning the Git repository) and familiarize yourself with the HBase source code. The only exception in the list above is the HBase Wiki, which is edited online. For information on downloading and building the source, see <a href="#developer">developer</a>.</p> </div> <div class="sect2"> <h3 id="_getting_access_to_the_wiki"><a class="anchor" href="#_getting_access_to_the_wiki"></a>A.1. Getting Access to the Wiki</h3> <div class="paragraph"> <p>The HBase Wiki is not well-maintained and much of its content has been moved into the HBase Reference Guide (this guide). However, some pages on the Wiki are well maintained, and it would be great to have some volunteers willing to help out with the Wiki. To request access to the Wiki, register a new account at <a href="https://wiki.apache.org/hadoop/Hbase?action=newaccount">https://wiki.apache.org/hadoop/Hbase?action=newaccount</a>. Contact one of the HBase committers, who can either give you access or refer you to someone who can.</p> </div> </div> <div class="sect2"> <h3 id="_contributing_to_documentation_or_other_strings"><a class="anchor" href="#_contributing_to_documentation_or_other_strings"></a>A.2. Contributing to Documentation or Other Strings</h3> <div class="paragraph"> <p>If you spot an error in a string in a UI, utility, script, log message, or elsewhere, or you think something could be made more clear, or you think text needs to be added where it doesn’t currently exist, the first step is to file a JIRA. Be sure to set the component to <code>Documentation</code> in addition any other involved components. Most components have one or more default owners, who monitor new issues which come into those queues. Regardless of whether you feel able to fix the bug, you should still file bugs where you see them.</p> </div> <div class="paragraph"> <p>If you want to try your hand at fixing your newly-filed bug, assign it to yourself. You will need to clone the HBase Git repository to your local system and work on the issue there. When you have developed a potential fix, submit it for review. If it addresses the issue and is seen as an improvement, one of the HBase committers will commit it to one or more branches, as appropriate.</p> </div> <div class="paragraph"> <div class="title">Procedure: Suggested Work flow for Submitting Patches</div> <p>This procedure goes into more detail than Git pros will need, but is included in this appendix so that people unfamiliar with Git can feel confident contributing to HBase while they learn.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>If you have not already done so, clone the Git repository locally. You only need to do this once.</p> </li> <li> <p>Fairly often, pull remote changes into your local repository by using the <code>git pull</code> command, while your master branch is checked out.</p> </li> <li> <p>For each issue you work on, create a new branch. One convention that works well for naming the branches is to name a given branch the same as the JIRA it relates to:</p> <div class="listingblock"> <div class="content"> <pre>$ git checkout -b HBASE-123456</pre> </div> </div> </li> <li> <p>Make your suggested changes on your branch, committing your changes to your local repository often. If you need to switch to working on a different issue, remember to check out the appropriate branch.</p> </li> <li> <p>When you are ready to submit your patch, first be sure that HBase builds cleanly and behaves as expected in your modified branch. If you have made documentation changes, be sure the documentation and website builds by running <code>mvn clean site</code>.</p> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Before you use the <code>site</code> target the very first time, be sure you have built HBase at least once, in order to fetch all the Maven dependencies you need. </td> </tr> </table> </div> <div class="listingblock"> <div class="content"> <pre>$ mvn clean install -DskipTests # Builds HBase</pre> </div> </div> <div class="listingblock"> <div class="content"> <pre>$ mvn clean site -DskipTests # Builds the website and documentation</pre> </div> </div> <div class="paragraph"> <p>If any errors occur, address them.</p> </div> </li> <li> <p>If it takes you several days or weeks to implement your fix, or you know that the area of the code you are working in has had a lot of changes lately, make sure you rebase your branch against the remote master and take care of any conflicts before submitting your patch.</p> <div class="listingblock"> <div class="content"> <pre>$ git checkout HBASE-123456 $ git rebase origin/master</pre> </div> </div> </li> <li> <p>Generate your patch against the remote master. Run the following command from the top level of your git repository (usually called <code>hbase</code>):</p> <div class="listingblock"> <div class="content"> <pre>$ git format-patch --stdout origin/master > HBASE-123456.patch</pre> </div> </div> <div class="paragraph"> <p>The name of the patch should contain the JIRA ID. Look over the patch file to be sure that you did not change any additional files by accident and that there are no other surprises. When you are satisfied, attach the patch to the JIRA and click the <b class="button">Patch Available</b> button. A reviewer will review your patch. If you need to submit a new version of the patch, leave the old one on the JIRA and add a version number to the name of the new patch.</p> </div> </li> <li> <p>After a change has been committed, there is no need to keep your local branch around. Instead you should run <code>git pull</code> to get the new change into your master branch.</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="_editing_the_hbase_website"><a class="anchor" href="#_editing_the_hbase_website"></a>A.3. Editing the HBase Website</h3> <div class="paragraph"> <p>The source for the HBase website is in the HBase source, in the <em>src/main/site/</em> directory. Within this directory, source for the individual pages is in the <em>xdocs/</em> directory, and images referenced in those pages are in the <em>images/</em> directory. This directory also stores images used in the HBase Reference Guide.</p> </div> <div class="paragraph"> <p>The website’s pages are written in an HTML-like XML dialect called xdoc, which has a reference guide at link:http://maven.apache.org/archives/maven-1.x/plugins/xdoc/reference/xdocs.html. You can edit these files in a plain-text editor, an IDE, or an XML editor such as XML Mind XML Editor (XXE) or Oxygen XML Author.</p> </div> <div class="paragraph"> <p>To preview your changes, build the website using the mvn clean site -DskipTests command. The HTML output resides in the <em>target/site/</em> directory. When you are satisfied with your changes, follow the procedure in <a href="#submit_doc_patch_procedure">submit doc patch procedure</a> to submit your patch.</p> </div> </div> <div class="sect2"> <h3 id="_hbase_reference_guide_style_guide_and_cheat_sheet"><a class="anchor" href="#_hbase_reference_guide_style_guide_and_cheat_sheet"></a>A.4. HBase Reference Guide Style Guide and Cheat Sheet</h3> <div class="paragraph"> <p>The HBase Reference Guide is written in Asciidoc and built using <a href="http://asciidoctor.org">AsciiDoctor</a>. The following cheat sheet is included for your reference. More nuanced and comprehensive documentation is available at link:http://asciidoctor.org/docs/user-manual/.</p> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 11. AsciiDoc Cheat Sheet</caption> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Element Type</th> <th class="tableblock halign-left valign-top">Desired Rendering</th> <th class="tableblock halign-left valign-top">How to do it</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">A paragraph</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">a paragraph</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>Just type some text with a blank line at the top and bottom.</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Add line breaks within a paragraph without adding blank lines</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Manual line breaks</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>This will break + at the plus sign. Or prefix the whole paragraph with a line containing '[%hardbreaks]'</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Give a title to anything</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Colored italic bold differently-sized text</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> </div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">In-Line Code or commands</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">monospace</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>`text`</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">In-line literal content (things to be typed exactly as shown)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">bold mono</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>*`typethis`*</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">In-line replaceable content (things to substitute with your own values)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">bold italic mono</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>*_typesomething_*</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Code blocks with highlighting</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">monospace, highlighted, preserve space</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="literalblock"> <div class="content"> <pre>[source,java] ---- myAwesomeCode() { } ----</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Code block included from a separate file</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">included just as though it were part of the main file</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="literalblock"> <div class="content"> <pre>[source,ruby] ---- include\::path/to/app.rb[] ----</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Include only part of a separate file</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Similar to Javadoc</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>See link:http://asciidoctor.org/docs/user-manual/#by-tagged-regions</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Filenames, directory names, new terms</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">italic</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>_hbase-default.xml_</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">External naked URLs</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">A link with the URL as link text</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>link:http://www.google.com</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">External URLs with text</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">A link with arbitrary link text</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>link:http://www.google.com[Google]</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Create an internal anchor to cross-reference</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">not rendered</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>[[anchor_name]]</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Cross-reference an existing anchor using its default title</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">an internal hyperlink using the element title if available, otherwise using the anchor name</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre><<anchor_name>></pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Cross-reference an existing anchor using custom text</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">an internal hyperlink using arbitrary text</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre><<anchor_name,Anchor Text>></pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">A block image</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The image with alt text</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>image::sunset.jpg[Alt Text]</pre> </div> </div> <div class="paragraph"> <p>(put the image in the src/main/site/resources/images directory)</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">An inline image</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The image with alt text, as part of the text flow</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>image:sunset.jpg [Alt Text]</pre> </div> </div> <div class="paragraph"> <p>(only one colon)</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Link to a remote image</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">show an image hosted elsewhere</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>image::http://inkscape.org/doc/examples/tux.svg[Tux,250,350]</pre> </div> </div> <div class="paragraph"> <p>(or <code>image:</code>)</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Add dimensions or a URL to the image</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">depends</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>inside the brackets after the alt text, specify width, height and/or link="http://my_link.com"</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">A footnote</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">subscript link which takes you to the footnote</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>Some text.footnote:[The footnote text.]</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">A note or warning with no title</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The admonition image followed by the admonition</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>NOTE:My note here</pre> </div> </div> <div class="listingblock"> <div class="content"> <pre>WARNING:My warning here</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">A complex note</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The note has a title and/or multiple paragraphs and/or code blocks or lists, etc</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="literalblock"> <div class="content"> <pre>.The Title [NOTE] ==== Here is the note text. Everything until the second set of four equals signs is part of the note. ---- some source code ---- ====</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Bullet lists</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">bullet lists</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>* list item 1</pre> </div> </div> <div class="paragraph"> <p>(see <a href="http://asciidoctor.org/docs/user-manual/#unordered-lists" class="bare">http://asciidoctor.org/docs/user-manual/#unordered-lists</a>)</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Numbered lists</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">numbered list</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>. list item 2</pre> </div> </div> <div class="paragraph"> <p>(see <a href="http://asciidoctor.org/docs/user-manual/#ordered-lists" class="bare">http://asciidoctor.org/docs/user-manual/#ordered-lists</a>)</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Checklists</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Checked or unchecked boxes</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>Checked:</p> </div> <div class="listingblock"> <div class="content"> <pre>- [*]</pre> </div> </div> <div class="paragraph"> <p>Unchecked:</p> </div> <div class="listingblock"> <div class="content"> <pre>- [ ]</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Multiple levels of lists</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">bulleted or numbered or combo</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>. Numbered (1), at top level * Bullet (2), nested under 1 * Bullet (3), nested under 1 . Numbered (4), at top level * Bullet (5), nested under 4 ** Bullet (6), nested under 5 - [x] Checked (7), at top level</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Labelled lists / variablelists</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">a list item title or summary followed by content</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>Title:: content Title:: content</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Sidebars, quotes, or other blocks of text</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">a block of text, formatted differently from the default</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>Delimited using different delimiters, see link:http://asciidoctor.org/docs/user-manual/#built-in-blocks-summary. Some of the examples above use delimiters like ...., ----,====.</p> </div> <div class="literalblock"> <div class="content"> <pre>[example] ==== This is an example block. ==== [source] ---- This is a source block. ---- [note] ==== This is a note block. ==== [quote] ____ This is a quote block. ____</pre> </div> </div> <div class="paragraph"> <p>If you want to insert literal Asciidoc content that keeps being interpreted, when in doubt, use eight dots as the delimiter at the top and bottom.</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Nested Sections</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">chapter, section, sub-section, etc</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>= Book (or chapter if the chapter can be built alone, see the leveloffset info below) == Chapter (or section if the chapter is standalone) === Section (or subsection, etc) ==== Subsection</pre> </div> </div> <div class="paragraph"> <p>and so on up to 6 levels (think carefully about going deeper than 4 levels, maybe you can just titled paragraphs or lists instead). Note that you can include a book inside another book by adding the <code>:leveloffset:+1</code> macro directive directly before your include, and resetting it to 0 directly after. See the <em>book.adoc</em> source for examples, as this is how this guide handles chapters. <strong>Don’t do it for prefaces, glossaries, appendixes, or other special types of chapters.</strong></p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Include one file from another</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Content is included as though it were inline</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>include::[/path/to/file.adoc]</pre> </div> </div> <div class="paragraph"> <p>For plenty of examples. see <em>book.adoc</em>.</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">A table</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">a table</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p>See <a href="http://asciidoctor.org/docs/user-manual/#tables" class="bare">http://asciidoctor.org/docs/user-manual/#tables</a>. Generally rows are separated by newlines and columns by pipes</p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Comment out a single line</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">A line is skipped during rendering</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="paragraph"> <p><code>// This line won’t show up</code></p> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Comment out a block</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">A section of the file is skipped during rendering</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>//// Nothing between the slashes will show up. ////</pre> </div> </div></div></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Highlight text for review</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">text shows up with yellow background</p></td> <td class="tableblock halign-left valign-top"><div><div id="toc" class="toc"> <div id="toctitle">Contents</div> </div> <div class="listingblock"> <div class="content"> <pre>Test between #hash marks# is highlighted yellow.</pre> </div> </div></div></td> </tr> </tbody> </table> </div> <div class="sect2"> <h3 id="_auto_generated_content"><a class="anchor" href="#_auto_generated_content"></a>A.5. Auto-Generated Content</h3> <div class="paragraph"> <p>Some parts of the HBase Reference Guide, most notably <a href="#config.files">config.files</a>, are generated automatically, so that this area of the documentation stays in sync with the code. This is done by means of an XSLT transform, which you can examine in the source at <em>src/main/xslt/configuration_to_asciidoc_chapter.xsl</em>. This transforms the <em>hbase-common/src/main/resources/hbase-default.xml</em> file into an Asciidoc output which can be included in the Reference Guide. Sometimes, it is necessary to add configuration parameters or modify their descriptions. Make the modifications to the source file, and they will be included in the Reference Guide when it is rebuilt.</p> </div> <div class="paragraph"> <p>It is possible that other types of content can and will be automatically generated from HBase source files in the future.</p> </div> </div> <div class="sect2"> <h3 id="_images_in_the_hbase_reference_guide"><a class="anchor" href="#_images_in_the_hbase_reference_guide"></a>A.6. Images in the HBase Reference Guide</h3> <div class="paragraph"> <p>You can include images in the HBase Reference Guide. It is important to include an image title if possible, and alternate text always. This allows screen readers to navigate to the image and also provides alternative text for the image. The following is an example of an image with a title and alternate text. Notice the double colon.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="asciidoc">.My Image Title image::sunset.jpg[Alt Text]</code></pre> </div> </div> <div class="paragraph"> <p>Here is an example of an inline image with alternate text. Notice the single colon. Inline images cannot have titles. They are generally small images like GUI buttons.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="asciidoc">image:sunset.jpg[Alt Text]</code></pre> </div> </div> <div class="paragraph"> <p>When doing a local build, save the image to the <em>src/main/site/resources/images/</em> directory. When you link to the image, do not include the directory portion of the path. The image will be copied to the appropriate target location during the build of the output.</p> </div> <div class="paragraph"> <p>When you submit a patch which includes adding an image to the HBase Reference Guide, attach the image to the JIRA. If the committer asks where the image should be committed, it should go into the above directory.</p> </div> </div> <div class="sect2"> <h3 id="_adding_a_new_chapter_to_the_hbase_reference_guide"><a class="anchor" href="#_adding_a_new_chapter_to_the_hbase_reference_guide"></a>A.7. Adding a New Chapter to the HBase Reference Guide</h3> <div class="paragraph"> <p>If you want to add a new chapter to the HBase Reference Guide, the easiest way is to copy an existing chapter file, rename it, and change the ID (in double brackets) and title. Chapters are located in the <em>src/main/asciidoc/_chapters/</em> directory.</p> </div> <div class="paragraph"> <p>Delete the existing content and create the new content. Then open the <em>src/main/asciidoc/book.adoc</em> file, which is the main file for the HBase Reference Guide, and copy an existing <code>include</code> element to include your new chapter in the appropriate location. Be sure to add your new file to your Git repository before creating your patch.</p> </div> <div class="paragraph"> <p>When in doubt, check to see how other files have been included.</p> </div> </div> <div class="sect2"> <h3 id="_common_documentation_issues"><a class="anchor" href="#_common_documentation_issues"></a>A.8. Common Documentation Issues</h3> <div class="paragraph"> <p>The following documentation issues come up often. Some of these are preferences, but others can create mysterious build errors or other problems.</p> </div> <div class="qlist qanda"> <ol> <li> <p><em>Isolate Changes for Easy Diff Review.</em></p> <p>Be careful with pretty-printing or re-formatting an entire XML file, even if the formatting has degraded over time. If you need to reformat a file, do that in a separate JIRA where you do not change any content. Be careful because some XML editors do a bulk-reformat when you open a new file, especially if you use GUI mode in the editor.</p> </li> <li> <p><em>Syntax Highlighting</em></p> <p>The HBase Reference Guide uses <code>coderay</code> for syntax highlighting. To enable syntax highlighting for a given code listing, use the following type of syntax:</p> <div class="literalblock"> <div class="content"> <pre>[source,xml] ---- <name>My Name</name> ----</pre> </div> </div> <div class="paragraph"> <p>Several syntax types are supported. The most interesting ones for the HBase Reference Guide are <code>java</code>, <code>xml</code>, <code>sql</code>, and <code>bash</code>.</p> </div> </li> </ol> </div> </div> </div> </div> <div class="sect1"> <h2 id="faq"><a class="anchor" href="#faq"></a>Appendix B: FAQ</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="_general"><a class="anchor" href="#_general"></a>B.1. General</h3> <div class="dlist"> <dl> <dt class="hdlist1">When should I use HBase?</dt> <dd> <p>See <a href="#arch.overview">Overview</a> in the Architecture chapter.</p> </dd> <dt class="hdlist1">Are there other HBase FAQs?</dt> <dd> <p>See the FAQ that is up on the wiki, <a href="http://wiki.apache.org/hadoop/Hbase/FAQ">HBase Wiki FAQ</a>.</p> </dd> <dt class="hdlist1">Does HBase support SQL?</dt> <dd> <p>Not really. SQL-ish support for HBase via <a href="http://hive.apache.org/">Hive</a> is in development, however Hive is based on MapReduce which is not generally suitable for low-latency requests. See the <a href="#datamodel">Data Model</a> section for examples on the HBase client.</p> </dd> <dt class="hdlist1">How can I find examples of NoSQL/HBase?</dt> <dd> <p>See the link to the BigTable paper in <a href="#other.info">Other Information About HBase</a>, as well as the other papers.</p> </dd> <dt class="hdlist1">What is the history of HBase?</dt> <dd> <p>See <a href="#hbase.history">hbase.history</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_upgrading"><a class="anchor" href="#_upgrading"></a>B.2. Upgrading</h3> <div class="dlist"> <dl> <dt class="hdlist1">How do I upgrade Maven-managed projects from HBase 0.94 to HBase 0.96+?</dt> <dd> <p>In HBase 0.96, the project moved to a modular structure. Adjust your project’s dependencies to rely upon the <code>hbase-client</code> module or another module as appropriate, rather than a single JAR. You can model your Maven depency after one of the following, depending on your targeted version of HBase. See Section 3.5, “Upgrading from 0.94.x to 0.96.x” or Section 3.3, “Upgrading from 0.96.x to 0.98.x” for more information.</p> <div class="listingblock"> <div class="title">Maven Dependency for HBase 0.98</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hbase<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hbase-client<span class="tag"></artifactId></span> <span class="tag"><version></span>0.98.5-hadoop2<span class="tag"></version></span> <span class="tag"></dependency></span></code></pre> </div> </div> <div class="listingblock"> <div class="title">Maven Dependency for HBase 0.96</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hbase<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hbase-client<span class="tag"></artifactId></span> <span class="tag"><version></span>0.96.2-hadoop2<span class="tag"></version></span> <span class="tag"></dependency></span></code></pre> </div> </div> <div class="listingblock"> <div class="title">Maven Dependency for HBase 0.94</div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><dependency></span> <span class="tag"><groupId></span>org.apache.hbase<span class="tag"></groupId></span> <span class="tag"><artifactId></span>hbase<span class="tag"></artifactId></span> <span class="tag"><version></span>0.94.3<span class="tag"></version></span> <span class="tag"></dependency></span></code></pre> </div> </div> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_architecture_2"><a class="anchor" href="#_architecture_2"></a>B.3. Architecture</h3> <div class="dlist"> <dl> <dt class="hdlist1">How does HBase handle Region-RegionServer assignment and locality?</dt> <dd> <p>See <a href="#regions.arch">Regions</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_configuration_2"><a class="anchor" href="#_configuration_2"></a>B.4. Configuration</h3> <div class="dlist"> <dl> <dt class="hdlist1">How can I get started with my first cluster?</dt> <dd> <p>See <a href="#quickstart">Quick Start - Standalone HBase</a>.</p> </dd> <dt class="hdlist1">Where can I learn about the rest of the configuration options?</dt> <dd> <p>See <a href="#configuration">Apache HBase Configuration</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_schema_design_data_access"><a class="anchor" href="#_schema_design_data_access"></a>B.5. Schema Design / Data Access</h3> <div class="dlist"> <dl> <dt class="hdlist1">How should I design my schema in HBase?</dt> <dd> <p>See <a href="#datamodel">Data Model</a> and <a href="#schema">HBase and Schema Design</a>.</p> </dd> <dt class="hdlist1">How can I store (fill in the blank) in HBase?</dt> <dd> <p>See <a href="#supported.datatypes">Supported Datatypes</a>.</p> </dd> <dt class="hdlist1">How can I handle secondary indexes in HBase?</dt> <dd> <p>See <a href="#secondary.indexes">Secondary Indexes and Alternate Query Paths</a>.</p> </dd> <dt class="hdlist1">Can I change a table’s rowkeys?</dt> <dd> <p>This is a very common question. You can’t. See <a href="#changing.rowkeys">Immutability of Rowkeys</a>.</p> </dd> <dt class="hdlist1">What APIs does HBase support?</dt> <dd> <p>See <a href="#datamodel">Data Model</a>, <a href="#architecture.client">Client</a>, and <a href="#nonjava.jvm">Non-Java Languages Talking to the JVM</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_mapreduce"><a class="anchor" href="#_mapreduce"></a>B.6. MapReduce</h3> <div class="dlist"> <dl> <dt class="hdlist1">How can I use MapReduce with HBase?</dt> <dd> <p>See <a href="#mapreduce">HBase and MapReduce</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_performance_and_troubleshooting"><a class="anchor" href="#_performance_and_troubleshooting"></a>B.7. Performance and Troubleshooting</h3> <div class="dlist"> <dl> <dt class="hdlist1">How can I improve HBase cluster performance?</dt> <dd> <p>See <a href="#performance">Apache HBase Performance Tuning</a>.</p> </dd> <dt class="hdlist1">How can I troubleshoot my HBase cluster?</dt> <dd> <p>See <a href="#trouble">Troubleshooting and Debugging Apache HBase</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_amazon_ec2"><a class="anchor" href="#_amazon_ec2"></a>B.8. Amazon EC2</h3> <div class="dlist"> <dl> <dt class="hdlist1">I am running HBase on Amazon EC2 and…</dt> <dd> <p>EC2 issues are a special case. See <a href="#trouble.ec2">Amazon EC2</a> and <a href="#perf.ec2">Amazon EC2</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_operations"><a class="anchor" href="#_operations"></a>B.9. Operations</h3> <div class="dlist"> <dl> <dt class="hdlist1">How do I manage my HBase cluster?</dt> <dd> <p>See <a href="#ops_mgt">Apache HBase Operational Management</a>.</p> </dd> <dt class="hdlist1">How do I back up my HBase cluster?</dt> <dd> <p>See <a href="#ops.backup">HBase Backup</a>.</p> </dd> </dl> </div> </div> <div class="sect2"> <h3 id="_hbase_in_action"><a class="anchor" href="#_hbase_in_action"></a>B.10. HBase in Action</h3> <div class="dlist"> <dl> <dt class="hdlist1">Where can I find interesting videos and presentations on HBase?</dt> <dd> <p>See <a href="#other.info">Other Information About HBase</a>.</p> </dd> </dl> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbck.in.depth"><a class="anchor" href="#hbck.in.depth"></a>Appendix C: hbck In Depth</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBaseFsck (hbck) is a tool for checking for region consistency and table integrity problems and repairing a corrupted HBase. It works in two basic modes — a read-only inconsistency identifying mode and a multi-phase read-write repair mode.</p> </div> <div class="sect2"> <h3 id="_running_hbck_to_identify_inconsistencies"><a class="anchor" href="#_running_hbck_to_identify_inconsistencies"></a>C.1. Running hbck to identify inconsistencies</h3> <div class="paragraph"> <p>To check to see if your HBase cluster has corruptions, run hbck against your HBase cluster:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck</code></pre> </div> </div> <div class="paragraph"> <p>At the end of the commands output it prints OK or tells you the number of INCONSISTENCIES present. You may also want to run run hbck a few times because some inconsistencies can be transient (e.g. cluster is starting up or a region is splitting). Operationally you may want to run hbck regularly and setup alert (e.g. via nagios) if it repeatedly reports inconsistencies . A run of hbck will report a list of inconsistencies along with a brief description of the regions and tables affected. The using the <code>-details</code> option will report more details including a representative listing of all the splits present in all the tables.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck -details</code></pre> </div> </div> <div class="paragraph"> <p>If you just want to know if some tables are corrupted, you can limit hbck to identify inconsistencies in only specific tables. For example the following command would only attempt to check table TableFoo and TableBar. The benefit is that hbck will run in less time.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck TableFoo TableBar</code></pre> </div> </div> </div> <div class="sect2"> <h3 id="_inconsistencies"><a class="anchor" href="#_inconsistencies"></a>C.2. Inconsistencies</h3> <div class="paragraph"> <p>If after several runs, inconsistencies continue to be reported, you may have encountered a corruption. These should be rare, but in the event they occur newer versions of HBase include the hbck tool enabled with automatic repair options.</p> </div> <div class="paragraph"> <p>There are two invariants that when violated create inconsistencies in HBase:</p> </div> <div class="ulist"> <ul> <li> <p>HBase’s region consistency invariant is satisfied if every region is assigned and deployed on exactly one region server, and all places where this state kept is in accordance.</p> </li> <li> <p>HBase’s table integrity invariant is satisfied if for each table, every possible row key resolves to exactly one region.</p> </li> </ul> </div> <div class="paragraph"> <p>Repairs generally work in three phases — a read-only information gathering phase that identifies inconsistencies, a table integrity repair phase that restores the table integrity invariant, and then finally a region consistency repair phase that restores the region consistency invariant. Starting from version 0.90.0, hbck could detect region consistency problems report on a subset of possible table integrity problems. It also included the ability to automatically fix the most common inconsistency, region assignment and deployment consistency problems. This repair could be done by using the <code>-fix</code> command line option. These problems close regions if they are open on the wrong server or on multiple region servers and also assigns regions to region servers if they are not open.</p> </div> <div class="paragraph"> <p>Starting from HBase versions 0.90.7, 0.92.2 and 0.94.0, several new command line options are introduced to aid repairing a corrupted HBase. This hbck sometimes goes by the nickname ``uberhbck''. Each particular version of uber hbck is compatible with the HBase’s of the same major version (0.90.7 uberhbck can repair a 0.90.4). However, versions ⇐0.90.6 and versions ⇐0.92.1 may require restarting the master or failing over to a backup master.</p> </div> </div> <div class="sect2"> <h3 id="_localized_repairs"><a class="anchor" href="#_localized_repairs"></a>C.3. Localized repairs</h3> <div class="paragraph"> <p>When repairing a corrupted HBase, it is best to repair the lowest risk inconsistencies first. These are generally region consistency repairs — localized single region repairs, that only modify in-memory data, ephemeral zookeeper data, or patch holes in the META table. Region consistency requires that the HBase instance has the state of the region’s data in HDFS (.regioninfo files), the region’s row in the hbase:meta table., and region’s deployment/assignments on region servers and the master in accordance. Options for repairing region consistency include:</p> </div> <div class="ulist"> <ul> <li> <p><code>-fixAssignments</code> (equivalent to the 0.90 <code>-fix</code> option) repairs unassigned, incorrectly assigned or multiply assigned regions.</p> </li> <li> <p><code>-fixMeta</code> which removes meta rows when corresponding regions are not present in HDFS and adds new meta rows if they regions are present in HDFS while not in META. To fix deployment and assignment problems you can run this command:</p> </li> </ul> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck -fixAssignments</code></pre> </div> </div> <div class="paragraph"> <p>To fix deployment and assignment problems as well as repairing incorrect meta rows you can run this command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck -fixAssignments -fixMeta</code></pre> </div> </div> <div class="paragraph"> <p>There are a few classes of table integrity problems that are low risk repairs. The first two are degenerate (startkey == endkey) regions and backwards regions (startkey > endkey). These are automatically handled by sidelining the data to a temporary directory (/hbck/xxxx). The third low-risk class is hdfs region holes. This can be repaired by using the:</p> </div> <div class="ulist"> <ul> <li> <p><code>-fixHdfsHoles</code> option for fabricating new empty regions on the file system. If holes are detected you can use -fixHdfsHoles and should include -fixMeta and -fixAssignments to make the new region consistent.</p> </li> </ul> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck -fixAssignments -fixMeta -fixHdfsHoles</code></pre> </div> </div> <div class="paragraph"> <p>Since this is a common operation, we’ve added a the <code>-repairHoles</code> flag that is equivalent to the previous command:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ ./bin/hbase hbck -repairHoles</code></pre> </div> </div> <div class="paragraph"> <p>If inconsistencies still remain after these steps, you most likely have table integrity problems related to orphaned or overlapping regions.</p> </div> </div> <div class="sect2"> <h3 id="_region_overlap_repairs"><a class="anchor" href="#_region_overlap_repairs"></a>C.4. Region Overlap Repairs</h3> <div class="paragraph"> <p>Table integrity problems can require repairs that deal with overlaps. This is a riskier operation because it requires modifications to the file system, requires some decision making, and may require some manual steps. For these repairs it is best to analyze the output of a <code>hbck -details</code> run so that you isolate repairs attempts only upon problems the checks identify. Because this is riskier, there are safeguard that should be used to limit the scope of the repairs. WARNING: This is a relatively new and have only been tested on online but idle HBase instances (no reads/writes). Use at your own risk in an active production environment! The options for repairing table integrity violations include:</p> </div> <div class="ulist"> <ul> <li> <p><code>-fixHdfsOrphans</code> option for ``adopting'' a region directory that is missing a region metadata file (the .regioninfo file).</p> </li> <li> <p><code>-fixHdfsOverlaps</code> ability for fixing overlapping regions</p> </li> </ul> </div> <div class="paragraph"> <p>When repairing overlapping regions, a region’s data can be modified on the file system in two ways: 1) by merging regions into a larger region or 2) by sidelining regions by moving data to ``sideline'' directory where data could be restored later. Merging a large number of regions is technically correct but could result in an extremely large region that requires series of costly compactions and splitting operations. In these cases, it is probably better to sideline the regions that overlap with the most other regions (likely the largest ranges) so that merges can happen on a more reasonable scale. Since these sidelined regions are already laid out in HBase’s native directory and HFile format, they can be restored by using HBase’s bulk load mechanism. The default safeguard thresholds are conservative. These options let you override the default thresholds and to enable the large region sidelining feature.</p> </div> <div class="ulist"> <ul> <li> <p><code>-maxMerge <n></code> maximum number of overlapping regions to merge</p> </li> <li> <p><code>-sidelineBigOverlaps</code> if more than maxMerge regions are overlapping, sideline attempt to sideline the regions overlapping with the most other regions.</p> </li> <li> <p><code>-maxOverlapsToSideline <n></code> if sidelining large overlapping regions, sideline at most n regions.</p> </li> </ul> </div> <div class="paragraph"> <p>Since often times you would just want to get the tables repaired, you can use this option to turn on all repair options:</p> </div> <div class="ulist"> <ul> <li> <p><code>-repair</code> includes all the region consistency options and only the hole repairing table integrity options.</p> </li> </ul> </div> <div class="paragraph"> <p>Finally, there are safeguards to limit repairs to only specific tables. For example the following command would only attempt to check and repair table TableFoo and TableBar.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase hbck -repair TableFoo TableBar</pre> </div> </div> <div class="sect3"> <h4 id="_special_cases_meta_is_not_properly_assigned"><a class="anchor" href="#_special_cases_meta_is_not_properly_assigned"></a>C.4.1. Special cases: Meta is not properly assigned</h4> <div class="paragraph"> <p>There are a few special cases that hbck can handle as well. Sometimes the meta table’s only region is inconsistently assigned or deployed. In this case there is a special <code>-fixMetaOnly</code> option that can try to fix meta assignments.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase hbck -fixMetaOnly -fixAssignments</pre> </div> </div> </div> <div class="sect3"> <h4 id="_special_cases_hbase_version_file_is_missing"><a class="anchor" href="#_special_cases_hbase_version_file_is_missing"></a>C.4.2. Special cases: HBase version file is missing</h4> <div class="paragraph"> <p>HBase’s data on the file system requires a version file in order to start. If this flie is missing, you can use the <code>-fixVersionFile</code> option to fabricating a new HBase version file. This assumes that the version of hbck you are running is the appropriate version for the HBase cluster.</p> </div> </div> <div class="sect3"> <h4 id="_special_case_root_and_meta_are_corrupt"><a class="anchor" href="#_special_case_root_and_meta_are_corrupt"></a>C.4.3. Special case: Root and META are corrupt.</h4> <div class="paragraph"> <p>The most drastic corruption scenario is the case where the ROOT or META is corrupted and HBase will not start. In this case you can use the OfflineMetaRepair tool create new ROOT and META regions and tables. This tool assumes that HBase is offline. It then marches through the existing HBase home directory, loads as much information from region metadata files (.regioninfo files) as possible from the file system. If the region metadata has proper table integrity, it sidelines the original root and meta table directories, and builds new ones with pointers to the region directories and their data.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ ./bin/hbase org.apache.hadoop.hbase.util.hbck.OfflineMetaRepair</pre> </div> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> This tool is not as clever as uberhbck but can be used to bootstrap repairs that uberhbck can complete. If the tool succeeds you should be able to start hbase and run online repairs if necessary. </td> </tr> </table> </div> </div> <div class="sect3"> <h4 id="_special_cases_offline_split_parent"><a class="anchor" href="#_special_cases_offline_split_parent"></a>C.4.4. Special cases: Offline split parent</h4> <div class="paragraph"> <p>Once a region is split, the offline parent will be cleaned up automatically. Sometimes, daughter regions are split again before their parents are cleaned up. HBase can clean up parents in the right order. However, there could be some lingering offline split parents sometimes. They are in META, in HDFS, and not deployed. But HBase can’t clean them up. In this case, you can use the <code>-fixSplitParents</code> option to reset them in META to be online and not split. Therefore, hbck can merge them with other regions if fixing overlapping regions option is used.</p> </div> <div class="paragraph"> <p>This option should not normally be used, and it is not in <code>-fixAll</code>.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="appendix_acl_matrix"><a class="anchor" href="#appendix_acl_matrix"></a>Appendix D: Access Control Matrix</h2> <div class="sectionbody"> <div class="paragraph"> <p>The following matrix shows the permission set required to perform operations in HBase. Before using the table, read through the information about how to interpret it.</p> </div> <div class="paragraph"> <div class="title">Interpreting the ACL Matrix Table</div> <p>The following conventions are used in the ACL Matrix table:</p> </div> <div class="sect2"> <h3 id="_scopes"><a class="anchor" href="#_scopes"></a>D.1. Scopes</h3> <div class="paragraph"> <p>Permissions are evaluated starting at the widest scope and working to the narrowest scope.</p> </div> <div class="paragraph"> <p>A scope corresponds to a level of the data model. From broadest to narrowest, the scopes are as follows:</p> </div> <div class="ulist"> <div class="title">Scopes</div> <ul> <li> <p>Global</p> </li> <li> <p>Namespace (NS)</p> </li> <li> <p>Table</p> </li> <li> <p>Column Family (CF)</p> </li> <li> <p>Column Qualifier (CQ)</p> </li> <li> <p>Cell</p> </li> </ul> </div> <div class="paragraph"> <p>For instance, a permission granted at table level dominates any grants done at the Column Family, Column Qualifier, or cell level. The user can do what that grant implies at any location in the table. A permission granted at global scope dominates all: the user is always allowed to take that action everywhere.</p> </div> </div> <div class="sect2"> <h3 id="_permissions"><a class="anchor" href="#_permissions"></a>D.2. Permissions</h3> <div class="paragraph"> <p>Possible permissions include the following:</p> </div> <div class="ulist"> <div class="title">Permissions</div> <ul> <li> <p>Superuser - a special user that belongs to group "supergroup" and has unlimited access</p> </li> <li> <p>Admin (A)</p> </li> <li> <p>Create (C)</p> </li> <li> <p>Write (W)</p> </li> <li> <p>Read (R)</p> </li> <li> <p>Execute (X)</p> </li> </ul> </div> <div class="paragraph"> <p>For the most part, permissions work in an expected way, with the following caveats:</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Having Write permission does not imply Read permission.</dt> <dd> <p>It is possible and sometimes desirable for a user to be able to write data that same user cannot read. One such example is a log-writing process.</p> </dd> <dt class="hdlist1">The <span class="systemitem">hbase:meta</span> table is readable by every user, regardless of the user’s other grants or restrictions.</dt> <dd> <p>This is a requirement for HBase to function correctly.</p> </dd> <dt class="hdlist1"><code>CheckAndPut</code> and <code>CheckAndDelete</code> operations will fail if the user does not have both Write and Read permission.</dt> <dt class="hdlist1"><code>Increment</code> and <code>Append</code> operations do not require Read access.</dt> <dt class="hdlist1">The <code>superuser</code>, as the name suggests has permissions to perform all possible operations.</dt> <dt class="hdlist1">And for the operations marked with *, the checks are done in post hook and only subset of results satisfying access checks are returned back to the user.</dt> <dd> <p>The following table is sorted by the interface that provides each operation. In case the table goes out of date, the unit tests which check for accuracy of permissions can be found in <em>hbase-server/src/test/java/org/apache/hadoop/hbase/security/access/TestAccessController.java</em>, and the access controls themselves can be examined in <em>hbase-server/src/main/java/org/apache/hadoop/hbase/security/access/AccessController.java</em>.</p> </dd> </dl> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 12. ACL Matrix</caption> <colgroup> <col style="width: 33%;"> <col style="width: 33%;"> <col style="width: 33%;"> </colgroup> <thead> <tr> <th class="tableblock halign-left valign-top">Interface</th> <th class="tableblock halign-left valign-top">Operation</th> <th class="tableblock halign-left valign-top">Permissions</th> </tr> </thead> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Master</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">createTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(C)|NS(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">modifyTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">deleteTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">truncateTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">addColumn</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">modifyColumn</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)|column(A)|column(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">deleteColumn</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)|column(A)|column(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">enableTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">disableTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">disableAclTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Not allowed</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">move</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">assign</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">unassign</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">regionOffline</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">balance</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">balanceSwitch</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">shutdown</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">stopMaster</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">snapshot</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">listSnapshot</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|SnapshotOwner</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">cloneSnapshot</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|(SnapshotOwner & TableName matches)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">restoreSnapshot</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|SnapshotOwner & (NS(A)|TableOwner|table(A))</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">deleteSnapshot</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|SnapshotOwner</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">createNamespace</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">deleteNamespace</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">modifyNamespace</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getNamespaceDescriptor</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">listNamespaceDescriptors*</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">flushTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getTableDescriptors*</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|NS(A)|NS(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getTableNames*</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|TableOwner|Any global or table perm</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">setUserQuota(global level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">setUserQuota(namespace level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">setUserQuota(Table level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">setTableQuota</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|NS(A)|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">setNamespaceQuota</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Region</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">openRegion</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">closeRegion</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">flush</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">split</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|TableOwner|TableOwner|table(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">compact</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)|global(C)|TableOwner|table(A)|table(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getClosestRowBefore</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getOp</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">exists</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">put</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|table(W)|TableOwner|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">delete</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|table(W)|TableOwner|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">batchMutate</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|TableOwner|table(W)|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">checkAndPut</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(RW)|NS(RW)|TableOwner|table(RW)|CF(RW)|CQ(RW)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">checkAndPutAfterRowLock</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|Table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">checkAndDelete</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(RW)|NS(RW)|TableOwner|table(RW)|CF(RW)|CQ(RW)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">checkAndDeleteAfterRowLock</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">incrementColumnValue</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|TableOwner|table(W)|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">append</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|TableOwner|table(W)|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">appendAfterRowLock</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|TableOwner|table(W)|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">increment</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|TableOwner|table(W)|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">incrementAfterRowLock</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)|NS(W)|TableOwner|table(W)|CF(W)|CQ(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">scannerOpen</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">scannerNext</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">scannerClose</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(R)|NS(R)|TableOwner|table(R)|CF(R)|CQ(R)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">bulkLoadHFile</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(C)|TableOwner|table(C)|CF(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">prepareBulkLoad</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(C)|TableOwner|table(C)|CF(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">cleanupBulkLoad</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(C)|TableOwner|table(C)|CF(C)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Endpoint</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">invoke</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(X)|NS(X)|TableOwner|table(X)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">AccessController</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">grant(global level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">grant(namespace level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)|NS(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">grant(table level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)|NS(A)|TableOwner|table(A)|CF(A)|CQ(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">revoke(global level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">revoke(namespace level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)|NS(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">revoke(table level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)|NS(A)|TableOwner|table(A)|CF(A)|CQ(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getUserPermissions(global level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getUserPermissions(namespace level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)|NS(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getUserPermissions(table level)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">global(A)|NS(A)|TableOwner|table(A)|CF(A)|CQ(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">RegionServer</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">stopRegionServer</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">mergeRegions</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">rollWALWriterRequest</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">replicateLogEntries</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(W)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">RSGroup</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">addRSGroup</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">balanceRSGroup</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getRSGroupInfo</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getRSGroupInfoOfTable</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">getRSGroupOfServer</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">listRSGroups</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">moveServers</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">moveServersAndTables</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">moveTables</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">removeRSGroup</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">removeServers</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">superuser|global(A)</p></td> </tr> </tbody> </table> </div> </div> </div> <div class="sect1"> <h2 id="compression"><a class="anchor" href="#compression"></a>Appendix E: Compression and Data Block Encoding In HBase</h2> <div class="sectionbody"> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> Codecs mentioned in this section are for encoding and decoding data blocks or row keys. For information about replication codecs, see <a href="#cluster.replication.preserving.tags">cluster.replication.preserving.tags</a>. </td> </tr> </table> </div> <div class="paragraph"> <p>Some of the information in this section is pulled from a <a href="http://search-hadoop.com/m/lL12B1PFVhp1/v=threaded">discussion</a> on the HBase Development mailing list.</p> </div> <div class="paragraph"> <p>HBase supports several different compression algorithms which can be enabled on a ColumnFamily. Data block encoding attempts to limit duplication of information in keys, taking advantage of some of the fundamental designs and patterns of HBase, such as sorted row keys and the schema of a given table. Compressors reduce the size of large, opaque byte arrays in cells, and can significantly reduce the storage space needed to store uncompressed data.</p> </div> <div class="paragraph"> <p>Compressors and data block encoding can be used together on the same ColumnFamily.</p> </div> <div class="paragraph"> <div class="title">Changes Take Effect Upon Compaction</div> <p>If you change compression or encoding for a ColumnFamily, the changes take effect during compaction.</p> </div> <div class="paragraph"> <p>Some codecs take advantage of capabilities built into Java, such as GZip compression. Others rely on native libraries. Native libraries may be available as part of Hadoop, such as LZ4. In this case, HBase only needs access to the appropriate shared library.</p> </div> <div class="paragraph"> <p>Other codecs, such as Google Snappy, need to be installed first. Some codecs are licensed in ways that conflict with HBase’s license and cannot be shipped as part of HBase.</p> </div> <div class="paragraph"> <p>This section discusses common codecs that are used and tested with HBase. No matter what codec you use, be sure to test that it is installed correctly and is available on all nodes in your cluster. Extra operational steps may be necessary to be sure that codecs are available on newly-deployed nodes. You can use the <a href="#compression.test">compression.test</a> utility to check that a given codec is correctly installed.</p> </div> <div class="paragraph"> <p>To configure HBase to use a compressor, see <a href="#compressor.install">compressor.install</a>. To enable a compressor for a ColumnFamily, see <a href="#changing.compression">changing.compression</a>. To enable data block encoding for a ColumnFamily, see <a href="#data.block.encoding.enable">data.block.encoding.enable</a>.</p> </div> <div class="ulist"> <div class="title">Block Compressors</div> <ul> <li> <p>none</p> </li> <li> <p>Snappy</p> </li> <li> <p>LZO</p> </li> <li> <p>LZ4</p> </li> <li> <p>GZ</p> </li> </ul> </div> <div class="dlist"> <div class="title">Data Block Encoding Types</div> <dl> <dt class="hdlist1">Prefix</dt> <dd> <p> Often, keys are very similar. Specifically, keys often share a common prefix and only differ near the end. For instance, one key might be <code>RowKey:Family:Qualifier0</code> and the next key might be <code>RowKey:Family:Qualifier1</code>. <br> In Prefix encoding, an extra column is added which holds the length of the prefix shared between the current key and the previous key. Assuming the first key here is totally different from the key before, its prefix length is 0.</p> <div class="paragraph"> <p>The second key’s prefix length is <code>23</code>, since they have the first 23 characters in common.</p> </div> <div class="paragraph"> <p>Obviously if the keys tend to have nothing in common, Prefix will not provide much benefit.</p> </div> <div class="paragraph"> <p>The following image shows a hypothetical ColumnFamily with no data block encoding.</p> </div> <div class="imageblock"> <div class="content"> <img src="images/data_block_no_encoding.png" alt="data block no encoding"> </div> <div class="title">Figure 11. ColumnFamily with No Encoding</div> </div> <div class="paragraph"> <p>Here is the same data with prefix data encoding.</p> </div> <div class="imageblock"> <div class="content"> <img src="images/data_block_prefix_encoding.png" alt="data block prefix encoding"> </div> <div class="title">Figure 12. ColumnFamily with Prefix Encoding</div> </div> </dd> <dt class="hdlist1">Diff</dt> <dd> <p>Diff encoding expands upon Prefix encoding. Instead of considering the key sequentially as a monolithic series of bytes, each key field is split so that each part of the key can be compressed more efficiently.</p> <div class="paragraph"> <p>Two new fields are added: timestamp and type.</p> </div> <div class="paragraph"> <p>If the ColumnFamily is the same as the previous row, it is omitted from the current row.</p> </div> <div class="paragraph"> <p>If the key length, value length or type are the same as the previous row, the field is omitted.</p> </div> <div class="paragraph"> <p>In addition, for increased compression, the timestamp is stored as a Diff from the previous row’s timestamp, rather than being stored in full. Given the two row keys in the Prefix example, and given an exact match on timestamp and the same type, neither the value length, or type needs to be stored for the second row, and the timestamp value for the second row is just 0, rather than a full timestamp.</p> </div> <div class="paragraph"> <p>Diff encoding is disabled by default because writing and scanning are slower but more data is cached.</p> </div> <div class="paragraph"> <p>This image shows the same ColumnFamily from the previous images, with Diff encoding.</p> </div> <div class="imageblock"> <div class="content"> <img src="images/data_block_diff_encoding.png" alt="data block diff encoding"> </div> <div class="title">Figure 13. ColumnFamily with Diff Encoding</div> </div> </dd> <dt class="hdlist1">Fast Diff</dt> <dd> <p>Fast Diff works similar to Diff, but uses a faster implementation. It also adds another field which stores a single bit to track whether the data itself is the same as the previous row. If it is, the data is not stored again.</p> <div class="paragraph"> <p>Fast Diff is the recommended codec to use if you have long keys or many columns.</p> </div> <div class="paragraph"> <p>The data format is nearly identical to Diff encoding, so there is not an image to illustrate it.</p> </div> </dd> <dt class="hdlist1">Prefix Tree</dt> <dd> <p>Prefix tree encoding was introduced as an experimental feature in HBase 0.96. It provides similar memory savings to the Prefix, Diff, and Fast Diff encoder, but provides faster random access at a cost of slower encoding speed.</p> </dd> </dl> </div> <div class="paragraph"> <p>Prefix Tree may be appropriate for applications that have high block cache hit ratios. It introduces new 'tree' fields for the row and column. The row tree field contains a list of offsets/references corresponding to the cells in that row. This allows for a good deal of compression. For more details about Prefix Tree encoding, see <a href="https://issues.apache.org/jira/browse/HBASE-4676">HBASE-4676</a>.</p> </div> <div class="paragraph"> <p>It is difficult to graphically illustrate a prefix tree, so no image is included. See the Wikipedia article for <a href="http://en.wikipedia.org/wiki/Trie">Trie</a> for more general information about this data structure.</p> </div> <div class="paragraph"> <p>Prefix Tree will be removed in hbase-2.0.0. It is a sweet feature but has seen little uptake and is not actively maintained. Come and write the dev list if you are interesting in carrying-on this encoding.</p> </div> <div class="sect2"> <h3 id="_which_compressor_or_data_block_encoder_to_use"><a class="anchor" href="#_which_compressor_or_data_block_encoder_to_use"></a>E.1. Which Compressor or Data Block Encoder To Use</h3> <div class="paragraph"> <p>The compression or codec type to use depends on the characteristics of your data. Choosing the wrong type could cause your data to take more space rather than less, and can have performance implications.</p> </div> <div class="paragraph"> <p>In general, you need to weigh your options between smaller size and faster compression/decompression. Following are some general guidelines, expanded from a discussion at <a href="http://search-hadoop.com/m/lL12B1PFVhp1">Documenting Guidance on compression and codecs</a>.</p> </div> <div class="ulist"> <ul> <li> <p>If you have long keys (compared to the values) or many columns, use a prefix encoder. FAST_DIFF is recommended, as more testing is needed for Prefix Tree encoding.</p> </li> <li> <p>If the values are large (and not precompressed, such as images), use a data block compressor.</p> </li> <li> <p>Use GZIP for <em class="firstterm">cold data</em>, which is accessed infrequently. GZIP compression uses more CPU resources than Snappy or LZO, but provides a higher compression ratio.</p> </li> <li> <p>Use Snappy or LZO for <em class="firstterm">hot data</em>, which is accessed frequently. Snappy and LZO use fewer CPU resources than GZIP, but do not provide as high of a compression ratio.</p> </li> <li> <p>In most cases, enabling Snappy or LZO by default is a good choice, because they have a low performance overhead and provide space savings.</p> </li> <li> <p>Before Snappy became available by Google in 2011, LZO was the default. Snappy has similar qualities as LZO but has been shown to perform better.</p> </li> </ul> </div> </div> <div class="sect2"> <h3 id="hadoop.native.lib"><a class="anchor" href="#hadoop.native.lib"></a>E.2. Making use of Hadoop Native Libraries in HBase</h3> <div class="paragraph"> <p>The Hadoop shared library has a bunch of facility including compression libraries and fast crc’ing. To make this facility available to HBase, do the following. HBase/Hadoop will fall back to use alternatives if it cannot find the native library versions — or fail outright if you asking for an explicit compressor and there is no alternative available.</p> </div> <div class="paragraph"> <p>If you see the following in your HBase logs, you know that HBase was unable to locate the Hadoop native libraries:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="integer">2014</span>-<span class="integer">08</span>-<span class="octal">07</span> <span class="integer">09</span>:<span class="integer">26</span>:<span class="integer">20</span>,<span class="integer">139</span> WARN [main] util.NativeCodeLoader: Unable to load <span class="directive">native</span>-hadoop library <span class="keyword">for</span> your platform... using builtin-java classes where applicable</code></pre> </div> </div> <div class="paragraph"> <p>If the libraries loaded successfully, the WARN message does not show.</p> </div> <div class="paragraph"> <p>Lets presume your Hadoop shipped with a native library that suits the platform you are running HBase on. To check if the Hadoop native library is available to HBase, run the following tool (available in Hadoop 2.1 and greater):</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="error">$</span> ./bin/hbase --config ~/conf_hbase org.apache.hadoop.util.NativeLibraryChecker <span class="integer">2014</span>-<span class="integer">08</span>-<span class="integer">26</span> <span class="integer">13</span>:<span class="integer">15</span>:<span class="integer">38</span>,<span class="integer">717</span> WARN [main] util.NativeCodeLoader: Unable to load <span class="directive">native</span>-hadoop library <span class="keyword">for</span> your platform... using builtin-java classes where applicable Native library checking: hadoop: <span class="predefined-constant">false</span> zlib: <span class="predefined-constant">false</span> snappy: <span class="predefined-constant">false</span> lz4: <span class="predefined-constant">false</span> bzip2: <span class="predefined-constant">false</span> <span class="integer">2014</span>-<span class="integer">08</span>-<span class="integer">26</span> <span class="integer">13</span>:<span class="integer">15</span>:<span class="integer">38</span>,<span class="integer">863</span> INFO [main] util.ExitUtil: Exiting with status <span class="integer">1</span></code></pre> </div> </div> <div class="paragraph"> <p>Above shows that the native hadoop library is not available in HBase context.</p> </div> <div class="paragraph"> <p>To fix the above, either copy the Hadoop native libraries local or symlink to them if the Hadoop and HBase stalls are adjacent in the filesystem. You could also point at their location by setting the <code>LD_LIBRARY_PATH</code> environment variable.</p> </div> <div class="paragraph"> <p>Where the JVM looks to find native librarys is "system dependent" (See <code>java.lang.System#loadLibrary(name)</code>). On linux, by default, is going to look in <em>lib/native/PLATFORM</em> where <code>PLATFORM</code> is the label for the platform your HBase is installed on. On a local linux machine, it seems to be the concatenation of the java properties <code>os.name</code> and <code>os.arch</code> followed by whether 32 or 64 bit. HBase on startup prints out all of the java system properties so find the os.name and os.arch in the log. For example:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">... <span class="integer">2014</span>-<span class="integer">08</span>-<span class="octal">06</span> <span class="integer">15</span>:<span class="integer">27</span>:<span class="integer">22</span>,<span class="integer">853</span> INFO [main] zookeeper.ZooKeeper: Client environment:os.name=Linux <span class="integer">2014</span>-<span class="integer">08</span>-<span class="octal">06</span> <span class="integer">15</span>:<span class="integer">27</span>:<span class="integer">22</span>,<span class="integer">853</span> INFO [main] zookeeper.ZooKeeper: Client environment:os.arch=amd64 ...</code></pre> </div> </div> <div class="paragraph"> <p>So in this case, the PLATFORM string is <code>Linux-amd64-64</code>. Copying the Hadoop native libraries or symlinking at <em>lib/native/Linux-amd64-64</em> will ensure they are found. Check with the Hadoop <em>NativeLibraryChecker</em>.</p> </div> <div class="paragraph"> <p>Here is example of how to point at the Hadoop libs with <code>LD_LIBRARY_PATH</code> environment variable:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="error">$</span> LD_LIBRARY_PATH=~/hadoop-<span class="float">2.5</span><span class="float">.0</span>-SNAPSHOT/lib/<span class="directive">native</span> ./bin/hbase --config ~/conf_hbase org.apache.hadoop.util.NativeLibraryChecker <span class="integer">2014</span>-<span class="integer">08</span>-<span class="integer">26</span> <span class="integer">13</span>:<span class="integer">42</span>:<span class="integer">49</span>,<span class="integer">332</span> INFO [main] bzip2.Bzip2Factory: Successfully loaded & initialized <span class="directive">native</span>-bzip2 library system-<span class="directive">native</span> <span class="integer">2014</span>-<span class="integer">08</span>-<span class="integer">26</span> <span class="integer">13</span>:<span class="integer">42</span>:<span class="integer">49</span>,<span class="integer">337</span> INFO [main] zlib.ZlibFactory: Successfully loaded & initialized <span class="directive">native</span>-zlib library Native library checking: hadoop: <span class="predefined-constant">true</span> /home/stack/hadoop-<span class="float">2.5</span><span class="float">.0</span>-SNAPSHOT/lib/<span class="directive">native</span>/libhadoop.so<span class="float">.1</span><span class="float">.0</span><span class="float">.0</span> zlib: <span class="predefined-constant">true</span> /lib64/libz.so<span class="float">.1</span> snappy: <span class="predefined-constant">true</span> /usr/lib64/libsnappy.so<span class="float">.1</span> lz4: <span class="predefined-constant">true</span> revision:<span class="integer">99</span> bzip2: <span class="predefined-constant">true</span> /lib64/libbz2.so<span class="float">.1</span></code></pre> </div> </div> <div class="paragraph"> <p>Set in <em>hbase-env.sh</em> the LD_LIBRARY_PATH environment variable when starting your HBase.</p> </div> </div> <div class="sect2"> <h3 id="_compressor_configuration_installation_and_use"><a class="anchor" href="#_compressor_configuration_installation_and_use"></a>E.3. Compressor Configuration, Installation, and Use</h3> <div class="sect3"> <h4 id="compressor.install"><a class="anchor" href="#compressor.install"></a>E.3.1. Configure HBase For Compressors</h4> <div class="paragraph"> <p>Before HBase can use a given compressor, its libraries need to be available. Due to licensing issues, only GZ compression is available to HBase (via native Java libraries) in a default installation. Other compression libraries are available via the shared library bundled with your hadoop. The hadoop native library needs to be findable when HBase starts. See</p> </div> <div class="paragraph"> <div class="title">Compressor Support On the Master</div> <p>A new configuration setting was introduced in HBase 0.95, to check the Master to determine which data block encoders are installed and configured on it, and assume that the entire cluster is configured the same. This option, <code>hbase.master.check.compression</code>, defaults to <code>true</code>. This prevents the situation described in <a href="https://issues.apache.org/jira/browse/HBASE-6370">HBASE-6370</a>, where a table is created or modified to support a codec that a region server does not support, leading to failures that take a long time to occur and are difficult to debug.</p> </div> <div class="paragraph"> <p>If <code>hbase.master.check.compression</code> is enabled, libraries for all desired compressors need to be installed and configured on the Master, even if the Master does not run a region server.</p> </div> <div class="paragraph"> <div class="title">Install GZ Support Via Native Libraries</div> <p>HBase uses Java’s built-in GZip support unless the native Hadoop libraries are available on the CLASSPATH. The recommended way to add libraries to the CLASSPATH is to set the environment variable <code>HBASE_LIBRARY_PATH</code> for the user running HBase. If native libraries are not available and Java’s GZIP is used, <code>Got brand-new compressor</code> reports will be present in the logs. See <a href="#brand.new.compressor">brand.new.compressor</a>).</p> </div> <div id="lzo.compression" class="paragraph"> <div class="title">Install LZO Support</div> <p>HBase cannot ship with LZO because of incompatibility between HBase, which uses an Apache Software License (ASL) and LZO, which uses a GPL license. See the <a href="http://wiki.apache.org/hadoop/UsingLzoCompression">Using LZO Compression</a> wiki page for information on configuring LZO support for HBase.</p> </div> <div class="paragraph"> <p>If you depend upon LZO compression, consider configuring your RegionServers to fail to start if LZO is not available. See <a href="#hbase.regionserver.codecs">hbase.regionserver.codecs</a>.</p> </div> <div id="lz4.compression" class="paragraph"> <div class="title">Configure LZ4 Support</div> <p>LZ4 support is bundled with Hadoop. Make sure the hadoop shared library (libhadoop.so) is accessible when you start HBase. After configuring your platform (see <a href="#hbase.native.platform">hbase.native.platform</a>), you can make a symbolic link from HBase to the native Hadoop libraries. This assumes the two software installs are colocated. For example, if my 'platform' is Linux-amd64-64:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="bourne">$ cd $HBASE_HOME $ mkdir lib/native $ ln -s $HADOOP_HOME/lib/native lib/native/Linux-amd64-64</code></pre> </div> </div> <div class="paragraph"> <p>Use the compression tool to check that LZ4 is installed on all nodes. Start up (or restart) HBase. Afterward, you can create and alter tables to enable LZ4 as a compression codec.:</p> </div> <div class="listingblock"> <div class="content"> <pre>hbase(main):003:0> alter 'TestTable', {NAME => 'info', COMPRESSION => 'LZ4'}</pre> </div> </div> <div id="snappy.compression.installation" class="paragraph"> <div class="title">Install Snappy Support</div> <p>HBase does not ship with Snappy support because of licensing issues. You can install Snappy binaries (for instance, by using yum install snappy on CentOS) or build Snappy from source. After installing Snappy, search for the shared library, which will be called <em>libsnappy.so.X</em> where X is a number. If you built from source, copy the shared library to a known location on your system, such as <em>/opt/snappy/lib/</em>.</p> </div> <div class="paragraph"> <p>In addition to the Snappy library, HBase also needs access to the Hadoop shared library, which will be called something like <em>libhadoop.so.X.Y</em>, where X and Y are both numbers. Make note of the location of the Hadoop library, or copy it to the same location as the Snappy library.</p> </div> <div class="admonitionblock note"> <table> <tr> <td class="icon"> <i class="fa icon-note" title="Note"></i> </td> <td class="content"> <div class="paragraph"> <p>The Snappy and Hadoop libraries need to be available on each node of your cluster. See <a href="#compression.test">compression.test</a> to find out how to test that this is the case.</p> </div> <div class="paragraph"> <p>See <a href="#hbase.regionserver.codecs">hbase.regionserver.codecs</a> to configure your RegionServers to fail to start if a given compressor is not available.</p> </div> </td> </tr> </table> </div> <div class="paragraph"> <p>Each of these library locations need to be added to the environment variable <code>HBASE_LIBRARY_PATH</code> for the operating system user that runs HBase. You need to restart the RegionServer for the changes to take effect.</p> </div> <div id="compression.test" class="paragraph"> <div class="title">CompressionTest</div> <p>You can use the CompressionTest tool to verify that your compressor is available to HBase:</p> </div> <div class="listingblock"> <div class="content"> <pre> $ hbase org.apache.hadoop.hbase.util.CompressionTest hdfs://host/path/to/hbase snappy</pre> </div> </div> <div id="hbase.regionserver.codecs" class="paragraph"> <div class="title">Enforce Compression Settings On a RegionServer</div> <p>You can configure a RegionServer so that it will fail to restart if compression is configured incorrectly, by adding the option hbase.regionserver.codecs to the <em>hbase-site.xml</em>, and setting its value to a comma-separated list of codecs that need to be available. For example, if you set this property to <code>lzo,gz</code>, the RegionServer would fail to start if both compressors were not available. This would prevent a new server from being added to the cluster without having codecs configured properly.</p> </div> </div> <div class="sect3"> <h4 id="changing.compression"><a class="anchor" href="#changing.compression"></a>E.3.2. Enable Compression On a ColumnFamily</h4> <div class="paragraph"> <p>To enable compression for a ColumnFamily, use an <code>alter</code> command. You do not need to re-create the table or copy data. If you are changing codecs, be sure the old codec is still available until all the old StoreFiles have been compacted.</p> </div> <div class="exampleblock"> <div class="title">Example 52. Enabling Compression on a ColumnFamily of an Existing Table using HBaseShell</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> disable 'test' hbase> alter 'test', {NAME => 'cf', COMPRESSION => 'GZ'} hbase> enable 'test'</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 53. Creating a New Table with Compression On a ColumnFamily</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> create 'test2', { NAME => 'cf2', COMPRESSION => 'SNAPPY' }</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 54. Verifying a ColumnFamily’s Compression Settings</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> describe 'test' DESCRIPTION ENABLED 'test', {NAME => 'cf', DATA_BLOCK_ENCODING => 'NONE false ', BLOOMFILTER => 'ROW', REPLICATION_SCOPE => '0', VERSIONS => '1', COMPRESSION => 'GZ', MIN_VERSIONS => '0', TTL => 'FOREVER', KEEP_DELETED_CELLS => 'fa lse', BLOCKSIZE => '65536', IN_MEMORY => 'false', B LOCKCACHE => 'true'} 1 row(s) in 0.1070 seconds</pre> </div> </div> </div> </div> </div> <div class="sect3"> <h4 id="_testing_compression_performance"><a class="anchor" href="#_testing_compression_performance"></a>E.3.3. Testing Compression Performance</h4> <div class="paragraph"> <p>HBase includes a tool called LoadTestTool which provides mechanisms to test your compression performance. You must specify either <code>-write</code> or <code>-update-read</code> as your first parameter, and if you do not specify another parameter, usage advice is printed for each option.</p> </div> <div class="exampleblock"> <div class="title">Example 55. LoadTestTool Usage</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>$ bin/hbase org.apache.hadoop.hbase.util.LoadTestTool -h usage: bin/hbase org.apache.hadoop.hbase.util.LoadTestTool <options> Options: -batchupdate Whether to use batch as opposed to separate updates for every column in a row -bloom <arg> Bloom filter type, one of [NONE, ROW, ROWCOL] -compression <arg> Compression type, one of [LZO, GZ, NONE, SNAPPY, LZ4] -data_block_encoding <arg> Encoding algorithm (e.g. prefix compression) to use for data blocks in the test column family, one of [NONE, PREFIX, DIFF, FAST_DIFF, PREFIX_TREE]. -encryption <arg> Enables transparent encryption on the test table, one of [AES] -generator <arg> The class which generates load for the tool. Any args for this class can be passed as colon separated after class name -h,--help Show usage -in_memory Tries to keep the HFiles of the CF inmemory as far as possible. Not guaranteed that reads are always served from inmemory -init_only Initialize the test table only, don't do any loading -key_window <arg> The 'key window' to maintain between reads and writes for concurrent write/read workload. The default is 0. -max_read_errors <arg> The maximum number of read errors to tolerate before terminating all reader threads. The default is 10. -multiput Whether to use multi-puts as opposed to separate puts for every column in a row -num_keys <arg> The number of keys to read/write -num_tables <arg> A positive integer number. When a number n is speicfied, load test tool will load n table parallely. -tn parameter value becomes table name prefix. Each table name is in format <tn>_1...<tn>_n -read <arg> <verify_percent>[:<#threads=20>] -regions_per_server <arg> A positive integer number. When a number n is specified, load test tool will create the test table with n regions per server -skip_init Skip the initialization; assume test table already exists -start_key <arg> The first key to read/write (a 0-based index). The default value is 0. -tn <arg> The name of the table to read or write -update <arg> <update_percent>[:<#threads=20>][:<#whether to ignore nonce collisions=0>] -write <arg> <avg_cols_per_key>:<avg_data_size>[:<#threads=20>] -zk <arg> ZK quorum as comma-separated host names without port numbers -zk_root <arg> name of parent znode in zookeeper</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 56. Example Usage of LoadTestTool</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>$ hbase org.apache.hadoop.hbase.util.LoadTestTool -write 1:10:100 -num_keys 1000000 -read 100:30 -num_tables 1 -data_block_encoding NONE -tn load_test_tool_NONE</pre> </div> </div> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="data.block.encoding.enable"><a class="anchor" href="#data.block.encoding.enable"></a>154. Enable Data Block Encoding</h2> <div class="sectionbody"> <div class="paragraph"> <p>Codecs are built into HBase so no extra configuration is needed. Codecs are enabled on a table by setting the <code>DATA_BLOCK_ENCODING</code> property. Disable the table before altering its DATA_BLOCK_ENCODING setting. Following is an example using HBase Shell:</p> </div> <div class="exampleblock"> <div class="title">Example 57. Enable Data Block Encoding On a Table</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> disable 'test' hbase> alter 'test', { NAME => 'cf', DATA_BLOCK_ENCODING => 'FAST_DIFF' } Updating all regions with the new schema... 0/1 regions updated. 1/1 regions updated. Done. 0 row(s) in 2.2820 seconds hbase> enable 'test' 0 row(s) in 0.1580 seconds</pre> </div> </div> </div> </div> <div class="exampleblock"> <div class="title">Example 58. Verifying a ColumnFamily’s Data Block Encoding</div> <div class="content"> <div class="listingblock"> <div class="content"> <pre>hbase> describe 'test' DESCRIPTION ENABLED 'test', {NAME => 'cf', DATA_BLOCK_ENCODING => 'FAST true _DIFF', BLOOMFILTER => 'ROW', REPLICATION_SCOPE => '0', VERSIONS => '1', COMPRESSION => 'GZ', MIN_VERS IONS => '0', TTL => 'FOREVER', KEEP_DELETED_CELLS = > 'false', BLOCKSIZE => '65536', IN_MEMORY => 'fals e', BLOCKCACHE => 'true'} 1 row(s) in 0.0650 seconds</pre> </div> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="sql"><a class="anchor" href="#sql"></a>Appendix F: SQL over HBase</h2> <div class="sectionbody"> <div class="paragraph"> <p>The following projects offer some support for SQL over HBase.</p> </div> <div class="sect2"> <h3 id="phoenix"><a class="anchor" href="#phoenix"></a>F.1. Apache Phoenix</h3> <div class="paragraph"> <p><a href="http://phoenix.apache.org">Apache Phoenix</a></p> </div> </div> <div class="sect2"> <h3 id="_trafodion"><a class="anchor" href="#_trafodion"></a>F.2. Trafodion</h3> <div class="paragraph"> <p><a href="https://wiki.trafodion.org/">Trafodion: Transactional SQL-on-HBase</a></p> </div> </div> </div> </div> <div class="sect1"> <h2 id="_ycsb"><a class="anchor" href="#_ycsb"></a>Appendix G: YCSB</h2> <div class="sectionbody"> <div class="paragraph"> <p><a href="https://github.com/brianfrankcooper/YCSB/">YCSB: The Yahoo! Cloud Serving Benchmark</a> and HBase</p> </div> <div class="paragraph"> <p>TODO: Describe how YCSB is poor for putting up a decent cluster load.</p> </div> <div class="paragraph"> <p>TODO: Describe setup of YCSB for HBase. In particular, presplit your tables before you start a run. See <a href="https://issues.apache.org/jira/browse/HBASE-4163">HBASE-4163 Create Split Strategy for YCSB Benchmark</a> for why and a little shell command for how to do it.</p> </div> <div class="paragraph"> <p>Ted Dunning redid YCSB so it’s mavenized and added facility for verifying workloads. See <a href="https://github.com/tdunning/YCSB">Ted Dunning’s YCSB</a>.</p> </div> </div> </div> <div class="sect1"> <h2 id="_hfile_format_2"><a class="anchor" href="#_hfile_format_2"></a>Appendix H: HFile format</h2> <div class="sectionbody"> <div class="paragraph"> <p>This appendix describes the evolution of the HFile format.</p> </div> <div class="sect2"> <h3 id="hfilev1"><a class="anchor" href="#hfilev1"></a>H.1. HBase File Format (version 1)</h3> <div class="paragraph"> <p>As we will be discussing changes to the HFile format, it is useful to give a short overview of the original (HFile version 1) format.</p> </div> <div class="sect3"> <h4 id="hfilev1.overview"><a class="anchor" href="#hfilev1.overview"></a>H.1.1. Overview of Version 1</h4> <div class="paragraph"> <p>An HFile in version 1 format is structured as follows:</p> </div> <div class="imageblock"> <div class="content"> <img src="images/hfile.png" alt="HFile Version 1"> </div> <div class="title">Figure 14. HFile V1 Format</div> </div> </div> <div class="sect3"> <h4 id="_block_index_format_in_version_1"><a class="anchor" href="#_block_index_format_in_version_1"></a>H.1.2. Block index format in version 1</h4> <div class="paragraph"> <p>The block index in version 1 is very straightforward. For each entry, it contains:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Offset (long)</p> </li> <li> <p>Uncompressed size (int)</p> </li> <li> <p>Key (a serialized byte array written using Bytes.writeByteArray)</p> <div class="olist loweralpha"> <ol class="loweralpha" type="a"> <li> <p>Key length as a variable-length integer (VInt)</p> </li> <li> <p>Key bytes</p> </li> </ol> </div> </li> </ol> </div> <div class="paragraph"> <p>The number of entries in the block index is stored in the fixed file trailer, and has to be passed in to the method that reads the block index. One of the limitations of the block index in version 1 is that it does not provide the compressed size of a block, which turns out to be necessary for decompression. Therefore, the HFile reader has to infer this compressed size from the offset difference between blocks. We fix this limitation in version 2, where we store on-disk block size instead of uncompressed size, and get uncompressed size from the block header.</p> </div> </div> </div> <div class="sect2"> <h3 id="hfilev2"><a class="anchor" href="#hfilev2"></a>H.2. HBase file format with inline blocks (version 2)</h3> <div class="paragraph"> <p>Note: this feature was introduced in HBase 0.92</p> </div> <div class="sect3"> <h4 id="_motivation"><a class="anchor" href="#_motivation"></a>H.2.1. Motivation</h4> <div class="paragraph"> <p>We found it necessary to revise the HFile format after encountering high memory usage and slow startup times caused by large Bloom filters and block indexes in the region server. Bloom filters can get as large as 100 MB per HFile, which adds up to 2 GB when aggregated over 20 regions. Block indexes can grow as large as 6 GB in aggregate size over the same set of regions. A region is not considered opened until all of its block index data is loaded. Large Bloom filters produce a different performance problem: the first get request that requires a Bloom filter lookup will incur the latency of loading the entire Bloom filter bit array.</p> </div> <div class="paragraph"> <p>To speed up region server startup we break Bloom filters and block indexes into multiple blocks and write those blocks out as they fill up, which also reduces the HFile writer’s memory footprint. In the Bloom filter case, "filling up a block" means accumulating enough keys to efficiently utilize a fixed-size bit array, and in the block index case we accumulate an "index block" of the desired size. Bloom filter blocks and index blocks (we call these "inline blocks") become interspersed with data blocks, and as a side effect we can no longer rely on the difference between block offsets to determine data block length, as it was done in version 1.</p> </div> <div class="paragraph"> <p>HFile is a low-level file format by design, and it should not deal with application-specific details such as Bloom filters, which are handled at StoreFile level. Therefore, we call Bloom filter blocks in an HFile "inline" blocks. We also supply HFile with an interface to write those inline blocks.</p> </div> <div class="paragraph"> <p>Another format modification aimed at reducing the region server startup time is to use a contiguous "load-on-open" section that has to be loaded in memory at the time an HFile is being opened. Currently, as an HFile opens, there are separate seek operations to read the trailer, data/meta indexes, and file info. To read the Bloom filter, there are two more seek operations for its "data" and "meta" portions. In version 2, we seek once to read the trailer and seek again to read everything else we need to open the file from a contiguous block.</p> </div> </div> <div class="sect3"> <h4 id="hfilev2.overview"><a class="anchor" href="#hfilev2.overview"></a>H.2.2. Overview of Version 2</h4> <div class="paragraph"> <p>The version of HBase introducing the above features reads both version 1 and 2 HFiles, but only writes version 2 HFiles. A version 2 HFile is structured as follows:</p> </div> <div class="paragraph"> <div class="title">HFile Version 2 Structure</div> <p><span class="image"><img src="images/hfilev2.png" alt="HFile Version 2"></span></p> </div> </div> <div class="sect3"> <h4 id="_unified_version_2_block_format"><a class="anchor" href="#_unified_version_2_block_format"></a>H.2.3. Unified version 2 block format</h4> <div class="paragraph"> <p>In the version 2 every block in the data section contains the following fields:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>8 bytes: Block type, a sequence of bytes equivalent to version 1’s "magic records". Supported block types are:</p> <div class="olist loweralpha"> <ol class="loweralpha" type="a"> <li> <p>DATA – data blocks</p> </li> <li> <p>LEAF_INDEX – leaf-level index blocks in a multi-level-block-index</p> </li> <li> <p>BLOOM_CHUNK – Bloom filter chunks</p> </li> <li> <p>META – meta blocks (not used for Bloom filters in version 2 anymore)</p> </li> <li> <p>INTERMEDIATE_INDEX – intermediate-level index blocks in a multi-level blockindex</p> </li> <li> <p>ROOT_INDEX – root>level index blocks in a multi>level block index</p> </li> <li> <p>FILE_INFO – the ``file info'' block, a small key>value map of metadata</p> </li> <li> <p>BLOOM_META – a Bloom filter metadata block in the load>on>open section</p> </li> <li> <p>TRAILER – a fixed>size file trailer. As opposed to the above, this is not an HFile v2 block but a fixed>size (for each HFile version) data structure</p> </li> <li> <p>INDEX_V1 – this block type is only used for legacy HFile v1 block</p> </li> </ol> </div> </li> <li> <p>Compressed size of the block’s data, not including the header (int).</p> <div class="paragraph"> <p>Can be used for skipping the current data block when scanning HFile data.</p> </div> </li> <li> <p>Uncompressed size of the block’s data, not including the header (int)</p> <div class="paragraph"> <p>This is equal to the compressed size if the compression algorithm is NONE</p> </div> </li> <li> <p>File offset of the previous block of the same type (long)</p> <div class="paragraph"> <p>Can be used for seeking to the previous data/index block</p> </div> </li> <li> <p>Compressed data (or uncompressed data if the compression algorithm is NONE).</p> </li> </ol> </div> <div class="paragraph"> <p>The above format of blocks is used in the following HFile sections:</p> </div> <div class="dlist"> <dl> <dt class="hdlist1">Scanned block section</dt> <dd> <p>The section is named so because it contains all data blocks that need to be read when an HFile is scanned sequentially. Also contains leaf block index and Bloom chunk blocks.</p> </dd> <dt class="hdlist1">Non-scanned block section</dt> <dd> <p>This section still contains unified-format v2 blocks but it does not have to be read when doing a sequential scan. This section contains "meta" blocks and intermediate-level index blocks.</p> </dd> </dl> </div> <div class="paragraph"> <p>We are supporting "meta" blocks in version 2 the same way they were supported in version 1, even though we do not store Bloom filter data in these blocks anymore.</p> </div> </div> <div class="sect3"> <h4 id="_block_index_in_version_2"><a class="anchor" href="#_block_index_in_version_2"></a>H.2.4. Block index in version 2</h4> <div class="paragraph"> <p>There are three types of block indexes in HFile version 2, stored in two different formats (root and non-root):</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Data index — version 2 multi-level block index, consisting of:</p> <div class="olist loweralpha"> <ol class="loweralpha" type="a"> <li> <p>Version 2 root index, stored in the data block index section of the file</p> </li> <li> <p>Optionally, version 2 intermediate levels, stored in the non%root format in the data index section of the file. Intermediate levels can only be present if leaf level blocks are present</p> </li> <li> <p>Optionally, version 2 leaf levels, stored in the non%root format inline with data blocks</p> </li> </ol> </div> </li> <li> <p>Meta index — version 2 root index format only, stored in the meta index section of the file</p> </li> <li> <p>Bloom index — version 2 root index format only, stored in the ``load-on-open'' section as part of Bloom filter metadata.</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="_root_block_index_format_in_version_2"><a class="anchor" href="#_root_block_index_format_in_version_2"></a>H.2.5. Root block index format in version 2</h4> <div class="paragraph"> <p>This format applies to:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Root level of the version 2 data index</p> </li> <li> <p>Entire meta and Bloom indexes in version 2, which are always single-level.</p> </li> </ol> </div> <div class="paragraph"> <p>A version 2 root index block is a sequence of entries of the following format, similar to entries of a version 1 block index, but storing on-disk size instead of uncompressed size.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Offset (long)</p> <div class="paragraph"> <p>This offset may point to a data block or to a deeper>level index block.</p> </div> </li> <li> <p>On-disk size (int)</p> </li> <li> <p>Key (a serialized byte array stored using Bytes.writeByteArray)</p> </li> <li> <p>Key (VInt)</p> </li> <li> <p>Key bytes</p> </li> </ol> </div> <div class="paragraph"> <p>A single-level version 2 block index consists of just a single root index block. To read a root index block of version 2, one needs to know the number of entries. For the data index and the meta index the number of entries is stored in the trailer, and for the Bloom index it is stored in the compound Bloom filter metadata.</p> </div> <div class="paragraph"> <p>For a multi-level block index we also store the following fields in the root index block in the load-on-open section of the HFile, in addition to the data structure described above:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Middle leaf index block offset</p> </li> <li> <p>Middle leaf block on-disk size (meaning the leaf index block containing the reference to the ``middle'' data block of the file)</p> </li> <li> <p>The index of the mid-key (defined below) in the middle leaf-level block.</p> </li> </ol> </div> <div class="paragraph"> <p>These additional fields are used to efficiently retrieve the mid-key of the HFile used in HFile splits, which we define as the first key of the block with a zero-based index of (n – 1) / 2, if the total number of blocks in the HFile is n. This definition is consistent with how the mid-key was determined in HFile version 1, and is reasonable in general, because blocks are likely to be the same size on average, but we don’t have any estimates on individual key/value pair sizes.</p> </div> <div class="paragraph"> <p>When writing a version 2 HFile, the total number of data blocks pointed to by every leaf-level index block is kept track of. When we finish writing and the total number of leaf-level blocks is determined, it is clear which leaf-level block contains the mid-key, and the fields listed above are computed. When reading the HFile and the mid-key is requested, we retrieve the middle leaf index block (potentially from the block cache) and get the mid-key value from the appropriate position inside that leaf block.</p> </div> </div> <div class="sect3"> <h4 id="_non_root_block_index_format_in_version_2"><a class="anchor" href="#_non_root_block_index_format_in_version_2"></a>H.2.6. Non-root block index format in version 2</h4> <div class="paragraph"> <p>This format applies to intermediate-level and leaf index blocks of a version 2 multi-level data block index. Every non-root index block is structured as follows.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>numEntries: the number of entries (int).</p> </li> <li> <p>entryOffsets: the <code>secondary index'' of offsets of entries in the block, to facilitate a quick binary search on the key (numEntries + 1 int values). The last value is the total length of all entries in this index block. For example, in a non-root index block with entry sizes 60, 80, 50 the </code>secondary index'' will contain the following int array: {0, 60, 140, 190}.</p> </li> <li> <p>Entries. Each entry contains:</p> </li> <li> <p>Offset of the block referenced by this entry in the file (long)</p> </li> <li> <p>On>disk size of the referenced block (int)</p> </li> <li> <p>Key. The length can be calculated from entryOffsets.</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="_bloom_filters_in_version_2"><a class="anchor" href="#_bloom_filters_in_version_2"></a>H.2.7. Bloom filters in version 2</h4> <div class="paragraph"> <p>In contrast with version 1, in a version 2 HFile Bloom filter metadata is stored in the load-on-open section of the HFile for quick startup.</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>A compound Bloom filter.</p> </li> <li> <p>Bloom filter version = 3 (int). There used to be a DynamicByteBloomFilter class that had the Bloom filter version number 2</p> </li> <li> <p>The total byte size of all compound Bloom filter chunks (long)</p> </li> <li> <p>Number of hash functions (int</p> </li> <li> <p>Type of hash functions (int)</p> </li> <li> <p>The total key count inserted into the Bloom filter (long)</p> </li> <li> <p>The maximum total number of keys in the Bloom filter (long)</p> </li> <li> <p>The number of chunks (int)</p> </li> <li> <p>Comparator class used for Bloom filter keys, a UTF>8 encoded string stored using Bytes.writeByteArray</p> </li> <li> <p>Bloom block index in the version 2 root block index format</p> </li> </ol> </div> </div> <div class="sect3"> <h4 id="_file_info_format_in_versions_1_and_2"><a class="anchor" href="#_file_info_format_in_versions_1_and_2"></a>H.2.8. File Info format in versions 1 and 2</h4> <div class="paragraph"> <p>The file info block is a serialized <a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/io/HbaseMapWritable.html">HbaseMapWritable</a> (essentially a map from byte arrays to byte arrays) with the following keys, among others. StoreFile-level logic adds more keys to this.</p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hfile.LASTKEY</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The last key of the file (byte array)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hfile.AVG_KEY_LEN</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The average key length in the file (int)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hfile.AVG_VALUE_LEN</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The average value length in the file (int)</p></td> </tr> </tbody> </table> <div class="paragraph"> <p>File info format did not change in version 2. However, we moved the file info to the final section of the file, which can be loaded as one block at the time the HFile is being opened. Also, we do not store comparator in the version 2 file info anymore. Instead, we store it in the fixed file trailer. This is because we need to know the comparator at the time of parsing the load-on-open section of the HFile.</p> </div> </div> <div class="sect3"> <h4 id="_fixed_file_trailer_format_differences_between_versions_1_and_2"><a class="anchor" href="#_fixed_file_trailer_format_differences_between_versions_1_and_2"></a>H.2.9. Fixed file trailer format differences between versions 1 and 2</h4> <div class="paragraph"> <p>The following table shows common and different fields between fixed file trailers in versions 1 and 2. Note that the size of the trailer is different depending on the version, so it is ``fixed'' only within one version. However, the version is always stored as the last four-byte integer in the file.</p> </div> <table class="tableblock frame-all grid-all spread"> <caption class="title">Table 13. Differences between HFile Versions 1 and 2</caption> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Version 1</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Version 2</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">File info offset (long)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Data index offset (long)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">loadOnOpenOffset (long) /The offset of the sectionthat we need toload when opening the file./</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Number of data index entries (int)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">metaIndexOffset (long) /This field is not being used by the version 1 reader, so we removed it from version 2./</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">uncompressedDataIndexSize (long) /The total uncompressed size of the whole data block index, including root-level, intermediate-level, and leaf-level blocks./</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Number of meta index entries (int)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Total uncompressed bytes (long)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">numEntries (int)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">numEntries (long)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Compression codec: 0 = LZO, 1 = GZ, 2 = NONE (int)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Compression codec: 0 = LZO, 1 = GZ, 2 = NONE (int)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The number of levels in the data block index (int)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">firstDataBlockOffset (long) /The offset of the first first data block. Used when scanning./</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">lastDataBlockEnd (long) /The offset of the first byte after the last key/value data block. We don’t need to go beyond this offset when scanning./</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Version: 1 (int)</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Version: 2 (int)</p></td> </tr> </tbody> </table> </div> <div class="sect3"> <h4 id="_getshortmidpointkey_an_optimization_for_data_index_block"><a class="anchor" href="#_getshortmidpointkey_an_optimization_for_data_index_block"></a>H.2.10. getShortMidpointKey(an optimization for data index block)</h4> <div class="paragraph"> <p>Note: this optimization was introduced in HBase 0.95+</p> </div> <div class="paragraph"> <p>HFiles contain many blocks that contain a range of sorted Cells. Each cell has a key. To save IO when reading Cells, the HFile also has an index that maps a Cell’s start key to the offset of the beginning of a particular block. Prior to this optimization, HBase would use the key of the first cell in each data block as the index key.</p> </div> <div class="paragraph"> <p>In HBASE-7845, we generate a new key that is lexicographically larger than the last key of the previous block and lexicographically equal or smaller than the start key of the current block. While actual keys can potentially be very long, this "fake key" or "virtual key" can be much shorter. For example, if the stop key of previous block is "the quick brown fox", the start key of current block is "the who", we could use "the r" as our virtual key in our hfile index.</p> </div> <div class="paragraph"> <p>There are two benefits to this:</p> </div> <div class="ulist"> <ul> <li> <p>having shorter keys reduces the hfile index size, (allowing us to keep more indexes in memory), and</p> </li> <li> <p>using something closer to the end key of the previous block allows us to avoid a potential extra IO when the target key lives in between the "virtual key" and the key of the first element in the target block.</p> </li> </ul> </div> <div class="paragraph"> <p>This optimization (implemented by the getShortMidpointKey method) is inspired by LevelDB’s ByteWiseComparatorImpl::FindShortestSeparator() and FindShortSuccessor().</p> </div> </div> </div> <div class="sect2"> <h3 id="hfilev3"><a class="anchor" href="#hfilev3"></a>H.3. HBase File Format with Security Enhancements (version 3)</h3> <div class="paragraph"> <p>Note: this feature was introduced in HBase 0.98</p> </div> <div class="sect3"> <h4 id="hfilev3.motivation"><a class="anchor" href="#hfilev3.motivation"></a>H.3.1. Motivation</h4> <div class="paragraph"> <p>Version 3 of HFile makes changes needed to ease management of encryption at rest and cell-level metadata (which in turn is needed for cell-level ACLs and cell-level visibility labels). For more information see <a href="#hbase.encryption.server">hbase.encryption.server</a>, <a href="#hbase.tags">hbase.tags</a>, <a href="#hbase.accesscontrol.configuration">hbase.accesscontrol.configuration</a>, and <a href="#hbase.visibility.labels">hbase.visibility.labels</a>.</p> </div> </div> <div class="sect3"> <h4 id="hfilev3.overview"><a class="anchor" href="#hfilev3.overview"></a>H.3.2. Overview</h4> <div class="paragraph"> <p>The version of HBase introducing the above features reads HFiles in versions 1, 2, and 3 but only writes version 3 HFiles. Version 3 HFiles are structured the same as version 2 HFiles. For more information see <a href="#hfilev2.overview">hfilev2.overview</a>.</p> </div> </div> <div class="sect3"> <h4 id="hvilev3.infoblock"><a class="anchor" href="#hvilev3.infoblock"></a>H.3.3. File Info Block in Version 3</h4> <div class="paragraph"> <p>Version 3 added two additional pieces of information to the reserved keys in the file info block.</p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hfile.MAX_TAGS_LEN</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">The maximum number of bytes needed to store the serialized tags for any single cell in this hfile (int)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">hfile.TAGS_COMPRESSED</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Does the block encoder for this hfile compress tags? (boolean). Should only be present if hfile.MAX_TAGS_LEN is also present.</p></td> </tr> </tbody> </table> <div class="paragraph"> <p>When reading a Version 3 HFile the presence of <code>MAX_TAGS_LEN</code> is used to determine how to deserialize the cells within a data block. Therefore, consumers must read the file’s info block prior to reading any data blocks.</p> </div> <div class="paragraph"> <p>When writing a Version 3 HFile, HBase will always include `MAX_TAGS_LEN ` when flushing the memstore to underlying filesystem and when using prefix tree encoding for data blocks, as described in <a href="#compression">compression</a>.</p> </div> <div class="paragraph"> <p>When compacting extant files, the default writer will omit <code>MAX_TAGS_LEN</code> if all of the files selected do not themselves contain any cells with tags.</p> </div> <div class="paragraph"> <p>See <a href="#compaction">compaction</a> for details on the compaction file selection algorithm.</p> </div> </div> <div class="sect3"> <h4 id="hfilev3.datablock"><a class="anchor" href="#hfilev3.datablock"></a>H.3.4. Data Blocks in Version 3</h4> <div class="paragraph"> <p>Within an HFile, HBase cells are stored in data blocks as a sequence of KeyValues (see <a href="#hfilev1.overview">hfilev1.overview</a>, or <a href="http://www.larsgeorge.com/2009/10/hbase-architecture-101-storage.html">Lars George’s excellent introduction to HBase Storage</a>). In version 3, these KeyValue optionally will include a set of 0 or more tags:</p> </div> <table class="tableblock frame-all grid-all spread"> <colgroup> <col style="width: 50%;"> <col style="width: 50%;"> </colgroup> <tbody> <tr> <td class="tableblock halign-left valign-top"><p class="tableblock">Version 1 & 2, Version 3 without MAX_TAGS_LEN</p></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Version 3 with MAX_TAGS_LEN</p></td> </tr> <tr> <td class="tableblock halign-left valign-top" colspan="2"><p class="tableblock">Key Length (4 bytes)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top" colspan="2"><p class="tableblock">Value Length (4 bytes)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top" colspan="2"><p class="tableblock">Key bytes (variable)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top" colspan="2"><p class="tableblock">Value bytes (variable)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Tags Length (2 bytes)</p></td> </tr> <tr> <td class="tableblock halign-left valign-top"></td> <td class="tableblock halign-left valign-top"><p class="tableblock">Tags bytes (variable)</p></td> </tr> </tbody> </table> <div class="paragraph"> <p>If the info block for a given HFile contains an entry for <code>MAX_TAGS_LEN</code> each cell will have the length of that cell’s tags included, even if that length is zero. The actual tags are stored as a sequence of tag length (2 bytes), tag type (1 byte), tag bytes (variable). The format an individual tag’s bytes depends on the tag type.</p> </div> <div class="paragraph"> <p>Note that the dependence on the contents of the info block implies that prior to reading any data blocks you must first process a file’s info block. It also implies that prior to writing a data block you must know if the file’s info block will include <code>MAX_TAGS_LEN</code>.</p> </div> </div> <div class="sect3"> <h4 id="hfilev3.fixedtrailer"><a class="anchor" href="#hfilev3.fixedtrailer"></a>H.3.5. Fixed File Trailer in Version 3</h4> <div class="paragraph"> <p>The fixed file trailers written with HFile version 3 are always serialized with protocol buffers. Additionally, it adds an optional field to the version 2 protocol buffer named encryption_key. If HBase is configured to encrypt HFiles this field will store a data encryption key for this particular HFile, encrypted with the current cluster master key using AES. For more information see <a href="#hbase.encryption.server">hbase.encryption.server</a>.</p> </div> </div> </div> </div> </div> <div class="sect1"> <h2 id="other.info"><a class="anchor" href="#other.info"></a>Appendix I: Other Information About HBase</h2> <div class="sectionbody"> <div class="sect2"> <h3 id="other.info.videos"><a class="anchor" href="#other.info.videos"></a>I.1. HBase Videos</h3> <div class="ulist"> <div class="title">Introduction to HBase</div> <ul> <li> <p><a href="http://www.cloudera.com/content/cloudera/en/resources/library/presentation/chicago_data_summit_apache_hbase_an_introduction_todd_lipcon.html">Introduction to HBase</a> by Todd Lipcon (Chicago Data Summit 2011).</p> </li> <li> <p><a href="http://www.cloudera.com/videos/intorduction-hbase-todd-lipcon">Introduction to HBase</a> by Todd Lipcon (2010). <a href="http://www.cloudera.com/videos/hadoop-world-2011-presentation-video-building-realtime-big-data-services-at-facebook-with-hadoop-and-hbase">Building Real Time Services at Facebook with HBase</a> by Jonathan Gray (Hadoop World 2011).</p> </li> </ul> </div> <div class="paragraph"> <p><a href="http://www.cloudera.com/videos/hw10_video_how_stumbleupon_built_and_advertising_platform_using_hbase_and_hadoop">HBase and Hadoop, Mixing Real-Time and Batch Processing at StumbleUpon</a> by JD Cryans (Hadoop World 2010).</p> </div> </div> <div class="sect2"> <h3 id="other.info.pres"><a class="anchor" href="#other.info.pres"></a>I.2. HBase Presentations (Slides)</h3> <div class="paragraph"> <p><a href="http://www.cloudera.com/content/cloudera/en/resources/library/hadoopworld/hadoop-world-2011-presentation-video-advanced-hbase-schema-design.html">Advanced HBase Schema Design</a> by Lars George (Hadoop World 2011).</p> </div> <div class="paragraph"> <p><a href="http://www.slideshare.net/cloudera/chicago-data-summit-apache-hbase-an-introduction">Introduction to HBase</a> by Todd Lipcon (Chicago Data Summit 2011).</p> </div> <div class="paragraph"> <p><a href="http://www.slideshare.net/cloudera/hw09-practical-h-base-getting-the-most-from-your-h-base-install">Getting The Most From Your HBase Install</a> by Ryan Rawson, Jonathan Gray (Hadoop World 2009).</p> </div> </div> <div class="sect2"> <h3 id="other.info.papers"><a class="anchor" href="#other.info.papers"></a>I.3. HBase Papers</h3> <div class="paragraph"> <p><a href="http://research.google.com/archive/bigtable.html">BigTable</a> by Google (2006).</p> </div> <div class="paragraph"> <p><a href="http://www.larsgeorge.com/2010/05/hbase-file-locality-in-hdfs.html">HBase and HDFS Locality</a> by Lars George (2010).</p> </div> <div class="paragraph"> <p><a href="http://ianvarley.com/UT/MR/Varley_MastersReport_Full_2009-08-07.pdf">No Relation: The Mixed Blessings of Non-Relational Databases</a> by Ian Varley (2009).</p> </div> </div> <div class="sect2"> <h3 id="other.info.sites"><a class="anchor" href="#other.info.sites"></a>I.4. HBase Sites</h3> <div class="paragraph"> <p><a href="http://www.cloudera.com/blog/category/hbase/">Cloudera’s HBase Blog</a> has a lot of links to useful HBase information.</p> </div> <div class="ulist"> <ul> <li> <p><a href="http://www.cloudera.com/blog/2010/04/cap-confusion-problems-with-partition-tolerance/">CAP Confusion</a> is a relevant entry for background information on distributed storage systems.</p> </li> </ul> </div> <div class="paragraph"> <p><a href="http://wiki.apache.org/hadoop/HBase/HBasePresentations">HBase Wiki</a> has a page with a number of presentations.</p> </div> <div class="paragraph"> <p><a href="http://refcardz.dzone.com/refcardz/hbase">HBase RefCard</a> from DZone.</p> </div> </div> <div class="sect2"> <h3 id="other.info.books"><a class="anchor" href="#other.info.books"></a>I.5. HBase Books</h3> <div class="paragraph"> <p><a href="http://shop.oreilly.com/product/0636920014348.do">HBase: The Definitive Guide</a> by Lars George.</p> </div> </div> <div class="sect2"> <h3 id="other.info.books.hadoop"><a class="anchor" href="#other.info.books.hadoop"></a>I.6. Hadoop Books</h3> <div class="paragraph"> <p><a href="http://shop.oreilly.com/product/9780596521981.do">Hadoop: The Definitive Guide</a> by Tom White.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbase.history"><a class="anchor" href="#hbase.history"></a>Appendix J: HBase History</h2> <div class="sectionbody"> <div class="ulist"> <ul> <li> <p>2006: <a href="http://research.google.com/archive/bigtable.html">BigTable</a> paper published by Google.</p> </li> <li> <p>2006 (end of year): HBase development starts.</p> </li> <li> <p>2008: HBase becomes Hadoop sub-project.</p> </li> <li> <p>2010: HBase becomes Apache top-level project.</p> </li> </ul> </div> </div> </div> <div class="sect1"> <h2 id="asf"><a class="anchor" href="#asf"></a>Appendix K: HBase and the Apache Software Foundation</h2> <div class="sectionbody"> <div class="paragraph"> <p>HBase is a project in the Apache Software Foundation and as such there are responsibilities to the ASF to ensure a healthy project.</p> </div> <div class="sect2"> <h3 id="asf.devprocess"><a class="anchor" href="#asf.devprocess"></a>K.1. ASF Development Process</h3> <div class="paragraph"> <p>See the <a href="http://www.apache.org/dev/#committers">Apache Development Process page</a> for all sorts of information on how the ASF is structured (e.g., PMC, committers, contributors), to tips on contributing and getting involved, and how open-source works at ASF.</p> </div> </div> <div class="sect2"> <h3 id="asf.reporting"><a class="anchor" href="#asf.reporting"></a>K.2. ASF Board Reporting</h3> <div class="paragraph"> <p>Once a quarter, each project in the ASF portfolio submits a report to the ASF board. This is done by the HBase project lead and the committers. See <a href="http://www.apache.org/foundation/board/reporting">ASF board reporting</a> for more information.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="orca"><a class="anchor" href="#orca"></a>Appendix L: Apache HBase Orca</h2> <div class="sectionbody"> <div class="imageblock"> <div class="content"> <img src="images/jumping-orca_rotated_25percent.png" alt="jumping orca rotated 25percent"> </div> <div class="title">Figure 15. Apache HBase Orca</div> </div> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-4920">An Orca is the Apache HBase mascot.</a> See NOTICES.txt. Our Orca logo we got here: <a href="http://www.vectorfree.com/jumping-orca" class="bare">http://www.vectorfree.com/jumping-orca</a> It is licensed Creative Commons Attribution 3.0. See <a href="https://creativecommons.org/licenses/by/3.0/us/" class="bare">https://creativecommons.org/licenses/by/3.0/us/</a> We changed the logo by stripping the colored background, inverting it and then rotating it some.</p> </div> </div> </div> <div class="sect1"> <h2 id="tracing"><a class="anchor" href="#tracing"></a>Appendix M: Enabling Dapper-like Tracing in HBase</h2> <div class="sectionbody"> <div class="paragraph"> <p><a href="https://issues.apache.org/jira/browse/HBASE-6449">HBASE-6449</a> added support for tracing requests through HBase, using the open source tracing library, <a href="http://github.com/cloudera/htrace">HTrace</a>. Setting up tracing is quite simple, however it currently requires some very minor changes to your client code (it would not be very difficult to remove this requirement).</p> </div> <div class="sect2"> <h3 id="tracing.spanreceivers"><a class="anchor" href="#tracing.spanreceivers"></a>M.1. SpanReceivers</h3> <div class="paragraph"> <p>The tracing system works by collecting information in structs called 'Spans'. It is up to you to choose how you want to receive this information by implementing the <code>SpanReceiver</code> interface, which defines one method:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">public</span> <span class="type">void</span> receiveSpan(Span span);</code></pre> </div> </div> <div class="paragraph"> <p>This method serves as a callback whenever a span is completed. HTrace allows you to use as many SpanReceivers as you want so you can easily send trace information to multiple destinations.</p> </div> <div class="paragraph"> <p>Configure what SpanReceivers you’d like to us by putting a comma separated list of the fully-qualified class name of classes implementing <code>SpanReceiver</code> in <em>hbase-site.xml</em> property: <code>hbase.trace.spanreceiver.classes</code>.</p> </div> <div class="paragraph"> <p>HTrace includes a <code>LocalFileSpanReceiver</code> that writes all span information to local files in a JSON-based format. The <code>LocalFileSpanReceiver</code> looks in <em>hbase-site.xml</em> for a <code>hbase.local-file-span-receiver.path</code> property with a value describing the name of the file to which nodes should write their span information.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><property> <name>hbase.trace.spanreceiver.classes</name> <value>org.htrace.impl.LocalFileSpanReceiver</value> </property> <property> <name>hbase.local-file-span-receiver.path</name> <value>/var/log/hbase/htrace.out</value> </property></code></pre> </div> </div> <div class="paragraph"> <p>HTrace also provides <code>ZipkinSpanReceiver</code> which converts spans to <a href="http://github.com/twitter/zipkin">Zipkin</a> span format and send them to Zipkin server. In order to use this span receiver, you need to install the jar of htrace-zipkin to your HBase’s classpath on all of the nodes in your cluster.</p> </div> <div class="paragraph"> <p><em>htrace-zipkin</em> is published to the maven central repository. You could get the latest version from there or just build it locally and then copy it out to all nodes, change your config to use zipkin receiver, distribute the new configuration and then (rolling) restart.</p> </div> <div class="paragraph"> <p>Here is the example of manual setup procedure.</p> </div> <div class="listingblock"> <div class="content"> <pre>$ git clone https://github.com/cloudera/htrace $ cd htrace/htrace-zipkin $ mvn compile assembly:single $ cp target/htrace-zipkin-*-jar-with-dependencies.jar $HBASE_HOME/lib/ # copy jar to all nodes...</pre> </div> </div> <div class="paragraph"> <p>The <code>ZipkinSpanReceiver</code> looks in <em>hbase-site.xml</em> for a <code>hbase.zipkin.collector-hostname</code> and <code>hbase.zipkin.collector-port</code> property with a value describing the Zipkin collector server to which span information are sent.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="xml"><span class="tag"><property></span> <span class="tag"><name></span>hbase.trace.spanreceiver.classes<span class="tag"></name></span> <span class="tag"><value></span>org.htrace.impl.ZipkinSpanReceiver<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zipkin.collector-hostname<span class="tag"></name></span> <span class="tag"><value></span>localhost<span class="tag"></value></span> <span class="tag"></property></span> <span class="tag"><property></span> <span class="tag"><name></span>hbase.zipkin.collector-port<span class="tag"></name></span> <span class="tag"><value></span>9410<span class="tag"></value></span> <span class="tag"></property></span></code></pre> </div> </div> <div class="paragraph"> <p>If you do not want to use the included span receivers, you are encouraged to write your own receiver (take a look at <code>LocalFileSpanReceiver</code> for an example). If you think others would benefit from your receiver, file a JIRA or send a pull request to <a href="http://github.com/cloudera/htrace">HTrace</a>.</p> </div> </div> </div> </div> <div class="sect1"> <h2 id="tracing.client.modifications"><a class="anchor" href="#tracing.client.modifications"></a>155. Client Modifications</h2> <div class="sectionbody"> <div class="paragraph"> <p>In order to turn on tracing in your client code, you must initialize the module sending spans to receiver once per client process.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="directive">private</span> SpanReceiverHost spanReceiverHost; ... Configuration conf = HBaseConfiguration.create(); SpanReceiverHost spanReceiverHost = SpanReceiverHost.getInstance(conf);</code></pre> </div> </div> <div class="paragraph"> <p>Then you simply start tracing span before requests you think are interesting, and close it when the request is done. For example, if you wanted to trace all of your get operations, you change this:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="predefined-type">Configuration</span> config = HBaseConfiguration.create(); <span class="predefined-type">Connection</span> connection = ConnectionFactory.createConnection(config); Table table = connection.getTable(TableName.valueOf(<span class="string"><span class="delimiter">"</span><span class="content">t1</span><span class="delimiter">"</span></span>)); Get get = <span class="keyword">new</span> Get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">r1</span><span class="delimiter">"</span></span>)); <span class="predefined-type">Result</span> res = table.get(get);</code></pre> </div> </div> <div class="paragraph"> <p>into:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">TraceScope ts = Trace.startSpan(<span class="string"><span class="delimiter">"</span><span class="content">Gets</span><span class="delimiter">"</span></span>, Sampler.ALWAYS); <span class="keyword">try</span> { Table table = connection.getTable(TableName.valueOf(<span class="string"><span class="delimiter">"</span><span class="content">t1</span><span class="delimiter">"</span></span>)); Get get = <span class="keyword">new</span> Get(Bytes.toBytes(<span class="string"><span class="delimiter">"</span><span class="content">r1</span><span class="delimiter">"</span></span>)); <span class="predefined-type">Result</span> res = table.get(get); } <span class="keyword">finally</span> { ts.close(); }</code></pre> </div> </div> <div class="paragraph"> <p>If you wanted to trace half of your 'get' operations, you would pass in:</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><span class="keyword">new</span> ProbabilitySampler(<span class="float">0.5</span>)</code></pre> </div> </div> <div class="paragraph"> <p>in lieu of <code>Sampler.ALWAYS</code> to <code>Trace.startSpan()</code>. See the HTrace <em>README</em> for more information on Samplers.</p> </div> </div> </div> <div class="sect1"> <h2 id="tracing.client.shell"><a class="anchor" href="#tracing.client.shell"></a>156. Tracing from HBase Shell</h2> <div class="sectionbody"> <div class="paragraph"> <p>You can use trace command for tracing requests from HBase Shell. trace 'start' command turns on tracing and trace 'stop' command turns off tracing.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">001</span>:<span class="integer">0</span>> trace <span class="string"><span class="delimiter">'</span><span class="content">start</span><span class="delimiter">'</span></span> hbase(main):<span class="octal">002</span>:<span class="integer">0</span>> put <span class="string"><span class="delimiter">'</span><span class="content">test</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">row1</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">f:</span><span class="delimiter">'</span></span>, <span class="string"><span class="delimiter">'</span><span class="content">val1</span><span class="delimiter">'</span></span> <span class="error">#</span> traced commands hbase(main):<span class="octal">003</span>:<span class="integer">0</span>> trace <span class="string"><span class="delimiter">'</span><span class="content">stop</span><span class="delimiter">'</span></span></code></pre> </div> </div> <div class="paragraph"> <p>trace 'start' and trace 'stop' always returns boolean value representing if or not there is ongoing tracing. As a result, trace 'stop' returns false on suceess. trace 'status' just returns if or not tracing is turned on.</p> </div> <div class="listingblock"> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java">hbase(main):<span class="octal">001</span>:<span class="integer">0</span>> trace <span class="string"><span class="delimiter">'</span><span class="content">start</span><span class="delimiter">'</span></span> => <span class="predefined-constant">true</span> hbase(main):<span class="octal">002</span>:<span class="integer">0</span>> trace <span class="string"><span class="delimiter">'</span><span class="content">status</span><span class="delimiter">'</span></span> => <span class="predefined-constant">true</span> hbase(main):<span class="octal">003</span>:<span class="integer">0</span>> trace <span class="string"><span class="delimiter">'</span><span class="content">stop</span><span class="delimiter">'</span></span> => <span class="predefined-constant">false</span> hbase(main):<span class="octal">004</span>:<span class="integer">0</span>> trace <span class="string"><span class="delimiter">'</span><span class="content">status</span><span class="delimiter">'</span></span> => <span class="predefined-constant">false</span></code></pre> </div> </div> </div> </div> <div class="sect1"> <h2 id="hbase.rpc"><a class="anchor" href="#hbase.rpc"></a>Appendix N: 0.95 RPC Specification</h2> <div class="sectionbody"> <div class="paragraph"> <p>In 0.95, all client/server communication is done with <a href="https://developers.google.com/protocol-buffers/">protobuf’ed</a> Messages rather than with <a href="http://hadoop.apache.org/docs/current/api/org/apache/hadoop/io/Writable.html">Hadoop Writables</a>. Our RPC wire format therefore changes. This document describes the client/server request/response protocol and our new RPC wire-format.</p> </div> <div class="paragraph"> <p>For what RPC is like in 0.94 and previous, see Benoît/Tsuna’s <a href="https://github.com/OpenTSDB/asynchbase/blob/master/src/HBaseRpc.java#L164">Unofficial Hadoop / HBase RPC protocol documentation</a>. For more background on how we arrived at this spec., see <a href="https://docs.google.com/document/d/1WCKwgaLDqBw2vpux0jPsAu2WPTRISob7HGCO8YhfDTA/edit#">HBase RPC: WIP</a></p> </div> <div class="sect2"> <h3 id="_goals"><a class="anchor" href="#_goals"></a>N.1. Goals</h3> <div class="olist arabic"> <ol class="arabic"> <li> <p>A wire-format we can evolve</p> </li> <li> <p>A format that does not require our rewriting server core or radically changing its current architecture (for later).</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="_todo"><a class="anchor" href="#_todo"></a>N.2. TODO</h3> <div class="olist arabic"> <ol class="arabic"> <li> <p>List of problems with currently specified format and where we would like to go in a version2, etc. For example, what would we have to change if anything to move server async or to support streaming/chunking?</p> </li> <li> <p>Diagram on how it works</p> </li> <li> <p>A grammar that succinctly describes the wire-format. Currently we have these words and the content of the rpc protobuf idl but a grammar for the back and forth would help with groking rpc. Also, a little state machine on client/server interactions would help with understanding (and ensuring correct implementation).</p> </li> </ol> </div> </div> <div class="sect2"> <h3 id="_rpc"><a class="anchor" href="#_rpc"></a>N.3. RPC</h3> <div class="paragraph"> <p>The client will send setup information on connection establish. Thereafter, the client invokes methods against the remote server sending a protobuf Message and receiving a protobuf Message in response. Communication is synchronous. All back and forth is preceded by an int that has the total length of the request/response. Optionally, Cells(KeyValues) can be passed outside of protobufs in follow-behind Cell blocks (because <a href="https://docs.google.com/document/d/1WEtrq-JTIUhlnlnvA0oYRLp0F8MKpEBeBSCFcQiacdw/edit#">we can’t protobuf megabytes of KeyValues</a> or Cells). These CellBlocks are encoded and optionally compressed.</p> </div> <div class="paragraph"> <p>For more detail on the protobufs involved, see the <a href="http://svn.apache.org/viewvc/hbase/trunk/hbase-protocol/src/main/protobuf/RPC.proto?view=markup">RPC.proto</a> file in trunk.</p> </div> <div class="sect3"> <h4 id="_connection_setup"><a class="anchor" href="#_connection_setup"></a>N.3.1. Connection Setup</h4> <div class="paragraph"> <p>Client initiates connection.</p> </div> <div class="sect4"> <h5 id="_client"><a class="anchor" href="#_client"></a>Client</h5> <div class="paragraph"> <p>On connection setup, client sends a preamble followed by a connection header.</p> </div> <div class="listingblock"> <div class="title"><preamble></div> <div class="content"> <pre class="CodeRay highlight"><code data-lang="java"><MAGIC <span class="integer">4</span> <span class="type">byte</span> integer> <<span class="integer">1</span> <span class="type">byte</span> RPC <span class="predefined-type">Format</span> Version> <<span class="integer">1</span> <span class="type">byte</span> auth type></code></pre> </div> </div> <div class="paragraph"> <p>We need the auth method spec. here so the connection header is encoded if auth enabled.</p> </div> <div class="paragraph"> <p>E.g.: HBas0x000x50 — 4 bytes of MAGIC — `HBas' — plus one-byte of version, 0 in this case, and one byte, 0x50 (SIMPLE). of an auth type.</p> </div> <div class="paragraph"> <div class="title"><Protobuf ConnectionHeader Message></div> <p>Has user info, and ``protocol'', as well as the encoders and compression the client will use sending CellBlocks. CellBlock encoders and compressors are for the life of the connection. CellBlock encoders implement org.apache.hadoop.hbase.codec.Codec. CellBlocks may then also be compressed. Compressors implement org.apache.hadoop.io.compress.CompressionCodec. This protobuf is written using writeDelimited so is prefaced by a pb varint with its serialized length</p> </div> </div> <div class="sect4"> <h5 id="_server"><a class="anchor" href="#_server"></a>Server</h5> <div class="paragraph"> <p>After client sends preamble and connection header, server does NOT respond if successful connection setup. No response means server is READY to accept requests and to give out response. If the version or authentication in the preamble is not agreeable or the server has trouble parsing the preamble, it will throw a org.apache.hadoop.hbase.ipc.FatalConnectionException explaining the error and will then disconnect. If the client in the connection header — i.e. the protobuf’d Message that comes after the connection preamble — asks for for a Service the server does not support or a codec the server does not have, again we throw a FatalConnectionException with explanation.</p> </div> </div> </div> <div class="sect3"> <h4 id="_request"><a class="anchor" href="#_request"></a>N.3.2. Request</h4> <div class="paragraph"> <p>After a Connection has been set up, client makes requests. Server responds.</p> </div> <div class="paragraph"> <p>A request is made up of a protobuf RequestHeader followed by a protobuf Message parameter. The header includes the method name and optionally, metadata on the optional CellBlock that may be following. The parameter type suits the method being invoked: i.e. if we are doing a getRegionInfo request, the protobuf Message param will be an instance of GetRegionInfoRequest. The response will be a GetRegionInfoResponse. The CellBlock is optionally used ferrying the bulk of the RPC data: i.e Cells/KeyValues.</p> </div> <div class="sect4"> <h5 id="_request_parts"><a class="anchor" href="#_request_parts"></a>Request Parts</h5> <div class="paragraph"> <div class="title"><Total Length></div> <p>The request is prefaced by an int that holds the total length of what follows.</p> </div> <div class="paragraph"> <div class="title"><Protobuf RequestHeader Message></div> <p>Will have call.id, trace.id, and method name, etc. including optional Metadata on the Cell block IFF one is following. Data is protobuf’d inline in this pb Message or optionally comes in the following CellBlock</p> </div> <div class="paragraph"> <div class="title"><Protobuf Param Message></div> <p>If the method being invoked is getRegionInfo, if you study the Service descriptor for the client to regionserver protocol, you will find that the request sends a GetRegionInfoRequest protobuf Message param in this position.</p> </div> <div class="paragraph"> <div class="title"><CellBlock></div> <p>An encoded and optionally compressed Cell block.</p> </div> </div> </div> <div class="sect3"> <h4 id="_response"><a class="anchor" href="#_response"></a>N.3.3. Response</h4> <div class="paragraph"> <p>Same as Request, it is a protobuf ResponseHeader followed by a protobuf Message response where the Message response type suits the method invoked. Bulk of the data may come in a following CellBlock.</p> </div> <div class="sect4"> <h5 id="_response_parts"><a class="anchor" href="#_response_parts"></a>Response Parts</h5> <div class="paragraph"> <div class="title"><Total Length></div> <p>The response is prefaced by an int that holds the total length of what follows.</p> </div> <div class="paragraph"> <div class="title"><Protobuf ResponseHeader Message></div> <p>Will have call.id, etc. Will include exception if failed processing. Optionally includes metadata on optional, IFF there is a CellBlock following.</p> </div> <div class="paragraph"> <div class="title"><Protobuf Response Message></div> <p>Return or may be nothing if exception. If the method being invoked is getRegionInfo, if you study the Service descriptor for the client to regionserver protocol, you will find that the response sends a GetRegionInfoResponse protobuf Message param in this position.</p> </div> <div class="paragraph"> <div class="title"><CellBlock></div> <p>An encoded and optionally compressed Cell block.</p> </div> </div> </div> <div class="sect3"> <h4 id="_exceptions"><a class="anchor" href="#_exceptions"></a>N.3.4. Exceptions</h4> <div class="paragraph"> <p>There are two distinct types. There is the request failed which is encapsulated inside the response header for the response. The connection stays open to receive new requests. The second type, the FatalConnectionException, kills the connection.</p> </div> <div class="paragraph"> <p>Exceptions can carry extra information. See the ExceptionResponse protobuf type. It has a flag to indicate do-no-retry as well as other miscellaneous payload to help improve client responsiveness.</p> </div> </div> <div class="sect3"> <h4 id="_cellblocks"><a class="anchor" href="#_cellblocks"></a>N.3.5. CellBlocks</h4> <div class="paragraph"> <p>These are not versioned. Server can do the codec or it cannot. If new version of a codec with say, tighter encoding, then give it a new class name. Codecs will live on the server for all time so old clients can connect.</p> </div> </div> </div> <div class="sect2"> <h3 id="_notes_2"><a class="anchor" href="#_notes_2"></a>N.4. Notes</h3> <div class="paragraph"> <div class="title">Constraints</div> <p>In some part, current wire-format — i.e. all requests and responses preceeded by a length — has been dictated by current server non-async architecture.</p> </div> <div class="paragraph"> <div class="title">One fat pb request or header+param</div> <p>We went with pb header followed by pb param making a request and a pb header followed by pb response for now. Doing header+param rather than a single protobuf Message with both header and param content:</p> </div> <div class="olist arabic"> <ol class="arabic"> <li> <p>Is closer to what we currently have</p> </li> <li> <p>Having a single fat pb requires extra copying putting the already pb’d param into the body of the fat request pb (and same making result)</p> </li> <li> <p>We can decide whether to accept the request or not before we read the param; for example, the request might be low priority. As is, we read header+param in one go as server is currently implemented so this is a TODO.</p> </li> </ol> </div> <div class="paragraph"> <p>The advantages are minor. If later, fat request has clear advantage, can roll out a v2 later.</p> </div> <div class="sect3"> <h4 id="rpc.configs"><a class="anchor" href="#rpc.configs"></a>N.4.1. RPC Configurations</h4> <div class="paragraph"> <div class="title">CellBlock Codecs</div> <p>To enable a codec other than the default <code>KeyValueCodec</code>, set <code>hbase.client.rpc.codec</code> to the name of the Codec class to use. Codec must implement hbase’s <code>Codec</code> Interface. After connection setup, all passed cellblocks will be sent with this codec. The server will return cellblocks using this same codec as long as the codec is on the servers' CLASSPATH (else you will get <code>UnsupportedCellCodecException</code>).</p> </div> <div class="paragraph"> <p>To change the default codec, set <code>hbase.client.default.rpc.codec</code>.</p> </div> <div class="paragraph"> <p>To disable cellblocks completely and to go pure protobuf, set the default to the empty String and do not specify a codec in your Configuration. So, set <code>hbase.client.default.rpc.codec</code> to the empty string and do not set <code>hbase.client.rpc.codec</code>. This will cause the client to connect to the server with no codec specified. If a server sees no codec, it will return all responses in pure protobuf. Running pure protobuf all the time will be slower than running with cellblocks.</p> </div> <div class="paragraph"> <div class="title">Compression</div> <p>Uses hadoops compression codecs. To enable compressing of passed CellBlocks, set <code>hbase.client.rpc.compressor</code> to the name of the Compressor to use. Compressor must implement Hadoops' CompressionCodec Interface. After connection setup, all passed cellblocks will be sent compressed. The server will return cellblocks compressed using this same compressor as long as the compressor is on its CLASSPATH (else you will get <code>UnsupportedCompressionCodecException</code>).</p> </div> </div> </div> </div> </div> </div> <div id="footnotes"> <hr> <div class="footnote" id="_footnote_1"> <a href="#_footnoteref_1">1</a>. See <a href="http://docs.oracle.com/javase/specs/jls/se7/html/jls-13.html" class="bare">http://docs.oracle.com/javase/specs/jls/se7/html/jls-13.html</a>. </div> <div class="footnote" id="_footnote_2"> <a href="#_footnoteref_2">2</a>. Note that this indicates what could break, not that it will break. We will/should add specifics in our release notes. </div> <div class="footnote" id="_footnote_3"> <a href="#_footnoteref_3">3</a>. comp_matrix_offline_upgrade_note,Running an offline upgrade tool without rollback might be needed. We will typically only support migrating data from major version X to major version X+1. </div> <div class="footnote" id="_footnote_4"> <a href="#_footnoteref_4">4</a>. The Metrics system was redone in HBase 0.96. See Migration to the New Metrics Hotness – Metrics2 by Elliot Clark for detail </div> </div> <div id="footer"> <div id="footer-text"> Version 1.4.11<br> Last updated 2019-10-14 18:05:56 CDT </div> </div> </body> </html>