blob: f282731820b6dc8da0cd7565940dc2e933e3e824 [file] [log] [blame]
<!DOCTYPE HTML>
<html lang="en">
<head>
<!-- Generated by javadoc (17) -->
<title>Source code</title>
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="source: package: org.apache.hadoop.hbase.regionserver, class: CompactionTool, class: CompactionMapper">
<meta name="generator" content="javadoc/SourceToHTMLConverter">
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body class="source-page">
<main role="main">
<div class="source-container">
<pre><span class="source-line-no">001</span><span id="line-1">/*</span>
<span class="source-line-no">002</span><span id="line-2"> * Licensed to the Apache Software Foundation (ASF) under one</span>
<span class="source-line-no">003</span><span id="line-3"> * or more contributor license agreements. See the NOTICE file</span>
<span class="source-line-no">004</span><span id="line-4"> * distributed with this work for additional information</span>
<span class="source-line-no">005</span><span id="line-5"> * regarding copyright ownership. The ASF licenses this file</span>
<span class="source-line-no">006</span><span id="line-6"> * to you under the Apache License, Version 2.0 (the</span>
<span class="source-line-no">007</span><span id="line-7"> * "License"); you may not use this file except in compliance</span>
<span class="source-line-no">008</span><span id="line-8"> * with the License. You may obtain a copy of the License at</span>
<span class="source-line-no">009</span><span id="line-9"> *</span>
<span class="source-line-no">010</span><span id="line-10"> * http://www.apache.org/licenses/LICENSE-2.0</span>
<span class="source-line-no">011</span><span id="line-11"> *</span>
<span class="source-line-no">012</span><span id="line-12"> * Unless required by applicable law or agreed to in writing, software</span>
<span class="source-line-no">013</span><span id="line-13"> * distributed under the License is distributed on an "AS IS" BASIS,</span>
<span class="source-line-no">014</span><span id="line-14"> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
<span class="source-line-no">015</span><span id="line-15"> * See the License for the specific language governing permissions and</span>
<span class="source-line-no">016</span><span id="line-16"> * limitations under the License.</span>
<span class="source-line-no">017</span><span id="line-17"> */</span>
<span class="source-line-no">018</span><span id="line-18">package org.apache.hadoop.hbase.regionserver;</span>
<span class="source-line-no">019</span><span id="line-19"></span>
<span class="source-line-no">020</span><span id="line-20">import static org.apache.hadoop.hbase.regionserver.Store.PRIORITY_USER;</span>
<span class="source-line-no">021</span><span id="line-21"></span>
<span class="source-line-no">022</span><span id="line-22">import java.io.IOException;</span>
<span class="source-line-no">023</span><span id="line-23">import java.util.ArrayList;</span>
<span class="source-line-no">024</span><span id="line-24">import java.util.HashSet;</span>
<span class="source-line-no">025</span><span id="line-25">import java.util.LinkedList;</span>
<span class="source-line-no">026</span><span id="line-26">import java.util.List;</span>
<span class="source-line-no">027</span><span id="line-27">import java.util.Optional;</span>
<span class="source-line-no">028</span><span id="line-28">import java.util.Set;</span>
<span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.conf.Configuration;</span>
<span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.conf.Configured;</span>
<span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.fs.FSDataOutputStream;</span>
<span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.fs.FileStatus;</span>
<span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.fs.FileSystem;</span>
<span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.fs.Path;</span>
<span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.HBaseConfiguration;</span>
<span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.HBaseInterfaceAudience;</span>
<span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.HDFSBlocksDistribution;</span>
<span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.client.RegionInfo;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.hbase.client.TableDescriptor;</span>
<span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.hbase.mapreduce.JobUtil;</span>
<span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;</span>
<span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.hbase.regionserver.compactions.CompactionContext;</span>
<span class="source-line-no">043</span><span id="line-43">import org.apache.hadoop.hbase.regionserver.compactions.CompactionLifeCycleTracker;</span>
<span class="source-line-no">044</span><span id="line-44">import org.apache.hadoop.hbase.regionserver.throttle.NoLimitThroughputController;</span>
<span class="source-line-no">045</span><span id="line-45">import org.apache.hadoop.hbase.util.Bytes;</span>
<span class="source-line-no">046</span><span id="line-46">import org.apache.hadoop.hbase.util.CommonFSUtils;</span>
<span class="source-line-no">047</span><span id="line-47">import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;</span>
<span class="source-line-no">048</span><span id="line-48">import org.apache.hadoop.hbase.util.FSTableDescriptors;</span>
<span class="source-line-no">049</span><span id="line-49">import org.apache.hadoop.hbase.util.FSUtils;</span>
<span class="source-line-no">050</span><span id="line-50">import org.apache.hadoop.io.LongWritable;</span>
<span class="source-line-no">051</span><span id="line-51">import org.apache.hadoop.io.NullWritable;</span>
<span class="source-line-no">052</span><span id="line-52">import org.apache.hadoop.io.Text;</span>
<span class="source-line-no">053</span><span id="line-53">import org.apache.hadoop.mapreduce.InputSplit;</span>
<span class="source-line-no">054</span><span id="line-54">import org.apache.hadoop.mapreduce.Job;</span>
<span class="source-line-no">055</span><span id="line-55">import org.apache.hadoop.mapreduce.JobContext;</span>
<span class="source-line-no">056</span><span id="line-56">import org.apache.hadoop.mapreduce.Mapper;</span>
<span class="source-line-no">057</span><span id="line-57">import org.apache.hadoop.mapreduce.lib.input.FileSplit;</span>
<span class="source-line-no">058</span><span id="line-58">import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;</span>
<span class="source-line-no">059</span><span id="line-59">import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;</span>
<span class="source-line-no">060</span><span id="line-60">import org.apache.hadoop.mapreduce.security.TokenCache;</span>
<span class="source-line-no">061</span><span id="line-61">import org.apache.hadoop.util.LineReader;</span>
<span class="source-line-no">062</span><span id="line-62">import org.apache.hadoop.util.Tool;</span>
<span class="source-line-no">063</span><span id="line-63">import org.apache.hadoop.util.ToolRunner;</span>
<span class="source-line-no">064</span><span id="line-64">import org.apache.yetus.audience.InterfaceAudience;</span>
<span class="source-line-no">065</span><span id="line-65">import org.slf4j.Logger;</span>
<span class="source-line-no">066</span><span id="line-66">import org.slf4j.LoggerFactory;</span>
<span class="source-line-no">067</span><span id="line-67"></span>
<span class="source-line-no">068</span><span id="line-68">/*</span>
<span class="source-line-no">069</span><span id="line-69"> * The CompactionTool allows to execute a compaction specifying a:</span>
<span class="source-line-no">070</span><span id="line-70"> * &lt;ul&gt;</span>
<span class="source-line-no">071</span><span id="line-71"> * &lt;li&gt;table folder (all regions and families will be compacted)</span>
<span class="source-line-no">072</span><span id="line-72"> * &lt;li&gt;region folder (all families in the region will be compacted)</span>
<span class="source-line-no">073</span><span id="line-73"> * &lt;li&gt;family folder (the store files will be compacted)</span>
<span class="source-line-no">074</span><span id="line-74"> * &lt;/ul&gt;</span>
<span class="source-line-no">075</span><span id="line-75"> */</span>
<span class="source-line-no">076</span><span id="line-76">@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.TOOLS)</span>
<span class="source-line-no">077</span><span id="line-77">public class CompactionTool extends Configured implements Tool {</span>
<span class="source-line-no">078</span><span id="line-78"> private static final Logger LOG = LoggerFactory.getLogger(CompactionTool.class);</span>
<span class="source-line-no">079</span><span id="line-79"></span>
<span class="source-line-no">080</span><span id="line-80"> private final static String CONF_COMPACT_ONCE = "hbase.compactiontool.compact.once";</span>
<span class="source-line-no">081</span><span id="line-81"> private final static String CONF_COMPACT_MAJOR = "hbase.compactiontool.compact.major";</span>
<span class="source-line-no">082</span><span id="line-82"> private final static String CONF_DELETE_COMPACTED = "hbase.compactiontool.delete";</span>
<span class="source-line-no">083</span><span id="line-83"></span>
<span class="source-line-no">084</span><span id="line-84"> /**</span>
<span class="source-line-no">085</span><span id="line-85"> * Class responsible to execute the Compaction on the specified path. The path can be a table,</span>
<span class="source-line-no">086</span><span id="line-86"> * region or family directory.</span>
<span class="source-line-no">087</span><span id="line-87"> */</span>
<span class="source-line-no">088</span><span id="line-88"> private static class CompactionWorker {</span>
<span class="source-line-no">089</span><span id="line-89"> private final boolean deleteCompacted;</span>
<span class="source-line-no">090</span><span id="line-90"> private final Configuration conf;</span>
<span class="source-line-no">091</span><span id="line-91"> private final FileSystem fs;</span>
<span class="source-line-no">092</span><span id="line-92"></span>
<span class="source-line-no">093</span><span id="line-93"> public CompactionWorker(final FileSystem fs, final Configuration conf) {</span>
<span class="source-line-no">094</span><span id="line-94"> this.conf = conf;</span>
<span class="source-line-no">095</span><span id="line-95"> this.deleteCompacted = conf.getBoolean(CONF_DELETE_COMPACTED, false);</span>
<span class="source-line-no">096</span><span id="line-96"> this.fs = fs;</span>
<span class="source-line-no">097</span><span id="line-97"> }</span>
<span class="source-line-no">098</span><span id="line-98"></span>
<span class="source-line-no">099</span><span id="line-99"> /**</span>
<span class="source-line-no">100</span><span id="line-100"> * Execute the compaction on the specified path.</span>
<span class="source-line-no">101</span><span id="line-101"> * @param path Directory path on which to run compaction.</span>
<span class="source-line-no">102</span><span id="line-102"> * @param compactOnce Execute just a single step of compaction.</span>
<span class="source-line-no">103</span><span id="line-103"> * @param major Request major compaction.</span>
<span class="source-line-no">104</span><span id="line-104"> */</span>
<span class="source-line-no">105</span><span id="line-105"> public void compact(final Path path, final boolean compactOnce, final boolean major)</span>
<span class="source-line-no">106</span><span id="line-106"> throws IOException {</span>
<span class="source-line-no">107</span><span id="line-107"> if (isFamilyDir(fs, path)) {</span>
<span class="source-line-no">108</span><span id="line-108"> Path regionDir = path.getParent();</span>
<span class="source-line-no">109</span><span id="line-109"> Path tableDir = regionDir.getParent();</span>
<span class="source-line-no">110</span><span id="line-110"> TableDescriptor htd = FSTableDescriptors.getTableDescriptorFromFs(fs, tableDir);</span>
<span class="source-line-no">111</span><span id="line-111"> RegionInfo hri = HRegionFileSystem.loadRegionInfoFileContent(fs, regionDir);</span>
<span class="source-line-no">112</span><span id="line-112"> compactStoreFiles(tableDir, htd, hri, path.getName(), compactOnce, major);</span>
<span class="source-line-no">113</span><span id="line-113"> } else if (isRegionDir(fs, path)) {</span>
<span class="source-line-no">114</span><span id="line-114"> Path tableDir = path.getParent();</span>
<span class="source-line-no">115</span><span id="line-115"> TableDescriptor htd = FSTableDescriptors.getTableDescriptorFromFs(fs, tableDir);</span>
<span class="source-line-no">116</span><span id="line-116"> compactRegion(tableDir, htd, path, compactOnce, major);</span>
<span class="source-line-no">117</span><span id="line-117"> } else if (isTableDir(fs, path)) {</span>
<span class="source-line-no">118</span><span id="line-118"> compactTable(path, compactOnce, major);</span>
<span class="source-line-no">119</span><span id="line-119"> } else {</span>
<span class="source-line-no">120</span><span id="line-120"> throw new IOException(</span>
<span class="source-line-no">121</span><span id="line-121"> "Specified path is not a table, region or family directory. path=" + path);</span>
<span class="source-line-no">122</span><span id="line-122"> }</span>
<span class="source-line-no">123</span><span id="line-123"> }</span>
<span class="source-line-no">124</span><span id="line-124"></span>
<span class="source-line-no">125</span><span id="line-125"> private void compactTable(final Path tableDir, final boolean compactOnce, final boolean major)</span>
<span class="source-line-no">126</span><span id="line-126"> throws IOException {</span>
<span class="source-line-no">127</span><span id="line-127"> TableDescriptor htd = FSTableDescriptors.getTableDescriptorFromFs(fs, tableDir);</span>
<span class="source-line-no">128</span><span id="line-128"> for (Path regionDir : FSUtils.getRegionDirs(fs, tableDir)) {</span>
<span class="source-line-no">129</span><span id="line-129"> compactRegion(tableDir, htd, regionDir, compactOnce, major);</span>
<span class="source-line-no">130</span><span id="line-130"> }</span>
<span class="source-line-no">131</span><span id="line-131"> }</span>
<span class="source-line-no">132</span><span id="line-132"></span>
<span class="source-line-no">133</span><span id="line-133"> private void compactRegion(final Path tableDir, final TableDescriptor htd, final Path regionDir,</span>
<span class="source-line-no">134</span><span id="line-134"> final boolean compactOnce, final boolean major) throws IOException {</span>
<span class="source-line-no">135</span><span id="line-135"> RegionInfo hri = HRegionFileSystem.loadRegionInfoFileContent(fs, regionDir);</span>
<span class="source-line-no">136</span><span id="line-136"> for (Path familyDir : FSUtils.getFamilyDirs(fs, regionDir)) {</span>
<span class="source-line-no">137</span><span id="line-137"> compactStoreFiles(tableDir, htd, hri, familyDir.getName(), compactOnce, major);</span>
<span class="source-line-no">138</span><span id="line-138"> }</span>
<span class="source-line-no">139</span><span id="line-139"> }</span>
<span class="source-line-no">140</span><span id="line-140"></span>
<span class="source-line-no">141</span><span id="line-141"> /**</span>
<span class="source-line-no">142</span><span id="line-142"> * Execute the actual compaction job. If the compact once flag is not specified, execute the</span>
<span class="source-line-no">143</span><span id="line-143"> * compaction until no more compactions are needed. Uses the Configuration settings provided.</span>
<span class="source-line-no">144</span><span id="line-144"> */</span>
<span class="source-line-no">145</span><span id="line-145"> private void compactStoreFiles(final Path tableDir, final TableDescriptor htd,</span>
<span class="source-line-no">146</span><span id="line-146"> final RegionInfo hri, final String familyName, final boolean compactOnce, final boolean major)</span>
<span class="source-line-no">147</span><span id="line-147"> throws IOException {</span>
<span class="source-line-no">148</span><span id="line-148"> HStore store = getStore(conf, fs, tableDir, htd, hri, familyName);</span>
<span class="source-line-no">149</span><span id="line-149"> LOG.info("Compact table=" + htd.getTableName() + " region=" + hri.getRegionNameAsString()</span>
<span class="source-line-no">150</span><span id="line-150"> + " family=" + familyName);</span>
<span class="source-line-no">151</span><span id="line-151"> if (major) {</span>
<span class="source-line-no">152</span><span id="line-152"> store.triggerMajorCompaction();</span>
<span class="source-line-no">153</span><span id="line-153"> }</span>
<span class="source-line-no">154</span><span id="line-154"> do {</span>
<span class="source-line-no">155</span><span id="line-155"> Optional&lt;CompactionContext&gt; compaction =</span>
<span class="source-line-no">156</span><span id="line-156"> store.requestCompaction(PRIORITY_USER, CompactionLifeCycleTracker.DUMMY, null);</span>
<span class="source-line-no">157</span><span id="line-157"> if (!compaction.isPresent()) {</span>
<span class="source-line-no">158</span><span id="line-158"> break;</span>
<span class="source-line-no">159</span><span id="line-159"> }</span>
<span class="source-line-no">160</span><span id="line-160"> List&lt;HStoreFile&gt; storeFiles =</span>
<span class="source-line-no">161</span><span id="line-161"> store.compact(compaction.get(), NoLimitThroughputController.INSTANCE, null);</span>
<span class="source-line-no">162</span><span id="line-162"> if (storeFiles != null &amp;&amp; !storeFiles.isEmpty()) {</span>
<span class="source-line-no">163</span><span id="line-163"> if (deleteCompacted) {</span>
<span class="source-line-no">164</span><span id="line-164"> for (HStoreFile storeFile : storeFiles) {</span>
<span class="source-line-no">165</span><span id="line-165"> fs.delete(storeFile.getPath(), false);</span>
<span class="source-line-no">166</span><span id="line-166"> }</span>
<span class="source-line-no">167</span><span id="line-167"> }</span>
<span class="source-line-no">168</span><span id="line-168"> }</span>
<span class="source-line-no">169</span><span id="line-169"> } while (store.needsCompaction() &amp;&amp; !compactOnce);</span>
<span class="source-line-no">170</span><span id="line-170"> // We need to close the store properly, to make sure it will archive compacted files</span>
<span class="source-line-no">171</span><span id="line-171"> store.close();</span>
<span class="source-line-no">172</span><span id="line-172"> }</span>
<span class="source-line-no">173</span><span id="line-173"></span>
<span class="source-line-no">174</span><span id="line-174"> private static HStore getStore(final Configuration conf, final FileSystem fs,</span>
<span class="source-line-no">175</span><span id="line-175"> final Path tableDir, final TableDescriptor htd, final RegionInfo hri, final String familyName)</span>
<span class="source-line-no">176</span><span id="line-176"> throws IOException {</span>
<span class="source-line-no">177</span><span id="line-177"> HRegionFileSystem regionFs = new HRegionFileSystem(conf, fs, tableDir, hri);</span>
<span class="source-line-no">178</span><span id="line-178"> HRegion region = new HRegion(regionFs, null, conf, htd, null);</span>
<span class="source-line-no">179</span><span id="line-179"> return new HStore(region, htd.getColumnFamily(Bytes.toBytes(familyName)), conf, false);</span>
<span class="source-line-no">180</span><span id="line-180"> }</span>
<span class="source-line-no">181</span><span id="line-181"> }</span>
<span class="source-line-no">182</span><span id="line-182"></span>
<span class="source-line-no">183</span><span id="line-183"> private static boolean isRegionDir(final FileSystem fs, final Path path) throws IOException {</span>
<span class="source-line-no">184</span><span id="line-184"> Path regionInfo = new Path(path, HRegionFileSystem.REGION_INFO_FILE);</span>
<span class="source-line-no">185</span><span id="line-185"> return fs.exists(regionInfo);</span>
<span class="source-line-no">186</span><span id="line-186"> }</span>
<span class="source-line-no">187</span><span id="line-187"></span>
<span class="source-line-no">188</span><span id="line-188"> private static boolean isTableDir(final FileSystem fs, final Path path) throws IOException {</span>
<span class="source-line-no">189</span><span id="line-189"> return FSTableDescriptors.isTableDir(fs, path);</span>
<span class="source-line-no">190</span><span id="line-190"> }</span>
<span class="source-line-no">191</span><span id="line-191"></span>
<span class="source-line-no">192</span><span id="line-192"> private static boolean isFamilyDir(final FileSystem fs, final Path path) throws IOException {</span>
<span class="source-line-no">193</span><span id="line-193"> return isRegionDir(fs, path.getParent());</span>
<span class="source-line-no">194</span><span id="line-194"> }</span>
<span class="source-line-no">195</span><span id="line-195"></span>
<span class="source-line-no">196</span><span id="line-196"> private static class CompactionMapper</span>
<span class="source-line-no">197</span><span id="line-197"> extends Mapper&lt;LongWritable, Text, NullWritable, NullWritable&gt; {</span>
<span class="source-line-no">198</span><span id="line-198"> private CompactionWorker compactor = null;</span>
<span class="source-line-no">199</span><span id="line-199"> private boolean compactOnce = false;</span>
<span class="source-line-no">200</span><span id="line-200"> private boolean major = false;</span>
<span class="source-line-no">201</span><span id="line-201"></span>
<span class="source-line-no">202</span><span id="line-202"> @Override</span>
<span class="source-line-no">203</span><span id="line-203"> public void setup(Context context) {</span>
<span class="source-line-no">204</span><span id="line-204"> Configuration conf = context.getConfiguration();</span>
<span class="source-line-no">205</span><span id="line-205"> compactOnce = conf.getBoolean(CONF_COMPACT_ONCE, false);</span>
<span class="source-line-no">206</span><span id="line-206"> major = conf.getBoolean(CONF_COMPACT_MAJOR, false);</span>
<span class="source-line-no">207</span><span id="line-207"></span>
<span class="source-line-no">208</span><span id="line-208"> try {</span>
<span class="source-line-no">209</span><span id="line-209"> FileSystem fs = CommonFSUtils.getRootDirFileSystem(conf);</span>
<span class="source-line-no">210</span><span id="line-210"> this.compactor = new CompactionWorker(fs, conf);</span>
<span class="source-line-no">211</span><span id="line-211"> } catch (IOException e) {</span>
<span class="source-line-no">212</span><span id="line-212"> throw new RuntimeException("Could not get the input FileSystem", e);</span>
<span class="source-line-no">213</span><span id="line-213"> }</span>
<span class="source-line-no">214</span><span id="line-214"> // Disable the MemStoreLAB as MemStore is not used by flow during compaction</span>
<span class="source-line-no">215</span><span id="line-215"> conf.setBoolean(MemStoreLAB.USEMSLAB_KEY, false);</span>
<span class="source-line-no">216</span><span id="line-216"> }</span>
<span class="source-line-no">217</span><span id="line-217"></span>
<span class="source-line-no">218</span><span id="line-218"> @Override</span>
<span class="source-line-no">219</span><span id="line-219"> public void map(LongWritable key, Text value, Context context)</span>
<span class="source-line-no">220</span><span id="line-220"> throws InterruptedException, IOException {</span>
<span class="source-line-no">221</span><span id="line-221"> Path path = new Path(value.toString());</span>
<span class="source-line-no">222</span><span id="line-222"> this.compactor.compact(path, compactOnce, major);</span>
<span class="source-line-no">223</span><span id="line-223"> }</span>
<span class="source-line-no">224</span><span id="line-224"> }</span>
<span class="source-line-no">225</span><span id="line-225"></span>
<span class="source-line-no">226</span><span id="line-226"> /**</span>
<span class="source-line-no">227</span><span id="line-227"> * Input format that uses store files block location as input split locality.</span>
<span class="source-line-no">228</span><span id="line-228"> */</span>
<span class="source-line-no">229</span><span id="line-229"> private static class CompactionInputFormat extends TextInputFormat {</span>
<span class="source-line-no">230</span><span id="line-230"> @Override</span>
<span class="source-line-no">231</span><span id="line-231"> protected boolean isSplitable(JobContext context, Path file) {</span>
<span class="source-line-no">232</span><span id="line-232"> return true;</span>
<span class="source-line-no">233</span><span id="line-233"> }</span>
<span class="source-line-no">234</span><span id="line-234"></span>
<span class="source-line-no">235</span><span id="line-235"> /**</span>
<span class="source-line-no">236</span><span id="line-236"> * Returns a split for each store files directory using the block location of each file as</span>
<span class="source-line-no">237</span><span id="line-237"> * locality reference.</span>
<span class="source-line-no">238</span><span id="line-238"> */</span>
<span class="source-line-no">239</span><span id="line-239"> @Override</span>
<span class="source-line-no">240</span><span id="line-240"> public List&lt;InputSplit&gt; getSplits(JobContext job) throws IOException {</span>
<span class="source-line-no">241</span><span id="line-241"> List&lt;InputSplit&gt; splits = new ArrayList&lt;&gt;();</span>
<span class="source-line-no">242</span><span id="line-242"> List&lt;FileStatus&gt; files = listStatus(job);</span>
<span class="source-line-no">243</span><span id="line-243"></span>
<span class="source-line-no">244</span><span id="line-244"> Text key = new Text();</span>
<span class="source-line-no">245</span><span id="line-245"> for (FileStatus file : files) {</span>
<span class="source-line-no">246</span><span id="line-246"> Path path = file.getPath();</span>
<span class="source-line-no">247</span><span id="line-247"> FileSystem fs = path.getFileSystem(job.getConfiguration());</span>
<span class="source-line-no">248</span><span id="line-248"> LineReader reader = new LineReader(fs.open(path));</span>
<span class="source-line-no">249</span><span id="line-249"> long pos = 0;</span>
<span class="source-line-no">250</span><span id="line-250"> int n;</span>
<span class="source-line-no">251</span><span id="line-251"> try {</span>
<span class="source-line-no">252</span><span id="line-252"> while ((n = reader.readLine(key)) &gt; 0) {</span>
<span class="source-line-no">253</span><span id="line-253"> String[] hosts = getStoreDirHosts(fs, path);</span>
<span class="source-line-no">254</span><span id="line-254"> splits.add(new FileSplit(path, pos, n, hosts));</span>
<span class="source-line-no">255</span><span id="line-255"> pos += n;</span>
<span class="source-line-no">256</span><span id="line-256"> }</span>
<span class="source-line-no">257</span><span id="line-257"> } finally {</span>
<span class="source-line-no">258</span><span id="line-258"> reader.close();</span>
<span class="source-line-no">259</span><span id="line-259"> }</span>
<span class="source-line-no">260</span><span id="line-260"> }</span>
<span class="source-line-no">261</span><span id="line-261"></span>
<span class="source-line-no">262</span><span id="line-262"> return splits;</span>
<span class="source-line-no">263</span><span id="line-263"> }</span>
<span class="source-line-no">264</span><span id="line-264"></span>
<span class="source-line-no">265</span><span id="line-265"> /**</span>
<span class="source-line-no">266</span><span id="line-266"> * return the top hosts of the store files, used by the Split</span>
<span class="source-line-no">267</span><span id="line-267"> */</span>
<span class="source-line-no">268</span><span id="line-268"> private static String[] getStoreDirHosts(final FileSystem fs, final Path path)</span>
<span class="source-line-no">269</span><span id="line-269"> throws IOException {</span>
<span class="source-line-no">270</span><span id="line-270"> FileStatus[] files = CommonFSUtils.listStatus(fs, path);</span>
<span class="source-line-no">271</span><span id="line-271"> if (files == null) {</span>
<span class="source-line-no">272</span><span id="line-272"> return new String[] {};</span>
<span class="source-line-no">273</span><span id="line-273"> }</span>
<span class="source-line-no">274</span><span id="line-274"></span>
<span class="source-line-no">275</span><span id="line-275"> HDFSBlocksDistribution hdfsBlocksDistribution = new HDFSBlocksDistribution();</span>
<span class="source-line-no">276</span><span id="line-276"> for (FileStatus hfileStatus : files) {</span>
<span class="source-line-no">277</span><span id="line-277"> HDFSBlocksDistribution storeFileBlocksDistribution =</span>
<span class="source-line-no">278</span><span id="line-278"> FSUtils.computeHDFSBlocksDistribution(fs, hfileStatus, 0, hfileStatus.getLen());</span>
<span class="source-line-no">279</span><span id="line-279"> hdfsBlocksDistribution.add(storeFileBlocksDistribution);</span>
<span class="source-line-no">280</span><span id="line-280"> }</span>
<span class="source-line-no">281</span><span id="line-281"></span>
<span class="source-line-no">282</span><span id="line-282"> List&lt;String&gt; hosts = hdfsBlocksDistribution.getTopHosts();</span>
<span class="source-line-no">283</span><span id="line-283"> return hosts.toArray(new String[hosts.size()]);</span>
<span class="source-line-no">284</span><span id="line-284"> }</span>
<span class="source-line-no">285</span><span id="line-285"></span>
<span class="source-line-no">286</span><span id="line-286"> /**</span>
<span class="source-line-no">287</span><span id="line-287"> * Create the input file for the given directories to compact. The file is a TextFile with each</span>
<span class="source-line-no">288</span><span id="line-288"> * line corrisponding to a store files directory to compact.</span>
<span class="source-line-no">289</span><span id="line-289"> */</span>
<span class="source-line-no">290</span><span id="line-290"> public static List&lt;Path&gt; createInputFile(final FileSystem fs, final FileSystem stagingFs,</span>
<span class="source-line-no">291</span><span id="line-291"> final Path path, final Set&lt;Path&gt; toCompactDirs) throws IOException {</span>
<span class="source-line-no">292</span><span id="line-292"> // Extract the list of store dirs</span>
<span class="source-line-no">293</span><span id="line-293"> List&lt;Path&gt; storeDirs = new LinkedList&lt;&gt;();</span>
<span class="source-line-no">294</span><span id="line-294"> for (Path compactDir : toCompactDirs) {</span>
<span class="source-line-no">295</span><span id="line-295"> if (isFamilyDir(fs, compactDir)) {</span>
<span class="source-line-no">296</span><span id="line-296"> storeDirs.add(compactDir);</span>
<span class="source-line-no">297</span><span id="line-297"> } else if (isRegionDir(fs, compactDir)) {</span>
<span class="source-line-no">298</span><span id="line-298"> storeDirs.addAll(FSUtils.getFamilyDirs(fs, compactDir));</span>
<span class="source-line-no">299</span><span id="line-299"> } else if (isTableDir(fs, compactDir)) {</span>
<span class="source-line-no">300</span><span id="line-300"> // Lookup regions</span>
<span class="source-line-no">301</span><span id="line-301"> for (Path regionDir : FSUtils.getRegionDirs(fs, compactDir)) {</span>
<span class="source-line-no">302</span><span id="line-302"> storeDirs.addAll(FSUtils.getFamilyDirs(fs, regionDir));</span>
<span class="source-line-no">303</span><span id="line-303"> }</span>
<span class="source-line-no">304</span><span id="line-304"> } else {</span>
<span class="source-line-no">305</span><span id="line-305"> throw new IOException(</span>
<span class="source-line-no">306</span><span id="line-306"> "Specified path is not a table, region or family directory. path=" + compactDir);</span>
<span class="source-line-no">307</span><span id="line-307"> }</span>
<span class="source-line-no">308</span><span id="line-308"> }</span>
<span class="source-line-no">309</span><span id="line-309"></span>
<span class="source-line-no">310</span><span id="line-310"> // Write Input File</span>
<span class="source-line-no">311</span><span id="line-311"> FSDataOutputStream stream = stagingFs.create(path);</span>
<span class="source-line-no">312</span><span id="line-312"> LOG.info("Create input file=" + path + " with " + storeDirs.size() + " dirs to compact.");</span>
<span class="source-line-no">313</span><span id="line-313"> try {</span>
<span class="source-line-no">314</span><span id="line-314"> final byte[] newLine = Bytes.toBytes("\n");</span>
<span class="source-line-no">315</span><span id="line-315"> for (Path storeDir : storeDirs) {</span>
<span class="source-line-no">316</span><span id="line-316"> stream.write(Bytes.toBytes(storeDir.toString()));</span>
<span class="source-line-no">317</span><span id="line-317"> stream.write(newLine);</span>
<span class="source-line-no">318</span><span id="line-318"> }</span>
<span class="source-line-no">319</span><span id="line-319"> } finally {</span>
<span class="source-line-no">320</span><span id="line-320"> stream.close();</span>
<span class="source-line-no">321</span><span id="line-321"> }</span>
<span class="source-line-no">322</span><span id="line-322"> return storeDirs;</span>
<span class="source-line-no">323</span><span id="line-323"> }</span>
<span class="source-line-no">324</span><span id="line-324"> }</span>
<span class="source-line-no">325</span><span id="line-325"></span>
<span class="source-line-no">326</span><span id="line-326"> /**</span>
<span class="source-line-no">327</span><span id="line-327"> * Execute compaction, using a Map-Reduce job.</span>
<span class="source-line-no">328</span><span id="line-328"> */</span>
<span class="source-line-no">329</span><span id="line-329"> private int doMapReduce(final FileSystem fs, final Set&lt;Path&gt; toCompactDirs,</span>
<span class="source-line-no">330</span><span id="line-330"> final boolean compactOnce, final boolean major) throws Exception {</span>
<span class="source-line-no">331</span><span id="line-331"> Configuration conf = getConf();</span>
<span class="source-line-no">332</span><span id="line-332"> conf.setBoolean(CONF_COMPACT_ONCE, compactOnce);</span>
<span class="source-line-no">333</span><span id="line-333"> conf.setBoolean(CONF_COMPACT_MAJOR, major);</span>
<span class="source-line-no">334</span><span id="line-334"></span>
<span class="source-line-no">335</span><span id="line-335"> Job job = new Job(conf);</span>
<span class="source-line-no">336</span><span id="line-336"> job.setJobName("CompactionTool");</span>
<span class="source-line-no">337</span><span id="line-337"> job.setJarByClass(CompactionTool.class);</span>
<span class="source-line-no">338</span><span id="line-338"> job.setMapperClass(CompactionMapper.class);</span>
<span class="source-line-no">339</span><span id="line-339"> job.setInputFormatClass(CompactionInputFormat.class);</span>
<span class="source-line-no">340</span><span id="line-340"> job.setOutputFormatClass(NullOutputFormat.class);</span>
<span class="source-line-no">341</span><span id="line-341"> job.setMapSpeculativeExecution(false);</span>
<span class="source-line-no">342</span><span id="line-342"> job.setNumReduceTasks(0);</span>
<span class="source-line-no">343</span><span id="line-343"></span>
<span class="source-line-no">344</span><span id="line-344"> // add dependencies (including HBase ones)</span>
<span class="source-line-no">345</span><span id="line-345"> TableMapReduceUtil.addDependencyJars(job);</span>
<span class="source-line-no">346</span><span id="line-346"></span>
<span class="source-line-no">347</span><span id="line-347"> Path stagingDir = JobUtil.getQualifiedStagingDir(conf);</span>
<span class="source-line-no">348</span><span id="line-348"> FileSystem stagingFs = stagingDir.getFileSystem(conf);</span>
<span class="source-line-no">349</span><span id="line-349"> try {</span>
<span class="source-line-no">350</span><span id="line-350"> // Create input file with the store dirs</span>
<span class="source-line-no">351</span><span id="line-351"> Path inputPath = new Path(stagingDir, "compact-" + EnvironmentEdgeManager.currentTime());</span>
<span class="source-line-no">352</span><span id="line-352"> List&lt;Path&gt; storeDirs =</span>
<span class="source-line-no">353</span><span id="line-353"> CompactionInputFormat.createInputFile(fs, stagingFs, inputPath, toCompactDirs);</span>
<span class="source-line-no">354</span><span id="line-354"> CompactionInputFormat.addInputPath(job, inputPath);</span>
<span class="source-line-no">355</span><span id="line-355"></span>
<span class="source-line-no">356</span><span id="line-356"> // Initialize credential for secure cluster</span>
<span class="source-line-no">357</span><span id="line-357"> TableMapReduceUtil.initCredentials(job);</span>
<span class="source-line-no">358</span><span id="line-358"> // Despite the method name this will get delegation token for the filesystem</span>
<span class="source-line-no">359</span><span id="line-359"> TokenCache.obtainTokensForNamenodes(job.getCredentials(), storeDirs.toArray(new Path[0]),</span>
<span class="source-line-no">360</span><span id="line-360"> conf);</span>
<span class="source-line-no">361</span><span id="line-361"></span>
<span class="source-line-no">362</span><span id="line-362"> // Start the MR Job and wait</span>
<span class="source-line-no">363</span><span id="line-363"> return job.waitForCompletion(true) ? 0 : 1;</span>
<span class="source-line-no">364</span><span id="line-364"> } finally {</span>
<span class="source-line-no">365</span><span id="line-365"> fs.delete(stagingDir, true);</span>
<span class="source-line-no">366</span><span id="line-366"> }</span>
<span class="source-line-no">367</span><span id="line-367"> }</span>
<span class="source-line-no">368</span><span id="line-368"></span>
<span class="source-line-no">369</span><span id="line-369"> /**</span>
<span class="source-line-no">370</span><span id="line-370"> * Execute compaction, from this client, one path at the time.</span>
<span class="source-line-no">371</span><span id="line-371"> */</span>
<span class="source-line-no">372</span><span id="line-372"> private int doClient(final FileSystem fs, final Set&lt;Path&gt; toCompactDirs,</span>
<span class="source-line-no">373</span><span id="line-373"> final boolean compactOnce, final boolean major) throws IOException {</span>
<span class="source-line-no">374</span><span id="line-374"> // Disable the MemStoreLAB as MemStore is not used by flow during compaction</span>
<span class="source-line-no">375</span><span id="line-375"> getConf().setBoolean(MemStoreLAB.USEMSLAB_KEY, false);</span>
<span class="source-line-no">376</span><span id="line-376"> CompactionWorker worker = new CompactionWorker(fs, getConf());</span>
<span class="source-line-no">377</span><span id="line-377"> for (Path path : toCompactDirs) {</span>
<span class="source-line-no">378</span><span id="line-378"> worker.compact(path, compactOnce, major);</span>
<span class="source-line-no">379</span><span id="line-379"> }</span>
<span class="source-line-no">380</span><span id="line-380"> return 0;</span>
<span class="source-line-no">381</span><span id="line-381"> }</span>
<span class="source-line-no">382</span><span id="line-382"></span>
<span class="source-line-no">383</span><span id="line-383"> @Override</span>
<span class="source-line-no">384</span><span id="line-384"> public int run(String[] args) throws Exception {</span>
<span class="source-line-no">385</span><span id="line-385"> Set&lt;Path&gt; toCompactDirs = new HashSet&lt;&gt;();</span>
<span class="source-line-no">386</span><span id="line-386"> boolean compactOnce = false;</span>
<span class="source-line-no">387</span><span id="line-387"> boolean major = false;</span>
<span class="source-line-no">388</span><span id="line-388"> boolean mapred = false;</span>
<span class="source-line-no">389</span><span id="line-389"></span>
<span class="source-line-no">390</span><span id="line-390"> Configuration conf = getConf();</span>
<span class="source-line-no">391</span><span id="line-391"> FileSystem fs = CommonFSUtils.getRootDirFileSystem(conf);</span>
<span class="source-line-no">392</span><span id="line-392"></span>
<span class="source-line-no">393</span><span id="line-393"> try {</span>
<span class="source-line-no">394</span><span id="line-394"> for (int i = 0; i &lt; args.length; ++i) {</span>
<span class="source-line-no">395</span><span id="line-395"> String opt = args[i];</span>
<span class="source-line-no">396</span><span id="line-396"> if (opt.equals("-compactOnce")) {</span>
<span class="source-line-no">397</span><span id="line-397"> compactOnce = true;</span>
<span class="source-line-no">398</span><span id="line-398"> } else if (opt.equals("-major")) {</span>
<span class="source-line-no">399</span><span id="line-399"> major = true;</span>
<span class="source-line-no">400</span><span id="line-400"> } else if (opt.equals("-mapred")) {</span>
<span class="source-line-no">401</span><span id="line-401"> mapred = true;</span>
<span class="source-line-no">402</span><span id="line-402"> } else if (!opt.startsWith("-")) {</span>
<span class="source-line-no">403</span><span id="line-403"> Path path = new Path(opt);</span>
<span class="source-line-no">404</span><span id="line-404"> FileStatus status = fs.getFileStatus(path);</span>
<span class="source-line-no">405</span><span id="line-405"> if (!status.isDirectory()) {</span>
<span class="source-line-no">406</span><span id="line-406"> printUsage("Specified path is not a directory. path=" + path);</span>
<span class="source-line-no">407</span><span id="line-407"> return 1;</span>
<span class="source-line-no">408</span><span id="line-408"> }</span>
<span class="source-line-no">409</span><span id="line-409"> toCompactDirs.add(path);</span>
<span class="source-line-no">410</span><span id="line-410"> } else {</span>
<span class="source-line-no">411</span><span id="line-411"> printUsage();</span>
<span class="source-line-no">412</span><span id="line-412"> }</span>
<span class="source-line-no">413</span><span id="line-413"> }</span>
<span class="source-line-no">414</span><span id="line-414"> } catch (Exception e) {</span>
<span class="source-line-no">415</span><span id="line-415"> printUsage(e.getMessage());</span>
<span class="source-line-no">416</span><span id="line-416"> return 1;</span>
<span class="source-line-no">417</span><span id="line-417"> }</span>
<span class="source-line-no">418</span><span id="line-418"></span>
<span class="source-line-no">419</span><span id="line-419"> if (toCompactDirs.isEmpty()) {</span>
<span class="source-line-no">420</span><span id="line-420"> printUsage("No directories to compact specified.");</span>
<span class="source-line-no">421</span><span id="line-421"> return 1;</span>
<span class="source-line-no">422</span><span id="line-422"> }</span>
<span class="source-line-no">423</span><span id="line-423"></span>
<span class="source-line-no">424</span><span id="line-424"> // Execute compaction!</span>
<span class="source-line-no">425</span><span id="line-425"> if (mapred) {</span>
<span class="source-line-no">426</span><span id="line-426"> return doMapReduce(fs, toCompactDirs, compactOnce, major);</span>
<span class="source-line-no">427</span><span id="line-427"> } else {</span>
<span class="source-line-no">428</span><span id="line-428"> return doClient(fs, toCompactDirs, compactOnce, major);</span>
<span class="source-line-no">429</span><span id="line-429"> }</span>
<span class="source-line-no">430</span><span id="line-430"> }</span>
<span class="source-line-no">431</span><span id="line-431"></span>
<span class="source-line-no">432</span><span id="line-432"> private void printUsage() {</span>
<span class="source-line-no">433</span><span id="line-433"> printUsage(null);</span>
<span class="source-line-no">434</span><span id="line-434"> }</span>
<span class="source-line-no">435</span><span id="line-435"></span>
<span class="source-line-no">436</span><span id="line-436"> private void printUsage(final String message) {</span>
<span class="source-line-no">437</span><span id="line-437"> if (message != null &amp;&amp; message.length() &gt; 0) {</span>
<span class="source-line-no">438</span><span id="line-438"> System.err.println(message);</span>
<span class="source-line-no">439</span><span id="line-439"> }</span>
<span class="source-line-no">440</span><span id="line-440"> System.err.println("Usage: java " + this.getClass().getName() + " \\");</span>
<span class="source-line-no">441</span><span id="line-441"> System.err.println(" [-compactOnce] [-major] [-mapred] [-D&lt;property=value&gt;]* files...");</span>
<span class="source-line-no">442</span><span id="line-442"> System.err.println();</span>
<span class="source-line-no">443</span><span id="line-443"> System.err.println("Options:");</span>
<span class="source-line-no">444</span><span id="line-444"> System.err.println(" mapred Use MapReduce to run compaction.");</span>
<span class="source-line-no">445</span><span id="line-445"> System.err.println(" compactOnce Execute just one compaction step. (default: while needed)");</span>
<span class="source-line-no">446</span><span id="line-446"> System.err.println(" major Trigger major compaction.");</span>
<span class="source-line-no">447</span><span id="line-447"> System.err.println();</span>
<span class="source-line-no">448</span><span id="line-448"> System.err.println("Note: -D properties will be applied to the conf used. ");</span>
<span class="source-line-no">449</span><span id="line-449"> System.err.println("For example: ");</span>
<span class="source-line-no">450</span><span id="line-450"> System.err</span>
<span class="source-line-no">451</span><span id="line-451"> .println(" To stop delete of compacted file, pass -D" + CONF_DELETE_COMPACTED + "=false");</span>
<span class="source-line-no">452</span><span id="line-452"> System.err.println();</span>
<span class="source-line-no">453</span><span id="line-453"> System.err.println("Examples:");</span>
<span class="source-line-no">454</span><span id="line-454"> System.err.println(" To compact the full 'TestTable' using MapReduce:");</span>
<span class="source-line-no">455</span><span id="line-455"> System.err.println(</span>
<span class="source-line-no">456</span><span id="line-456"> " $ hbase " + this.getClass().getName() + " -mapred hdfs://hbase/data/default/TestTable");</span>
<span class="source-line-no">457</span><span id="line-457"> System.err.println();</span>
<span class="source-line-no">458</span><span id="line-458"> System.err.println(" To compact column family 'x' of the table 'TestTable' region 'abc':");</span>
<span class="source-line-no">459</span><span id="line-459"> System.err.println(</span>
<span class="source-line-no">460</span><span id="line-460"> " $ hbase " + this.getClass().getName() + " hdfs://hbase/data/default/TestTable/abc/x");</span>
<span class="source-line-no">461</span><span id="line-461"> }</span>
<span class="source-line-no">462</span><span id="line-462"></span>
<span class="source-line-no">463</span><span id="line-463"> public static void main(String[] args) throws Exception {</span>
<span class="source-line-no">464</span><span id="line-464"> System.exit(ToolRunner.run(HBaseConfiguration.create(), new CompactionTool(), args));</span>
<span class="source-line-no">465</span><span id="line-465"> }</span>
<span class="source-line-no">466</span><span id="line-466">}</span>
</pre>
</div>
</main>
</body>
</html>