blob: e72019e3351c5cf934765d8cb017422b4b915ac7 [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.regionserver;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import static org.apache.hadoop.hbase.regionserver.Store.PRIORITY_USER;<a name="line.20"></a>
<span class="sourceLineNo">021</span><a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.IOException;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.ArrayList;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.util.HashSet;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.LinkedList;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.List;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.util.Optional;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import java.util.Set;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.conf.Configuration;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configured;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.FSDataOutputStream;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.fs.FileStatus;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.FileSystem;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.Path;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.HBaseInterfaceAudience;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.HDFSBlocksDistribution;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.client.RegionInfo;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.mapreduce.JobUtil;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.regionserver.compactions.CompactionContext;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.regionserver.compactions.CompactionLifeCycleTracker;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.regionserver.throttle.NoLimitThroughputController;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.util.CommonFSUtils;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.util.FSTableDescriptors;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.util.FSUtils;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.hadoop.io.LongWritable;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.hadoop.io.NullWritable;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.apache.hadoop.io.Text;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.apache.hadoop.mapreduce.InputSplit;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.apache.hadoop.mapreduce.Job;<a name="line.54"></a>
<span class="sourceLineNo">055</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.55"></a>
<span class="sourceLineNo">056</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.56"></a>
<span class="sourceLineNo">057</span>import org.apache.hadoop.mapreduce.lib.input.FileSplit;<a name="line.57"></a>
<span class="sourceLineNo">058</span>import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;<a name="line.58"></a>
<span class="sourceLineNo">059</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.59"></a>
<span class="sourceLineNo">060</span>import org.apache.hadoop.mapreduce.security.TokenCache;<a name="line.60"></a>
<span class="sourceLineNo">061</span>import org.apache.hadoop.util.LineReader;<a name="line.61"></a>
<span class="sourceLineNo">062</span>import org.apache.hadoop.util.Tool;<a name="line.62"></a>
<span class="sourceLineNo">063</span>import org.apache.hadoop.util.ToolRunner;<a name="line.63"></a>
<span class="sourceLineNo">064</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.64"></a>
<span class="sourceLineNo">065</span>import org.slf4j.Logger;<a name="line.65"></a>
<span class="sourceLineNo">066</span>import org.slf4j.LoggerFactory;<a name="line.66"></a>
<span class="sourceLineNo">067</span><a name="line.67"></a>
<span class="sourceLineNo">068</span>/*<a name="line.68"></a>
<span class="sourceLineNo">069</span> * The CompactionTool allows to execute a compaction specifying a:<a name="line.69"></a>
<span class="sourceLineNo">070</span> * &lt;ul&gt;<a name="line.70"></a>
<span class="sourceLineNo">071</span> * &lt;li&gt;table folder (all regions and families will be compacted)<a name="line.71"></a>
<span class="sourceLineNo">072</span> * &lt;li&gt;region folder (all families in the region will be compacted)<a name="line.72"></a>
<span class="sourceLineNo">073</span> * &lt;li&gt;family folder (the store files will be compacted)<a name="line.73"></a>
<span class="sourceLineNo">074</span> * &lt;/ul&gt;<a name="line.74"></a>
<span class="sourceLineNo">075</span> */<a name="line.75"></a>
<span class="sourceLineNo">076</span>@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.TOOLS)<a name="line.76"></a>
<span class="sourceLineNo">077</span>public class CompactionTool extends Configured implements Tool {<a name="line.77"></a>
<span class="sourceLineNo">078</span> private static final Logger LOG = LoggerFactory.getLogger(CompactionTool.class);<a name="line.78"></a>
<span class="sourceLineNo">079</span><a name="line.79"></a>
<span class="sourceLineNo">080</span> private final static String CONF_COMPACT_ONCE = "hbase.compactiontool.compact.once";<a name="line.80"></a>
<span class="sourceLineNo">081</span> private final static String CONF_COMPACT_MAJOR = "hbase.compactiontool.compact.major";<a name="line.81"></a>
<span class="sourceLineNo">082</span> private final static String CONF_DELETE_COMPACTED = "hbase.compactiontool.delete";<a name="line.82"></a>
<span class="sourceLineNo">083</span><a name="line.83"></a>
<span class="sourceLineNo">084</span> /**<a name="line.84"></a>
<span class="sourceLineNo">085</span> * Class responsible to execute the Compaction on the specified path.<a name="line.85"></a>
<span class="sourceLineNo">086</span> * The path can be a table, region or family directory.<a name="line.86"></a>
<span class="sourceLineNo">087</span> */<a name="line.87"></a>
<span class="sourceLineNo">088</span> private static class CompactionWorker {<a name="line.88"></a>
<span class="sourceLineNo">089</span> private final boolean deleteCompacted;<a name="line.89"></a>
<span class="sourceLineNo">090</span> private final Configuration conf;<a name="line.90"></a>
<span class="sourceLineNo">091</span> private final FileSystem fs;<a name="line.91"></a>
<span class="sourceLineNo">092</span><a name="line.92"></a>
<span class="sourceLineNo">093</span> public CompactionWorker(final FileSystem fs, final Configuration conf) {<a name="line.93"></a>
<span class="sourceLineNo">094</span> this.conf = conf;<a name="line.94"></a>
<span class="sourceLineNo">095</span> this.deleteCompacted = conf.getBoolean(CONF_DELETE_COMPACTED, false);<a name="line.95"></a>
<span class="sourceLineNo">096</span> this.fs = fs;<a name="line.96"></a>
<span class="sourceLineNo">097</span> }<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
<span class="sourceLineNo">099</span> /**<a name="line.99"></a>
<span class="sourceLineNo">100</span> * Execute the compaction on the specified path.<a name="line.100"></a>
<span class="sourceLineNo">101</span> *<a name="line.101"></a>
<span class="sourceLineNo">102</span> * @param path Directory path on which to run compaction.<a name="line.102"></a>
<span class="sourceLineNo">103</span> * @param compactOnce Execute just a single step of compaction.<a name="line.103"></a>
<span class="sourceLineNo">104</span> * @param major Request major compaction.<a name="line.104"></a>
<span class="sourceLineNo">105</span> */<a name="line.105"></a>
<span class="sourceLineNo">106</span> public void compact(final Path path, final boolean compactOnce, final boolean major)<a name="line.106"></a>
<span class="sourceLineNo">107</span> throws IOException {<a name="line.107"></a>
<span class="sourceLineNo">108</span> if (isFamilyDir(fs, path)) {<a name="line.108"></a>
<span class="sourceLineNo">109</span> Path regionDir = path.getParent();<a name="line.109"></a>
<span class="sourceLineNo">110</span> Path tableDir = regionDir.getParent();<a name="line.110"></a>
<span class="sourceLineNo">111</span> TableDescriptor htd = FSTableDescriptors.getTableDescriptorFromFs(fs, tableDir);<a name="line.111"></a>
<span class="sourceLineNo">112</span> RegionInfo hri = HRegionFileSystem.loadRegionInfoFileContent(fs, regionDir);<a name="line.112"></a>
<span class="sourceLineNo">113</span> compactStoreFiles(tableDir, htd, hri,<a name="line.113"></a>
<span class="sourceLineNo">114</span> path.getName(), compactOnce, major);<a name="line.114"></a>
<span class="sourceLineNo">115</span> } else if (isRegionDir(fs, path)) {<a name="line.115"></a>
<span class="sourceLineNo">116</span> Path tableDir = path.getParent();<a name="line.116"></a>
<span class="sourceLineNo">117</span> TableDescriptor htd = FSTableDescriptors.getTableDescriptorFromFs(fs, tableDir);<a name="line.117"></a>
<span class="sourceLineNo">118</span> compactRegion(tableDir, htd, path, compactOnce, major);<a name="line.118"></a>
<span class="sourceLineNo">119</span> } else if (isTableDir(fs, path)) {<a name="line.119"></a>
<span class="sourceLineNo">120</span> compactTable(path, compactOnce, major);<a name="line.120"></a>
<span class="sourceLineNo">121</span> } else {<a name="line.121"></a>
<span class="sourceLineNo">122</span> throw new IOException(<a name="line.122"></a>
<span class="sourceLineNo">123</span> "Specified path is not a table, region or family directory. path=" + path);<a name="line.123"></a>
<span class="sourceLineNo">124</span> }<a name="line.124"></a>
<span class="sourceLineNo">125</span> }<a name="line.125"></a>
<span class="sourceLineNo">126</span><a name="line.126"></a>
<span class="sourceLineNo">127</span> private void compactTable(final Path tableDir, final boolean compactOnce, final boolean major)<a name="line.127"></a>
<span class="sourceLineNo">128</span> throws IOException {<a name="line.128"></a>
<span class="sourceLineNo">129</span> TableDescriptor htd = FSTableDescriptors.getTableDescriptorFromFs(fs, tableDir);<a name="line.129"></a>
<span class="sourceLineNo">130</span> for (Path regionDir: FSUtils.getRegionDirs(fs, tableDir)) {<a name="line.130"></a>
<span class="sourceLineNo">131</span> compactRegion(tableDir, htd, regionDir, compactOnce, major);<a name="line.131"></a>
<span class="sourceLineNo">132</span> }<a name="line.132"></a>
<span class="sourceLineNo">133</span> }<a name="line.133"></a>
<span class="sourceLineNo">134</span><a name="line.134"></a>
<span class="sourceLineNo">135</span> private void compactRegion(final Path tableDir, final TableDescriptor htd,<a name="line.135"></a>
<span class="sourceLineNo">136</span> final Path regionDir, final boolean compactOnce, final boolean major)<a name="line.136"></a>
<span class="sourceLineNo">137</span> throws IOException {<a name="line.137"></a>
<span class="sourceLineNo">138</span> RegionInfo hri = HRegionFileSystem.loadRegionInfoFileContent(fs, regionDir);<a name="line.138"></a>
<span class="sourceLineNo">139</span> for (Path familyDir: FSUtils.getFamilyDirs(fs, regionDir)) {<a name="line.139"></a>
<span class="sourceLineNo">140</span> compactStoreFiles(tableDir, htd, hri, familyDir.getName(), compactOnce, major);<a name="line.140"></a>
<span class="sourceLineNo">141</span> }<a name="line.141"></a>
<span class="sourceLineNo">142</span> }<a name="line.142"></a>
<span class="sourceLineNo">143</span><a name="line.143"></a>
<span class="sourceLineNo">144</span> /**<a name="line.144"></a>
<span class="sourceLineNo">145</span> * Execute the actual compaction job.<a name="line.145"></a>
<span class="sourceLineNo">146</span> * If the compact once flag is not specified, execute the compaction until<a name="line.146"></a>
<span class="sourceLineNo">147</span> * no more compactions are needed. Uses the Configuration settings provided.<a name="line.147"></a>
<span class="sourceLineNo">148</span> */<a name="line.148"></a>
<span class="sourceLineNo">149</span> private void compactStoreFiles(final Path tableDir, final TableDescriptor htd,<a name="line.149"></a>
<span class="sourceLineNo">150</span> final RegionInfo hri, final String familyName, final boolean compactOnce,<a name="line.150"></a>
<span class="sourceLineNo">151</span> final boolean major) throws IOException {<a name="line.151"></a>
<span class="sourceLineNo">152</span> HStore store = getStore(conf, fs, tableDir, htd, hri, familyName);<a name="line.152"></a>
<span class="sourceLineNo">153</span> LOG.info("Compact table=" + htd.getTableName() +<a name="line.153"></a>
<span class="sourceLineNo">154</span> " region=" + hri.getRegionNameAsString() +<a name="line.154"></a>
<span class="sourceLineNo">155</span> " family=" + familyName);<a name="line.155"></a>
<span class="sourceLineNo">156</span> if (major) {<a name="line.156"></a>
<span class="sourceLineNo">157</span> store.triggerMajorCompaction();<a name="line.157"></a>
<span class="sourceLineNo">158</span> }<a name="line.158"></a>
<span class="sourceLineNo">159</span> do {<a name="line.159"></a>
<span class="sourceLineNo">160</span> Optional&lt;CompactionContext&gt; compaction =<a name="line.160"></a>
<span class="sourceLineNo">161</span> store.requestCompaction(PRIORITY_USER, CompactionLifeCycleTracker.DUMMY, null);<a name="line.161"></a>
<span class="sourceLineNo">162</span> if (!compaction.isPresent()) {<a name="line.162"></a>
<span class="sourceLineNo">163</span> break;<a name="line.163"></a>
<span class="sourceLineNo">164</span> }<a name="line.164"></a>
<span class="sourceLineNo">165</span> List&lt;HStoreFile&gt; storeFiles =<a name="line.165"></a>
<span class="sourceLineNo">166</span> store.compact(compaction.get(), NoLimitThroughputController.INSTANCE, null);<a name="line.166"></a>
<span class="sourceLineNo">167</span> if (storeFiles != null &amp;&amp; !storeFiles.isEmpty()) {<a name="line.167"></a>
<span class="sourceLineNo">168</span> if (deleteCompacted) {<a name="line.168"></a>
<span class="sourceLineNo">169</span> for (HStoreFile storeFile: storeFiles) {<a name="line.169"></a>
<span class="sourceLineNo">170</span> fs.delete(storeFile.getPath(), false);<a name="line.170"></a>
<span class="sourceLineNo">171</span> }<a name="line.171"></a>
<span class="sourceLineNo">172</span> }<a name="line.172"></a>
<span class="sourceLineNo">173</span> }<a name="line.173"></a>
<span class="sourceLineNo">174</span> } while (store.needsCompaction() &amp;&amp; !compactOnce);<a name="line.174"></a>
<span class="sourceLineNo">175</span> //We need to close the store properly, to make sure it will archive compacted files<a name="line.175"></a>
<span class="sourceLineNo">176</span> store.close();<a name="line.176"></a>
<span class="sourceLineNo">177</span> }<a name="line.177"></a>
<span class="sourceLineNo">178</span><a name="line.178"></a>
<span class="sourceLineNo">179</span> private static HStore getStore(final Configuration conf, final FileSystem fs,<a name="line.179"></a>
<span class="sourceLineNo">180</span> final Path tableDir, final TableDescriptor htd, final RegionInfo hri,<a name="line.180"></a>
<span class="sourceLineNo">181</span> final String familyName) throws IOException {<a name="line.181"></a>
<span class="sourceLineNo">182</span> HRegionFileSystem regionFs = new HRegionFileSystem(conf, fs, tableDir, hri);<a name="line.182"></a>
<span class="sourceLineNo">183</span> HRegion region = new HRegion(regionFs, null, conf, htd, null);<a name="line.183"></a>
<span class="sourceLineNo">184</span> return new HStore(region, htd.getColumnFamily(Bytes.toBytes(familyName)), conf, false);<a name="line.184"></a>
<span class="sourceLineNo">185</span> }<a name="line.185"></a>
<span class="sourceLineNo">186</span> }<a name="line.186"></a>
<span class="sourceLineNo">187</span><a name="line.187"></a>
<span class="sourceLineNo">188</span> private static boolean isRegionDir(final FileSystem fs, final Path path) throws IOException {<a name="line.188"></a>
<span class="sourceLineNo">189</span> Path regionInfo = new Path(path, HRegionFileSystem.REGION_INFO_FILE);<a name="line.189"></a>
<span class="sourceLineNo">190</span> return fs.exists(regionInfo);<a name="line.190"></a>
<span class="sourceLineNo">191</span> }<a name="line.191"></a>
<span class="sourceLineNo">192</span><a name="line.192"></a>
<span class="sourceLineNo">193</span> private static boolean isTableDir(final FileSystem fs, final Path path) throws IOException {<a name="line.193"></a>
<span class="sourceLineNo">194</span> return FSTableDescriptors.getTableInfoPath(fs, path) != null;<a name="line.194"></a>
<span class="sourceLineNo">195</span> }<a name="line.195"></a>
<span class="sourceLineNo">196</span><a name="line.196"></a>
<span class="sourceLineNo">197</span> private static boolean isFamilyDir(final FileSystem fs, final Path path) throws IOException {<a name="line.197"></a>
<span class="sourceLineNo">198</span> return isRegionDir(fs, path.getParent());<a name="line.198"></a>
<span class="sourceLineNo">199</span> }<a name="line.199"></a>
<span class="sourceLineNo">200</span><a name="line.200"></a>
<span class="sourceLineNo">201</span> private static class CompactionMapper<a name="line.201"></a>
<span class="sourceLineNo">202</span> extends Mapper&lt;LongWritable, Text, NullWritable, NullWritable&gt; {<a name="line.202"></a>
<span class="sourceLineNo">203</span> private CompactionWorker compactor = null;<a name="line.203"></a>
<span class="sourceLineNo">204</span> private boolean compactOnce = false;<a name="line.204"></a>
<span class="sourceLineNo">205</span> private boolean major = false;<a name="line.205"></a>
<span class="sourceLineNo">206</span><a name="line.206"></a>
<span class="sourceLineNo">207</span> @Override<a name="line.207"></a>
<span class="sourceLineNo">208</span> public void setup(Context context) {<a name="line.208"></a>
<span class="sourceLineNo">209</span> Configuration conf = context.getConfiguration();<a name="line.209"></a>
<span class="sourceLineNo">210</span> compactOnce = conf.getBoolean(CONF_COMPACT_ONCE, false);<a name="line.210"></a>
<span class="sourceLineNo">211</span> major = conf.getBoolean(CONF_COMPACT_MAJOR, false);<a name="line.211"></a>
<span class="sourceLineNo">212</span><a name="line.212"></a>
<span class="sourceLineNo">213</span> try {<a name="line.213"></a>
<span class="sourceLineNo">214</span> FileSystem fs = CommonFSUtils.getRootDirFileSystem(conf);<a name="line.214"></a>
<span class="sourceLineNo">215</span> this.compactor = new CompactionWorker(fs, conf);<a name="line.215"></a>
<span class="sourceLineNo">216</span> } catch (IOException e) {<a name="line.216"></a>
<span class="sourceLineNo">217</span> throw new RuntimeException("Could not get the input FileSystem", e);<a name="line.217"></a>
<span class="sourceLineNo">218</span> }<a name="line.218"></a>
<span class="sourceLineNo">219</span> }<a name="line.219"></a>
<span class="sourceLineNo">220</span><a name="line.220"></a>
<span class="sourceLineNo">221</span> @Override<a name="line.221"></a>
<span class="sourceLineNo">222</span> public void map(LongWritable key, Text value, Context context)<a name="line.222"></a>
<span class="sourceLineNo">223</span> throws InterruptedException, IOException {<a name="line.223"></a>
<span class="sourceLineNo">224</span> Path path = new Path(value.toString());<a name="line.224"></a>
<span class="sourceLineNo">225</span> this.compactor.compact(path, compactOnce, major);<a name="line.225"></a>
<span class="sourceLineNo">226</span> }<a name="line.226"></a>
<span class="sourceLineNo">227</span> }<a name="line.227"></a>
<span class="sourceLineNo">228</span><a name="line.228"></a>
<span class="sourceLineNo">229</span> /**<a name="line.229"></a>
<span class="sourceLineNo">230</span> * Input format that uses store files block location as input split locality.<a name="line.230"></a>
<span class="sourceLineNo">231</span> */<a name="line.231"></a>
<span class="sourceLineNo">232</span> private static class CompactionInputFormat extends TextInputFormat {<a name="line.232"></a>
<span class="sourceLineNo">233</span> @Override<a name="line.233"></a>
<span class="sourceLineNo">234</span> protected boolean isSplitable(JobContext context, Path file) {<a name="line.234"></a>
<span class="sourceLineNo">235</span> return true;<a name="line.235"></a>
<span class="sourceLineNo">236</span> }<a name="line.236"></a>
<span class="sourceLineNo">237</span><a name="line.237"></a>
<span class="sourceLineNo">238</span> /**<a name="line.238"></a>
<span class="sourceLineNo">239</span> * Returns a split for each store files directory using the block location<a name="line.239"></a>
<span class="sourceLineNo">240</span> * of each file as locality reference.<a name="line.240"></a>
<span class="sourceLineNo">241</span> */<a name="line.241"></a>
<span class="sourceLineNo">242</span> @Override<a name="line.242"></a>
<span class="sourceLineNo">243</span> public List&lt;InputSplit&gt; getSplits(JobContext job) throws IOException {<a name="line.243"></a>
<span class="sourceLineNo">244</span> List&lt;InputSplit&gt; splits = new ArrayList&lt;&gt;();<a name="line.244"></a>
<span class="sourceLineNo">245</span> List&lt;FileStatus&gt; files = listStatus(job);<a name="line.245"></a>
<span class="sourceLineNo">246</span><a name="line.246"></a>
<span class="sourceLineNo">247</span> Text key = new Text();<a name="line.247"></a>
<span class="sourceLineNo">248</span> for (FileStatus file: files) {<a name="line.248"></a>
<span class="sourceLineNo">249</span> Path path = file.getPath();<a name="line.249"></a>
<span class="sourceLineNo">250</span> FileSystem fs = path.getFileSystem(job.getConfiguration());<a name="line.250"></a>
<span class="sourceLineNo">251</span> LineReader reader = new LineReader(fs.open(path));<a name="line.251"></a>
<span class="sourceLineNo">252</span> long pos = 0;<a name="line.252"></a>
<span class="sourceLineNo">253</span> int n;<a name="line.253"></a>
<span class="sourceLineNo">254</span> try {<a name="line.254"></a>
<span class="sourceLineNo">255</span> while ((n = reader.readLine(key)) &gt; 0) {<a name="line.255"></a>
<span class="sourceLineNo">256</span> String[] hosts = getStoreDirHosts(fs, path);<a name="line.256"></a>
<span class="sourceLineNo">257</span> splits.add(new FileSplit(path, pos, n, hosts));<a name="line.257"></a>
<span class="sourceLineNo">258</span> pos += n;<a name="line.258"></a>
<span class="sourceLineNo">259</span> }<a name="line.259"></a>
<span class="sourceLineNo">260</span> } finally {<a name="line.260"></a>
<span class="sourceLineNo">261</span> reader.close();<a name="line.261"></a>
<span class="sourceLineNo">262</span> }<a name="line.262"></a>
<span class="sourceLineNo">263</span> }<a name="line.263"></a>
<span class="sourceLineNo">264</span><a name="line.264"></a>
<span class="sourceLineNo">265</span> return splits;<a name="line.265"></a>
<span class="sourceLineNo">266</span> }<a name="line.266"></a>
<span class="sourceLineNo">267</span><a name="line.267"></a>
<span class="sourceLineNo">268</span> /**<a name="line.268"></a>
<span class="sourceLineNo">269</span> * return the top hosts of the store files, used by the Split<a name="line.269"></a>
<span class="sourceLineNo">270</span> */<a name="line.270"></a>
<span class="sourceLineNo">271</span> private static String[] getStoreDirHosts(final FileSystem fs, final Path path)<a name="line.271"></a>
<span class="sourceLineNo">272</span> throws IOException {<a name="line.272"></a>
<span class="sourceLineNo">273</span> FileStatus[] files = CommonFSUtils.listStatus(fs, path);<a name="line.273"></a>
<span class="sourceLineNo">274</span> if (files == null) {<a name="line.274"></a>
<span class="sourceLineNo">275</span> return new String[] {};<a name="line.275"></a>
<span class="sourceLineNo">276</span> }<a name="line.276"></a>
<span class="sourceLineNo">277</span><a name="line.277"></a>
<span class="sourceLineNo">278</span> HDFSBlocksDistribution hdfsBlocksDistribution = new HDFSBlocksDistribution();<a name="line.278"></a>
<span class="sourceLineNo">279</span> for (FileStatus hfileStatus: files) {<a name="line.279"></a>
<span class="sourceLineNo">280</span> HDFSBlocksDistribution storeFileBlocksDistribution =<a name="line.280"></a>
<span class="sourceLineNo">281</span> FSUtils.computeHDFSBlocksDistribution(fs, hfileStatus, 0, hfileStatus.getLen());<a name="line.281"></a>
<span class="sourceLineNo">282</span> hdfsBlocksDistribution.add(storeFileBlocksDistribution);<a name="line.282"></a>
<span class="sourceLineNo">283</span> }<a name="line.283"></a>
<span class="sourceLineNo">284</span><a name="line.284"></a>
<span class="sourceLineNo">285</span> List&lt;String&gt; hosts = hdfsBlocksDistribution.getTopHosts();<a name="line.285"></a>
<span class="sourceLineNo">286</span> return hosts.toArray(new String[hosts.size()]);<a name="line.286"></a>
<span class="sourceLineNo">287</span> }<a name="line.287"></a>
<span class="sourceLineNo">288</span><a name="line.288"></a>
<span class="sourceLineNo">289</span> /**<a name="line.289"></a>
<span class="sourceLineNo">290</span> * Create the input file for the given directories to compact.<a name="line.290"></a>
<span class="sourceLineNo">291</span> * The file is a TextFile with each line corrisponding to a<a name="line.291"></a>
<span class="sourceLineNo">292</span> * store files directory to compact.<a name="line.292"></a>
<span class="sourceLineNo">293</span> */<a name="line.293"></a>
<span class="sourceLineNo">294</span> public static List&lt;Path&gt; createInputFile(final FileSystem fs, final FileSystem stagingFs,<a name="line.294"></a>
<span class="sourceLineNo">295</span> final Path path, final Set&lt;Path&gt; toCompactDirs) throws IOException {<a name="line.295"></a>
<span class="sourceLineNo">296</span> // Extract the list of store dirs<a name="line.296"></a>
<span class="sourceLineNo">297</span> List&lt;Path&gt; storeDirs = new LinkedList&lt;&gt;();<a name="line.297"></a>
<span class="sourceLineNo">298</span> for (Path compactDir: toCompactDirs) {<a name="line.298"></a>
<span class="sourceLineNo">299</span> if (isFamilyDir(fs, compactDir)) {<a name="line.299"></a>
<span class="sourceLineNo">300</span> storeDirs.add(compactDir);<a name="line.300"></a>
<span class="sourceLineNo">301</span> } else if (isRegionDir(fs, compactDir)) {<a name="line.301"></a>
<span class="sourceLineNo">302</span> storeDirs.addAll(FSUtils.getFamilyDirs(fs, compactDir));<a name="line.302"></a>
<span class="sourceLineNo">303</span> } else if (isTableDir(fs, compactDir)) {<a name="line.303"></a>
<span class="sourceLineNo">304</span> // Lookup regions<a name="line.304"></a>
<span class="sourceLineNo">305</span> for (Path regionDir: FSUtils.getRegionDirs(fs, compactDir)) {<a name="line.305"></a>
<span class="sourceLineNo">306</span> storeDirs.addAll(FSUtils.getFamilyDirs(fs, regionDir));<a name="line.306"></a>
<span class="sourceLineNo">307</span> }<a name="line.307"></a>
<span class="sourceLineNo">308</span> } else {<a name="line.308"></a>
<span class="sourceLineNo">309</span> throw new IOException(<a name="line.309"></a>
<span class="sourceLineNo">310</span> "Specified path is not a table, region or family directory. path=" + compactDir);<a name="line.310"></a>
<span class="sourceLineNo">311</span> }<a name="line.311"></a>
<span class="sourceLineNo">312</span> }<a name="line.312"></a>
<span class="sourceLineNo">313</span><a name="line.313"></a>
<span class="sourceLineNo">314</span> // Write Input File<a name="line.314"></a>
<span class="sourceLineNo">315</span> FSDataOutputStream stream = stagingFs.create(path);<a name="line.315"></a>
<span class="sourceLineNo">316</span> LOG.info("Create input file=" + path + " with " + storeDirs.size() + " dirs to compact.");<a name="line.316"></a>
<span class="sourceLineNo">317</span> try {<a name="line.317"></a>
<span class="sourceLineNo">318</span> final byte[] newLine = Bytes.toBytes("\n");<a name="line.318"></a>
<span class="sourceLineNo">319</span> for (Path storeDir: storeDirs) {<a name="line.319"></a>
<span class="sourceLineNo">320</span> stream.write(Bytes.toBytes(storeDir.toString()));<a name="line.320"></a>
<span class="sourceLineNo">321</span> stream.write(newLine);<a name="line.321"></a>
<span class="sourceLineNo">322</span> }<a name="line.322"></a>
<span class="sourceLineNo">323</span> } finally {<a name="line.323"></a>
<span class="sourceLineNo">324</span> stream.close();<a name="line.324"></a>
<span class="sourceLineNo">325</span> }<a name="line.325"></a>
<span class="sourceLineNo">326</span> return storeDirs;<a name="line.326"></a>
<span class="sourceLineNo">327</span> }<a name="line.327"></a>
<span class="sourceLineNo">328</span> }<a name="line.328"></a>
<span class="sourceLineNo">329</span><a name="line.329"></a>
<span class="sourceLineNo">330</span> /**<a name="line.330"></a>
<span class="sourceLineNo">331</span> * Execute compaction, using a Map-Reduce job.<a name="line.331"></a>
<span class="sourceLineNo">332</span> */<a name="line.332"></a>
<span class="sourceLineNo">333</span> private int doMapReduce(final FileSystem fs, final Set&lt;Path&gt; toCompactDirs,<a name="line.333"></a>
<span class="sourceLineNo">334</span> final boolean compactOnce, final boolean major) throws Exception {<a name="line.334"></a>
<span class="sourceLineNo">335</span> Configuration conf = getConf();<a name="line.335"></a>
<span class="sourceLineNo">336</span> conf.setBoolean(CONF_COMPACT_ONCE, compactOnce);<a name="line.336"></a>
<span class="sourceLineNo">337</span> conf.setBoolean(CONF_COMPACT_MAJOR, major);<a name="line.337"></a>
<span class="sourceLineNo">338</span><a name="line.338"></a>
<span class="sourceLineNo">339</span> Job job = new Job(conf);<a name="line.339"></a>
<span class="sourceLineNo">340</span> job.setJobName("CompactionTool");<a name="line.340"></a>
<span class="sourceLineNo">341</span> job.setJarByClass(CompactionTool.class);<a name="line.341"></a>
<span class="sourceLineNo">342</span> job.setMapperClass(CompactionMapper.class);<a name="line.342"></a>
<span class="sourceLineNo">343</span> job.setInputFormatClass(CompactionInputFormat.class);<a name="line.343"></a>
<span class="sourceLineNo">344</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.344"></a>
<span class="sourceLineNo">345</span> job.setMapSpeculativeExecution(false);<a name="line.345"></a>
<span class="sourceLineNo">346</span> job.setNumReduceTasks(0);<a name="line.346"></a>
<span class="sourceLineNo">347</span><a name="line.347"></a>
<span class="sourceLineNo">348</span> // add dependencies (including HBase ones)<a name="line.348"></a>
<span class="sourceLineNo">349</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.349"></a>
<span class="sourceLineNo">350</span><a name="line.350"></a>
<span class="sourceLineNo">351</span> Path stagingDir = JobUtil.getQualifiedStagingDir(conf);<a name="line.351"></a>
<span class="sourceLineNo">352</span> FileSystem stagingFs = stagingDir.getFileSystem(conf);<a name="line.352"></a>
<span class="sourceLineNo">353</span> try {<a name="line.353"></a>
<span class="sourceLineNo">354</span> // Create input file with the store dirs<a name="line.354"></a>
<span class="sourceLineNo">355</span> Path inputPath = new Path(stagingDir, "compact-"+ EnvironmentEdgeManager.currentTime());<a name="line.355"></a>
<span class="sourceLineNo">356</span> List&lt;Path&gt; storeDirs = CompactionInputFormat.createInputFile(fs, stagingFs,<a name="line.356"></a>
<span class="sourceLineNo">357</span> inputPath, toCompactDirs);<a name="line.357"></a>
<span class="sourceLineNo">358</span> CompactionInputFormat.addInputPath(job, inputPath);<a name="line.358"></a>
<span class="sourceLineNo">359</span><a name="line.359"></a>
<span class="sourceLineNo">360</span> // Initialize credential for secure cluster<a name="line.360"></a>
<span class="sourceLineNo">361</span> TableMapReduceUtil.initCredentials(job);<a name="line.361"></a>
<span class="sourceLineNo">362</span> // Despite the method name this will get delegation token for the filesystem<a name="line.362"></a>
<span class="sourceLineNo">363</span> TokenCache.obtainTokensForNamenodes(job.getCredentials(),<a name="line.363"></a>
<span class="sourceLineNo">364</span> storeDirs.toArray(new Path[0]), conf);<a name="line.364"></a>
<span class="sourceLineNo">365</span><a name="line.365"></a>
<span class="sourceLineNo">366</span> // Start the MR Job and wait<a name="line.366"></a>
<span class="sourceLineNo">367</span> return job.waitForCompletion(true) ? 0 : 1;<a name="line.367"></a>
<span class="sourceLineNo">368</span> } finally {<a name="line.368"></a>
<span class="sourceLineNo">369</span> fs.delete(stagingDir, true);<a name="line.369"></a>
<span class="sourceLineNo">370</span> }<a name="line.370"></a>
<span class="sourceLineNo">371</span> }<a name="line.371"></a>
<span class="sourceLineNo">372</span><a name="line.372"></a>
<span class="sourceLineNo">373</span> /**<a name="line.373"></a>
<span class="sourceLineNo">374</span> * Execute compaction, from this client, one path at the time.<a name="line.374"></a>
<span class="sourceLineNo">375</span> */<a name="line.375"></a>
<span class="sourceLineNo">376</span> private int doClient(final FileSystem fs, final Set&lt;Path&gt; toCompactDirs,<a name="line.376"></a>
<span class="sourceLineNo">377</span> final boolean compactOnce, final boolean major) throws IOException {<a name="line.377"></a>
<span class="sourceLineNo">378</span> CompactionWorker worker = new CompactionWorker(fs, getConf());<a name="line.378"></a>
<span class="sourceLineNo">379</span> for (Path path: toCompactDirs) {<a name="line.379"></a>
<span class="sourceLineNo">380</span> worker.compact(path, compactOnce, major);<a name="line.380"></a>
<span class="sourceLineNo">381</span> }<a name="line.381"></a>
<span class="sourceLineNo">382</span> return 0;<a name="line.382"></a>
<span class="sourceLineNo">383</span> }<a name="line.383"></a>
<span class="sourceLineNo">384</span><a name="line.384"></a>
<span class="sourceLineNo">385</span> @Override<a name="line.385"></a>
<span class="sourceLineNo">386</span> public int run(String[] args) throws Exception {<a name="line.386"></a>
<span class="sourceLineNo">387</span> Set&lt;Path&gt; toCompactDirs = new HashSet&lt;&gt;();<a name="line.387"></a>
<span class="sourceLineNo">388</span> boolean compactOnce = false;<a name="line.388"></a>
<span class="sourceLineNo">389</span> boolean major = false;<a name="line.389"></a>
<span class="sourceLineNo">390</span> boolean mapred = false;<a name="line.390"></a>
<span class="sourceLineNo">391</span><a name="line.391"></a>
<span class="sourceLineNo">392</span> Configuration conf = getConf();<a name="line.392"></a>
<span class="sourceLineNo">393</span> FileSystem fs = CommonFSUtils.getRootDirFileSystem(conf);<a name="line.393"></a>
<span class="sourceLineNo">394</span><a name="line.394"></a>
<span class="sourceLineNo">395</span> try {<a name="line.395"></a>
<span class="sourceLineNo">396</span> for (int i = 0; i &lt; args.length; ++i) {<a name="line.396"></a>
<span class="sourceLineNo">397</span> String opt = args[i];<a name="line.397"></a>
<span class="sourceLineNo">398</span> if (opt.equals("-compactOnce")) {<a name="line.398"></a>
<span class="sourceLineNo">399</span> compactOnce = true;<a name="line.399"></a>
<span class="sourceLineNo">400</span> } else if (opt.equals("-major")) {<a name="line.400"></a>
<span class="sourceLineNo">401</span> major = true;<a name="line.401"></a>
<span class="sourceLineNo">402</span> } else if (opt.equals("-mapred")) {<a name="line.402"></a>
<span class="sourceLineNo">403</span> mapred = true;<a name="line.403"></a>
<span class="sourceLineNo">404</span> } else if (!opt.startsWith("-")) {<a name="line.404"></a>
<span class="sourceLineNo">405</span> Path path = new Path(opt);<a name="line.405"></a>
<span class="sourceLineNo">406</span> FileStatus status = fs.getFileStatus(path);<a name="line.406"></a>
<span class="sourceLineNo">407</span> if (!status.isDirectory()) {<a name="line.407"></a>
<span class="sourceLineNo">408</span> printUsage("Specified path is not a directory. path=" + path);<a name="line.408"></a>
<span class="sourceLineNo">409</span> return 1;<a name="line.409"></a>
<span class="sourceLineNo">410</span> }<a name="line.410"></a>
<span class="sourceLineNo">411</span> toCompactDirs.add(path);<a name="line.411"></a>
<span class="sourceLineNo">412</span> } else {<a name="line.412"></a>
<span class="sourceLineNo">413</span> printUsage();<a name="line.413"></a>
<span class="sourceLineNo">414</span> }<a name="line.414"></a>
<span class="sourceLineNo">415</span> }<a name="line.415"></a>
<span class="sourceLineNo">416</span> } catch (Exception e) {<a name="line.416"></a>
<span class="sourceLineNo">417</span> printUsage(e.getMessage());<a name="line.417"></a>
<span class="sourceLineNo">418</span> return 1;<a name="line.418"></a>
<span class="sourceLineNo">419</span> }<a name="line.419"></a>
<span class="sourceLineNo">420</span><a name="line.420"></a>
<span class="sourceLineNo">421</span> if (toCompactDirs.isEmpty()) {<a name="line.421"></a>
<span class="sourceLineNo">422</span> printUsage("No directories to compact specified.");<a name="line.422"></a>
<span class="sourceLineNo">423</span> return 1;<a name="line.423"></a>
<span class="sourceLineNo">424</span> }<a name="line.424"></a>
<span class="sourceLineNo">425</span><a name="line.425"></a>
<span class="sourceLineNo">426</span> // Execute compaction!<a name="line.426"></a>
<span class="sourceLineNo">427</span> if (mapred) {<a name="line.427"></a>
<span class="sourceLineNo">428</span> return doMapReduce(fs, toCompactDirs, compactOnce, major);<a name="line.428"></a>
<span class="sourceLineNo">429</span> } else {<a name="line.429"></a>
<span class="sourceLineNo">430</span> return doClient(fs, toCompactDirs, compactOnce, major);<a name="line.430"></a>
<span class="sourceLineNo">431</span> }<a name="line.431"></a>
<span class="sourceLineNo">432</span> }<a name="line.432"></a>
<span class="sourceLineNo">433</span><a name="line.433"></a>
<span class="sourceLineNo">434</span> private void printUsage() {<a name="line.434"></a>
<span class="sourceLineNo">435</span> printUsage(null);<a name="line.435"></a>
<span class="sourceLineNo">436</span> }<a name="line.436"></a>
<span class="sourceLineNo">437</span><a name="line.437"></a>
<span class="sourceLineNo">438</span> private void printUsage(final String message) {<a name="line.438"></a>
<span class="sourceLineNo">439</span> if (message != null &amp;&amp; message.length() &gt; 0) {<a name="line.439"></a>
<span class="sourceLineNo">440</span> System.err.println(message);<a name="line.440"></a>
<span class="sourceLineNo">441</span> }<a name="line.441"></a>
<span class="sourceLineNo">442</span> System.err.println("Usage: java " + this.getClass().getName() + " \\");<a name="line.442"></a>
<span class="sourceLineNo">443</span> System.err.println(" [-compactOnce] [-major] [-mapred] [-D&lt;property=value&gt;]* files...");<a name="line.443"></a>
<span class="sourceLineNo">444</span> System.err.println();<a name="line.444"></a>
<span class="sourceLineNo">445</span> System.err.println("Options:");<a name="line.445"></a>
<span class="sourceLineNo">446</span> System.err.println(" mapred Use MapReduce to run compaction.");<a name="line.446"></a>
<span class="sourceLineNo">447</span> System.err.println(" compactOnce Execute just one compaction step. (default: while needed)");<a name="line.447"></a>
<span class="sourceLineNo">448</span> System.err.println(" major Trigger major compaction.");<a name="line.448"></a>
<span class="sourceLineNo">449</span> System.err.println();<a name="line.449"></a>
<span class="sourceLineNo">450</span> System.err.println("Note: -D properties will be applied to the conf used. ");<a name="line.450"></a>
<span class="sourceLineNo">451</span> System.err.println("For example: ");<a name="line.451"></a>
<span class="sourceLineNo">452</span> System.err.println(" To stop delete of compacted file, pass -D"+CONF_DELETE_COMPACTED+"=false");<a name="line.452"></a>
<span class="sourceLineNo">453</span> System.err.println();<a name="line.453"></a>
<span class="sourceLineNo">454</span> System.err.println("Examples:");<a name="line.454"></a>
<span class="sourceLineNo">455</span> System.err.println(" To compact the full 'TestTable' using MapReduce:");<a name="line.455"></a>
<span class="sourceLineNo">456</span> System.err.println(" $ hbase " + this.getClass().getName() +<a name="line.456"></a>
<span class="sourceLineNo">457</span> " -mapred hdfs://hbase/data/default/TestTable");<a name="line.457"></a>
<span class="sourceLineNo">458</span> System.err.println();<a name="line.458"></a>
<span class="sourceLineNo">459</span> System.err.println(" To compact column family 'x' of the table 'TestTable' region 'abc':");<a name="line.459"></a>
<span class="sourceLineNo">460</span> System.err.println(" $ hbase " + this.getClass().getName() +<a name="line.460"></a>
<span class="sourceLineNo">461</span> " hdfs://hbase/data/default/TestTable/abc/x");<a name="line.461"></a>
<span class="sourceLineNo">462</span> }<a name="line.462"></a>
<span class="sourceLineNo">463</span><a name="line.463"></a>
<span class="sourceLineNo">464</span> public static void main(String[] args) throws Exception {<a name="line.464"></a>
<span class="sourceLineNo">465</span> System.exit(ToolRunner.run(HBaseConfiguration.create(), new CompactionTool(), args));<a name="line.465"></a>
<span class="sourceLineNo">466</span> }<a name="line.466"></a>
<span class="sourceLineNo">467</span>}<a name="line.467"></a>
</pre>
</div>
</body>
</html>