| <!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.mapred, class: TableMapReduceUtil"> |
| <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.mapred;</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 java.io.IOException;</span> |
| <span class="source-line-no">021</span><span id="line-21">import java.util.Collection;</span> |
| <span class="source-line-no">022</span><span id="line-22">import java.util.Map;</span> |
| <span class="source-line-no">023</span><span id="line-23">import org.apache.hadoop.conf.Configuration;</span> |
| <span class="source-line-no">024</span><span id="line-24">import org.apache.hadoop.fs.Path;</span> |
| <span class="source-line-no">025</span><span id="line-25">import org.apache.hadoop.hbase.HBaseConfiguration;</span> |
| <span class="source-line-no">026</span><span id="line-26">import org.apache.hadoop.hbase.TableName;</span> |
| <span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.hbase.client.Connection;</span> |
| <span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.hbase.client.ConnectionFactory;</span> |
| <span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.hbase.client.Put;</span> |
| <span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.hbase.client.RegionLocator;</span> |
| <span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.hbase.client.Scan;</span> |
| <span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span> |
| <span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.mapreduce.MutationSerialization;</span> |
| <span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.mapreduce.ResultSerialization;</span> |
| <span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.security.User;</span> |
| <span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.security.UserProvider;</span> |
| <span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.security.token.TokenUtil;</span> |
| <span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.util.RegionSplitter;</span> |
| <span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.mapred.FileInputFormat;</span> |
| <span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.mapred.InputFormat;</span> |
| <span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.mapred.JobConf;</span> |
| <span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.mapred.OutputFormat;</span> |
| <span class="source-line-no">043</span><span id="line-43">import org.apache.hadoop.mapred.TextInputFormat;</span> |
| <span class="source-line-no">044</span><span id="line-44">import org.apache.hadoop.mapred.TextOutputFormat;</span> |
| <span class="source-line-no">045</span><span id="line-45">import org.apache.yetus.audience.InterfaceAudience;</span> |
| <span class="source-line-no">046</span><span id="line-46">import org.slf4j.Logger;</span> |
| <span class="source-line-no">047</span><span id="line-47">import org.slf4j.LoggerFactory;</span> |
| <span class="source-line-no">048</span><span id="line-48"></span> |
| <span class="source-line-no">049</span><span id="line-49">/**</span> |
| <span class="source-line-no">050</span><span id="line-50"> * Utility for {@link TableMap} and {@link TableReduce}</span> |
| <span class="source-line-no">051</span><span id="line-51"> */</span> |
| <span class="source-line-no">052</span><span id="line-52">@InterfaceAudience.Public</span> |
| <span class="source-line-no">053</span><span id="line-53">@SuppressWarnings({ "rawtypes", "unchecked" })</span> |
| <span class="source-line-no">054</span><span id="line-54">public class TableMapReduceUtil {</span> |
| <span class="source-line-no">055</span><span id="line-55"> private static final Logger LOG = LoggerFactory.getLogger(TableMapReduceUtil.class);</span> |
| <span class="source-line-no">056</span><span id="line-56"></span> |
| <span class="source-line-no">057</span><span id="line-57"> /**</span> |
| <span class="source-line-no">058</span><span id="line-58"> * Use this before submitting a TableMap job. It will appropriately set up the JobConf.</span> |
| <span class="source-line-no">059</span><span id="line-59"> * @param table The table name to read from.</span> |
| <span class="source-line-no">060</span><span id="line-60"> * @param columns The columns to scan.</span> |
| <span class="source-line-no">061</span><span id="line-61"> * @param mapper The mapper class to use.</span> |
| <span class="source-line-no">062</span><span id="line-62"> * @param outputKeyClass The class of the output key.</span> |
| <span class="source-line-no">063</span><span id="line-63"> * @param outputValueClass The class of the output value.</span> |
| <span class="source-line-no">064</span><span id="line-64"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">065</span><span id="line-65"> */</span> |
| <span class="source-line-no">066</span><span id="line-66"> public static void initTableMapJob(String table, String columns, Class<? extends TableMap> mapper,</span> |
| <span class="source-line-no">067</span><span id="line-67"> Class<?> outputKeyClass, Class<?> outputValueClass, JobConf job) {</span> |
| <span class="source-line-no">068</span><span id="line-68"> initTableMapJob(table, columns, mapper, outputKeyClass, outputValueClass, job, true,</span> |
| <span class="source-line-no">069</span><span id="line-69"> TableInputFormat.class);</span> |
| <span class="source-line-no">070</span><span id="line-70"> }</span> |
| <span class="source-line-no">071</span><span id="line-71"></span> |
| <span class="source-line-no">072</span><span id="line-72"> public static void initTableMapJob(String table, String columns, Class<? extends TableMap> mapper,</span> |
| <span class="source-line-no">073</span><span id="line-73"> Class<?> outputKeyClass, Class<?> outputValueClass, JobConf job, boolean addDependencyJars) {</span> |
| <span class="source-line-no">074</span><span id="line-74"> initTableMapJob(table, columns, mapper, outputKeyClass, outputValueClass, job,</span> |
| <span class="source-line-no">075</span><span id="line-75"> addDependencyJars, TableInputFormat.class);</span> |
| <span class="source-line-no">076</span><span id="line-76"> }</span> |
| <span class="source-line-no">077</span><span id="line-77"></span> |
| <span class="source-line-no">078</span><span id="line-78"> /**</span> |
| <span class="source-line-no">079</span><span id="line-79"> * Use this before submitting a TableMap job. It will appropriately set up the JobConf.</span> |
| <span class="source-line-no">080</span><span id="line-80"> * @param table The table name to read from.</span> |
| <span class="source-line-no">081</span><span id="line-81"> * @param columns The columns to scan.</span> |
| <span class="source-line-no">082</span><span id="line-82"> * @param mapper The mapper class to use.</span> |
| <span class="source-line-no">083</span><span id="line-83"> * @param outputKeyClass The class of the output key.</span> |
| <span class="source-line-no">084</span><span id="line-84"> * @param outputValueClass The class of the output value.</span> |
| <span class="source-line-no">085</span><span id="line-85"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">086</span><span id="line-86"> * @param addDependencyJars upload HBase jars and jars for any of the configured job classes via</span> |
| <span class="source-line-no">087</span><span id="line-87"> * the distributed cache (tmpjars).</span> |
| <span class="source-line-no">088</span><span id="line-88"> */</span> |
| <span class="source-line-no">089</span><span id="line-89"> public static void initTableMapJob(String table, String columns, Class<? extends TableMap> mapper,</span> |
| <span class="source-line-no">090</span><span id="line-90"> Class<?> outputKeyClass, Class<?> outputValueClass, JobConf job, boolean addDependencyJars,</span> |
| <span class="source-line-no">091</span><span id="line-91"> Class<? extends InputFormat> inputFormat) {</span> |
| <span class="source-line-no">092</span><span id="line-92"></span> |
| <span class="source-line-no">093</span><span id="line-93"> job.setInputFormat(inputFormat);</span> |
| <span class="source-line-no">094</span><span id="line-94"> job.setMapOutputValueClass(outputValueClass);</span> |
| <span class="source-line-no">095</span><span id="line-95"> job.setMapOutputKeyClass(outputKeyClass);</span> |
| <span class="source-line-no">096</span><span id="line-96"> job.setMapperClass(mapper);</span> |
| <span class="source-line-no">097</span><span id="line-97"> job.setStrings("io.serializations", job.get("io.serializations"),</span> |
| <span class="source-line-no">098</span><span id="line-98"> MutationSerialization.class.getName(), ResultSerialization.class.getName());</span> |
| <span class="source-line-no">099</span><span id="line-99"> FileInputFormat.addInputPaths(job, table);</span> |
| <span class="source-line-no">100</span><span id="line-100"> job.set(TableInputFormat.COLUMN_LIST, columns);</span> |
| <span class="source-line-no">101</span><span id="line-101"> if (addDependencyJars) {</span> |
| <span class="source-line-no">102</span><span id="line-102"> try {</span> |
| <span class="source-line-no">103</span><span id="line-103"> addDependencyJars(job);</span> |
| <span class="source-line-no">104</span><span id="line-104"> } catch (IOException e) {</span> |
| <span class="source-line-no">105</span><span id="line-105"> LOG.error("IOException encountered while adding dependency jars", e);</span> |
| <span class="source-line-no">106</span><span id="line-106"> }</span> |
| <span class="source-line-no">107</span><span id="line-107"> }</span> |
| <span class="source-line-no">108</span><span id="line-108"> try {</span> |
| <span class="source-line-no">109</span><span id="line-109"> initCredentials(job);</span> |
| <span class="source-line-no">110</span><span id="line-110"> } catch (IOException ioe) {</span> |
| <span class="source-line-no">111</span><span id="line-111"> // just spit out the stack trace? really?</span> |
| <span class="source-line-no">112</span><span id="line-112"> LOG.error("IOException encountered while initializing credentials", ioe);</span> |
| <span class="source-line-no">113</span><span id="line-113"> }</span> |
| <span class="source-line-no">114</span><span id="line-114"> }</span> |
| <span class="source-line-no">115</span><span id="line-115"></span> |
| <span class="source-line-no">116</span><span id="line-116"> /**</span> |
| <span class="source-line-no">117</span><span id="line-117"> * Sets up the job for reading from one or more multiple table snapshots, with one or more scans</span> |
| <span class="source-line-no">118</span><span id="line-118"> * per snapshot. It bypasses hbase servers and read directly from snapshot files.</span> |
| <span class="source-line-no">119</span><span id="line-119"> * @param snapshotScans map of snapshot name to scans on that snapshot.</span> |
| <span class="source-line-no">120</span><span id="line-120"> * @param mapper The mapper class to use.</span> |
| <span class="source-line-no">121</span><span id="line-121"> * @param outputKeyClass The class of the output key.</span> |
| <span class="source-line-no">122</span><span id="line-122"> * @param outputValueClass The class of the output value.</span> |
| <span class="source-line-no">123</span><span id="line-123"> * @param job The current job to adjust. Make sure the passed job is carrying all</span> |
| <span class="source-line-no">124</span><span id="line-124"> * necessary HBase configuration.</span> |
| <span class="source-line-no">125</span><span id="line-125"> * @param addDependencyJars upload HBase jars and jars for any of the configured job classes via</span> |
| <span class="source-line-no">126</span><span id="line-126"> * the distributed cache (tmpjars).</span> |
| <span class="source-line-no">127</span><span id="line-127"> */</span> |
| <span class="source-line-no">128</span><span id="line-128"> public static void initMultiTableSnapshotMapperJob(Map<String, Collection<Scan>> snapshotScans,</span> |
| <span class="source-line-no">129</span><span id="line-129"> Class<? extends TableMap> mapper, Class<?> outputKeyClass, Class<?> outputValueClass,</span> |
| <span class="source-line-no">130</span><span id="line-130"> JobConf job, boolean addDependencyJars, Path tmpRestoreDir) throws IOException {</span> |
| <span class="source-line-no">131</span><span id="line-131"> MultiTableSnapshotInputFormat.setInput(job, snapshotScans, tmpRestoreDir);</span> |
| <span class="source-line-no">132</span><span id="line-132"></span> |
| <span class="source-line-no">133</span><span id="line-133"> job.setInputFormat(MultiTableSnapshotInputFormat.class);</span> |
| <span class="source-line-no">134</span><span id="line-134"> if (outputValueClass != null) {</span> |
| <span class="source-line-no">135</span><span id="line-135"> job.setMapOutputValueClass(outputValueClass);</span> |
| <span class="source-line-no">136</span><span id="line-136"> }</span> |
| <span class="source-line-no">137</span><span id="line-137"> if (outputKeyClass != null) {</span> |
| <span class="source-line-no">138</span><span id="line-138"> job.setMapOutputKeyClass(outputKeyClass);</span> |
| <span class="source-line-no">139</span><span id="line-139"> }</span> |
| <span class="source-line-no">140</span><span id="line-140"> job.setMapperClass(mapper);</span> |
| <span class="source-line-no">141</span><span id="line-141"> if (addDependencyJars) {</span> |
| <span class="source-line-no">142</span><span id="line-142"> addDependencyJars(job);</span> |
| <span class="source-line-no">143</span><span id="line-143"> }</span> |
| <span class="source-line-no">144</span><span id="line-144"></span> |
| <span class="source-line-no">145</span><span id="line-145"> org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.resetCacheConfig(job);</span> |
| <span class="source-line-no">146</span><span id="line-146"> }</span> |
| <span class="source-line-no">147</span><span id="line-147"></span> |
| <span class="source-line-no">148</span><span id="line-148"> /**</span> |
| <span class="source-line-no">149</span><span id="line-149"> * Sets up the job for reading from a table snapshot. It bypasses hbase servers and read directly</span> |
| <span class="source-line-no">150</span><span id="line-150"> * from snapshot files.</span> |
| <span class="source-line-no">151</span><span id="line-151"> * @param snapshotName The name of the snapshot (of a table) to read from.</span> |
| <span class="source-line-no">152</span><span id="line-152"> * @param columns The columns to scan.</span> |
| <span class="source-line-no">153</span><span id="line-153"> * @param mapper The mapper class to use.</span> |
| <span class="source-line-no">154</span><span id="line-154"> * @param outputKeyClass The class of the output key.</span> |
| <span class="source-line-no">155</span><span id="line-155"> * @param outputValueClass The class of the output value.</span> |
| <span class="source-line-no">156</span><span id="line-156"> * @param job The current job to adjust. Make sure the passed job is carrying all</span> |
| <span class="source-line-no">157</span><span id="line-157"> * necessary HBase configuration.</span> |
| <span class="source-line-no">158</span><span id="line-158"> * @param addDependencyJars upload HBase jars and jars for any of the configured job classes via</span> |
| <span class="source-line-no">159</span><span id="line-159"> * the distributed cache (tmpjars).</span> |
| <span class="source-line-no">160</span><span id="line-160"> * @param tmpRestoreDir a temporary directory to copy the snapshot files into. Current user</span> |
| <span class="source-line-no">161</span><span id="line-161"> * should have write permissions to this directory, and this should not</span> |
| <span class="source-line-no">162</span><span id="line-162"> * be a subdirectory of rootdir. After the job is finished, restore</span> |
| <span class="source-line-no">163</span><span id="line-163"> * directory can be deleted.</span> |
| <span class="source-line-no">164</span><span id="line-164"> * @throws IOException When setting up the details fails.</span> |
| <span class="source-line-no">165</span><span id="line-165"> * @see TableSnapshotInputFormat</span> |
| <span class="source-line-no">166</span><span id="line-166"> */</span> |
| <span class="source-line-no">167</span><span id="line-167"> public static void initTableSnapshotMapJob(String snapshotName, String columns,</span> |
| <span class="source-line-no">168</span><span id="line-168"> Class<? extends TableMap> mapper, Class<?> outputKeyClass, Class<?> outputValueClass,</span> |
| <span class="source-line-no">169</span><span id="line-169"> JobConf job, boolean addDependencyJars, Path tmpRestoreDir) throws IOException {</span> |
| <span class="source-line-no">170</span><span id="line-170"> TableSnapshotInputFormat.setInput(job, snapshotName, tmpRestoreDir);</span> |
| <span class="source-line-no">171</span><span id="line-171"> initTableMapJob(snapshotName, columns, mapper, outputKeyClass, outputValueClass, job,</span> |
| <span class="source-line-no">172</span><span id="line-172"> addDependencyJars, TableSnapshotInputFormat.class);</span> |
| <span class="source-line-no">173</span><span id="line-173"> org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.resetCacheConfig(job);</span> |
| <span class="source-line-no">174</span><span id="line-174"> }</span> |
| <span class="source-line-no">175</span><span id="line-175"></span> |
| <span class="source-line-no">176</span><span id="line-176"> /**</span> |
| <span class="source-line-no">177</span><span id="line-177"> * Sets up the job for reading from a table snapshot. It bypasses hbase servers and read directly</span> |
| <span class="source-line-no">178</span><span id="line-178"> * from snapshot files.</span> |
| <span class="source-line-no">179</span><span id="line-179"> * @param snapshotName The name of the snapshot (of a table) to read from.</span> |
| <span class="source-line-no">180</span><span id="line-180"> * @param columns The columns to scan.</span> |
| <span class="source-line-no">181</span><span id="line-181"> * @param mapper The mapper class to use.</span> |
| <span class="source-line-no">182</span><span id="line-182"> * @param outputKeyClass The class of the output key.</span> |
| <span class="source-line-no">183</span><span id="line-183"> * @param outputValueClass The class of the output value.</span> |
| <span class="source-line-no">184</span><span id="line-184"> * @param jobConf The current job to adjust. Make sure the passed job is carrying all</span> |
| <span class="source-line-no">185</span><span id="line-185"> * necessary HBase configuration.</span> |
| <span class="source-line-no">186</span><span id="line-186"> * @param addDependencyJars upload HBase jars and jars for any of the configured job classes via</span> |
| <span class="source-line-no">187</span><span id="line-187"> * the distributed cache (tmpjars).</span> |
| <span class="source-line-no">188</span><span id="line-188"> * @param tmpRestoreDir a temporary directory to copy the snapshot files into. Current user</span> |
| <span class="source-line-no">189</span><span id="line-189"> * should have write permissions to this directory, and this should not</span> |
| <span class="source-line-no">190</span><span id="line-190"> * be a subdirectory of rootdir. After the job is finished, restore</span> |
| <span class="source-line-no">191</span><span id="line-191"> * directory can be deleted.</span> |
| <span class="source-line-no">192</span><span id="line-192"> * @param splitAlgo algorithm to split</span> |
| <span class="source-line-no">193</span><span id="line-193"> * @param numSplitsPerRegion how many input splits to generate per one region</span> |
| <span class="source-line-no">194</span><span id="line-194"> * @throws IOException When setting up the details fails.</span> |
| <span class="source-line-no">195</span><span id="line-195"> * @see TableSnapshotInputFormat</span> |
| <span class="source-line-no">196</span><span id="line-196"> */</span> |
| <span class="source-line-no">197</span><span id="line-197"> public static void initTableSnapshotMapJob(String snapshotName, String columns,</span> |
| <span class="source-line-no">198</span><span id="line-198"> Class<? extends TableMap> mapper, Class<?> outputKeyClass, Class<?> outputValueClass,</span> |
| <span class="source-line-no">199</span><span id="line-199"> JobConf jobConf, boolean addDependencyJars, Path tmpRestoreDir,</span> |
| <span class="source-line-no">200</span><span id="line-200"> RegionSplitter.SplitAlgorithm splitAlgo, int numSplitsPerRegion) throws IOException {</span> |
| <span class="source-line-no">201</span><span id="line-201"> TableSnapshotInputFormat.setInput(jobConf, snapshotName, tmpRestoreDir, splitAlgo,</span> |
| <span class="source-line-no">202</span><span id="line-202"> numSplitsPerRegion);</span> |
| <span class="source-line-no">203</span><span id="line-203"> initTableMapJob(snapshotName, columns, mapper, outputKeyClass, outputValueClass, jobConf,</span> |
| <span class="source-line-no">204</span><span id="line-204"> addDependencyJars, TableSnapshotInputFormat.class);</span> |
| <span class="source-line-no">205</span><span id="line-205"> org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.resetCacheConfig(jobConf);</span> |
| <span class="source-line-no">206</span><span id="line-206"> }</span> |
| <span class="source-line-no">207</span><span id="line-207"></span> |
| <span class="source-line-no">208</span><span id="line-208"> /**</span> |
| <span class="source-line-no">209</span><span id="line-209"> * Use this before submitting a TableReduce job. It will appropriately set up the JobConf.</span> |
| <span class="source-line-no">210</span><span id="line-210"> * @param table The output table.</span> |
| <span class="source-line-no">211</span><span id="line-211"> * @param reducer The reducer class to use.</span> |
| <span class="source-line-no">212</span><span id="line-212"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">213</span><span id="line-213"> * @throws IOException When determining the region count fails.</span> |
| <span class="source-line-no">214</span><span id="line-214"> */</span> |
| <span class="source-line-no">215</span><span id="line-215"> public static void initTableReduceJob(String table, Class<? extends TableReduce> reducer,</span> |
| <span class="source-line-no">216</span><span id="line-216"> JobConf job) throws IOException {</span> |
| <span class="source-line-no">217</span><span id="line-217"> initTableReduceJob(table, reducer, job, null);</span> |
| <span class="source-line-no">218</span><span id="line-218"> }</span> |
| <span class="source-line-no">219</span><span id="line-219"></span> |
| <span class="source-line-no">220</span><span id="line-220"> /**</span> |
| <span class="source-line-no">221</span><span id="line-221"> * Use this before submitting a TableReduce job. It will appropriately set up the JobConf.</span> |
| <span class="source-line-no">222</span><span id="line-222"> * @param table The output table.</span> |
| <span class="source-line-no">223</span><span id="line-223"> * @param reducer The reducer class to use.</span> |
| <span class="source-line-no">224</span><span id="line-224"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">225</span><span id="line-225"> * @param partitioner Partitioner to use. Pass <code>null</code> to use default partitioner.</span> |
| <span class="source-line-no">226</span><span id="line-226"> * @throws IOException When determining the region count fails.</span> |
| <span class="source-line-no">227</span><span id="line-227"> */</span> |
| <span class="source-line-no">228</span><span id="line-228"> public static void initTableReduceJob(String table, Class<? extends TableReduce> reducer,</span> |
| <span class="source-line-no">229</span><span id="line-229"> JobConf job, Class partitioner) throws IOException {</span> |
| <span class="source-line-no">230</span><span id="line-230"> initTableReduceJob(table, reducer, job, partitioner, true);</span> |
| <span class="source-line-no">231</span><span id="line-231"> }</span> |
| <span class="source-line-no">232</span><span id="line-232"></span> |
| <span class="source-line-no">233</span><span id="line-233"> /**</span> |
| <span class="source-line-no">234</span><span id="line-234"> * Use this before submitting a TableReduce job. It will appropriately set up the JobConf.</span> |
| <span class="source-line-no">235</span><span id="line-235"> * @param table The output table.</span> |
| <span class="source-line-no">236</span><span id="line-236"> * @param reducer The reducer class to use.</span> |
| <span class="source-line-no">237</span><span id="line-237"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">238</span><span id="line-238"> * @param partitioner Partitioner to use. Pass <code>null</code> to use default partitioner.</span> |
| <span class="source-line-no">239</span><span id="line-239"> * @param addDependencyJars upload HBase jars and jars for any of the configured job classes via</span> |
| <span class="source-line-no">240</span><span id="line-240"> * the distributed cache (tmpjars).</span> |
| <span class="source-line-no">241</span><span id="line-241"> * @throws IOException When determining the region count fails.</span> |
| <span class="source-line-no">242</span><span id="line-242"> */</span> |
| <span class="source-line-no">243</span><span id="line-243"> public static void initTableReduceJob(String table, Class<? extends TableReduce> reducer,</span> |
| <span class="source-line-no">244</span><span id="line-244"> JobConf job, Class partitioner, boolean addDependencyJars) throws IOException {</span> |
| <span class="source-line-no">245</span><span id="line-245"> job.setOutputFormat(TableOutputFormat.class);</span> |
| <span class="source-line-no">246</span><span id="line-246"> job.setReducerClass(reducer);</span> |
| <span class="source-line-no">247</span><span id="line-247"> job.set(TableOutputFormat.OUTPUT_TABLE, table);</span> |
| <span class="source-line-no">248</span><span id="line-248"> job.setOutputKeyClass(ImmutableBytesWritable.class);</span> |
| <span class="source-line-no">249</span><span id="line-249"> job.setOutputValueClass(Put.class);</span> |
| <span class="source-line-no">250</span><span id="line-250"> job.setStrings("io.serializations", job.get("io.serializations"),</span> |
| <span class="source-line-no">251</span><span id="line-251"> MutationSerialization.class.getName(), ResultSerialization.class.getName());</span> |
| <span class="source-line-no">252</span><span id="line-252"> if (partitioner == HRegionPartitioner.class) {</span> |
| <span class="source-line-no">253</span><span id="line-253"> job.setPartitionerClass(HRegionPartitioner.class);</span> |
| <span class="source-line-no">254</span><span id="line-254"> int regions = getRegionCount(HBaseConfiguration.create(job), TableName.valueOf(table));</span> |
| <span class="source-line-no">255</span><span id="line-255"> if (job.getNumReduceTasks() > regions) {</span> |
| <span class="source-line-no">256</span><span id="line-256"> job.setNumReduceTasks(regions);</span> |
| <span class="source-line-no">257</span><span id="line-257"> }</span> |
| <span class="source-line-no">258</span><span id="line-258"> } else if (partitioner != null) {</span> |
| <span class="source-line-no">259</span><span id="line-259"> job.setPartitionerClass(partitioner);</span> |
| <span class="source-line-no">260</span><span id="line-260"> }</span> |
| <span class="source-line-no">261</span><span id="line-261"> if (addDependencyJars) {</span> |
| <span class="source-line-no">262</span><span id="line-262"> addDependencyJars(job);</span> |
| <span class="source-line-no">263</span><span id="line-263"> }</span> |
| <span class="source-line-no">264</span><span id="line-264"> initCredentials(job);</span> |
| <span class="source-line-no">265</span><span id="line-265"> }</span> |
| <span class="source-line-no">266</span><span id="line-266"></span> |
| <span class="source-line-no">267</span><span id="line-267"> public static void initCredentials(JobConf job) throws IOException {</span> |
| <span class="source-line-no">268</span><span id="line-268"> UserProvider userProvider = UserProvider.instantiate(job);</span> |
| <span class="source-line-no">269</span><span id="line-269"> if (userProvider.isHadoopSecurityEnabled()) {</span> |
| <span class="source-line-no">270</span><span id="line-270"> // propagate delegation related props from launcher job to MR job</span> |
| <span class="source-line-no">271</span><span id="line-271"> if (System.getenv("HADOOP_TOKEN_FILE_LOCATION") != null) {</span> |
| <span class="source-line-no">272</span><span id="line-272"> job.set("mapreduce.job.credentials.binary", System.getenv("HADOOP_TOKEN_FILE_LOCATION"));</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"></span> |
| <span class="source-line-no">276</span><span id="line-276"> if (userProvider.isHBaseSecurityEnabled()) {</span> |
| <span class="source-line-no">277</span><span id="line-277"> Connection conn = ConnectionFactory.createConnection(job);</span> |
| <span class="source-line-no">278</span><span id="line-278"> try {</span> |
| <span class="source-line-no">279</span><span id="line-279"> // login the server principal (if using secure Hadoop)</span> |
| <span class="source-line-no">280</span><span id="line-280"> User user = userProvider.getCurrent();</span> |
| <span class="source-line-no">281</span><span id="line-281"> TokenUtil.addTokenForJob(conn, job, user);</span> |
| <span class="source-line-no">282</span><span id="line-282"> } catch (InterruptedException ie) {</span> |
| <span class="source-line-no">283</span><span id="line-283"> LOG.error("Interrupted obtaining user authentication token", ie);</span> |
| <span class="source-line-no">284</span><span id="line-284"> Thread.currentThread().interrupt();</span> |
| <span class="source-line-no">285</span><span id="line-285"> } finally {</span> |
| <span class="source-line-no">286</span><span id="line-286"> conn.close();</span> |
| <span class="source-line-no">287</span><span id="line-287"> }</span> |
| <span class="source-line-no">288</span><span id="line-288"> }</span> |
| <span class="source-line-no">289</span><span id="line-289"> }</span> |
| <span class="source-line-no">290</span><span id="line-290"></span> |
| <span class="source-line-no">291</span><span id="line-291"> /**</span> |
| <span class="source-line-no">292</span><span id="line-292"> * Ensures that the given number of reduce tasks for the given job configuration does not exceed</span> |
| <span class="source-line-no">293</span><span id="line-293"> * the number of regions for the given table.</span> |
| <span class="source-line-no">294</span><span id="line-294"> * @param table The table to get the region count for.</span> |
| <span class="source-line-no">295</span><span id="line-295"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">296</span><span id="line-296"> * @throws IOException When retrieving the table details fails.</span> |
| <span class="source-line-no">297</span><span id="line-297"> */</span> |
| <span class="source-line-no">298</span><span id="line-298"> // Used by tests.</span> |
| <span class="source-line-no">299</span><span id="line-299"> public static void limitNumReduceTasks(String table, JobConf job) throws IOException {</span> |
| <span class="source-line-no">300</span><span id="line-300"> int regions = getRegionCount(HBaseConfiguration.create(job), TableName.valueOf(table));</span> |
| <span class="source-line-no">301</span><span id="line-301"> if (job.getNumReduceTasks() > regions) {</span> |
| <span class="source-line-no">302</span><span id="line-302"> job.setNumReduceTasks(regions);</span> |
| <span class="source-line-no">303</span><span id="line-303"> }</span> |
| <span class="source-line-no">304</span><span id="line-304"> }</span> |
| <span class="source-line-no">305</span><span id="line-305"></span> |
| <span class="source-line-no">306</span><span id="line-306"> /**</span> |
| <span class="source-line-no">307</span><span id="line-307"> * Ensures that the given number of map tasks for the given job configuration does not exceed the</span> |
| <span class="source-line-no">308</span><span id="line-308"> * number of regions for the given table.</span> |
| <span class="source-line-no">309</span><span id="line-309"> * @param table The table to get the region count for.</span> |
| <span class="source-line-no">310</span><span id="line-310"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">311</span><span id="line-311"> * @throws IOException When retrieving the table details fails.</span> |
| <span class="source-line-no">312</span><span id="line-312"> */</span> |
| <span class="source-line-no">313</span><span id="line-313"> // Used by tests.</span> |
| <span class="source-line-no">314</span><span id="line-314"> public static void limitNumMapTasks(String table, JobConf job) throws IOException {</span> |
| <span class="source-line-no">315</span><span id="line-315"> int regions = getRegionCount(HBaseConfiguration.create(job), TableName.valueOf(table));</span> |
| <span class="source-line-no">316</span><span id="line-316"> if (job.getNumMapTasks() > regions) {</span> |
| <span class="source-line-no">317</span><span id="line-317"> job.setNumMapTasks(regions);</span> |
| <span class="source-line-no">318</span><span id="line-318"> }</span> |
| <span class="source-line-no">319</span><span id="line-319"> }</span> |
| <span class="source-line-no">320</span><span id="line-320"></span> |
| <span class="source-line-no">321</span><span id="line-321"> /**</span> |
| <span class="source-line-no">322</span><span id="line-322"> * Sets the number of reduce tasks for the given job configuration to the number of regions the</span> |
| <span class="source-line-no">323</span><span id="line-323"> * given table has.</span> |
| <span class="source-line-no">324</span><span id="line-324"> * @param table The table to get the region count for.</span> |
| <span class="source-line-no">325</span><span id="line-325"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">326</span><span id="line-326"> * @throws IOException When retrieving the table details fails.</span> |
| <span class="source-line-no">327</span><span id="line-327"> */</span> |
| <span class="source-line-no">328</span><span id="line-328"> public static void setNumReduceTasks(String table, JobConf job) throws IOException {</span> |
| <span class="source-line-no">329</span><span id="line-329"> job.setNumReduceTasks(getRegionCount(HBaseConfiguration.create(job), TableName.valueOf(table)));</span> |
| <span class="source-line-no">330</span><span id="line-330"> }</span> |
| <span class="source-line-no">331</span><span id="line-331"></span> |
| <span class="source-line-no">332</span><span id="line-332"> /**</span> |
| <span class="source-line-no">333</span><span id="line-333"> * Sets the number of map tasks for the given job configuration to the number of regions the given</span> |
| <span class="source-line-no">334</span><span id="line-334"> * table has.</span> |
| <span class="source-line-no">335</span><span id="line-335"> * @param table The table to get the region count for.</span> |
| <span class="source-line-no">336</span><span id="line-336"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">337</span><span id="line-337"> * @throws IOException When retrieving the table details fails.</span> |
| <span class="source-line-no">338</span><span id="line-338"> */</span> |
| <span class="source-line-no">339</span><span id="line-339"> public static void setNumMapTasks(String table, JobConf job) throws IOException {</span> |
| <span class="source-line-no">340</span><span id="line-340"> job.setNumMapTasks(getRegionCount(HBaseConfiguration.create(job), TableName.valueOf(table)));</span> |
| <span class="source-line-no">341</span><span id="line-341"> }</span> |
| <span class="source-line-no">342</span><span id="line-342"></span> |
| <span class="source-line-no">343</span><span id="line-343"> /**</span> |
| <span class="source-line-no">344</span><span id="line-344"> * Sets the number of rows to return and cache with each scanner iteration. Higher caching values</span> |
| <span class="source-line-no">345</span><span id="line-345"> * will enable faster mapreduce jobs at the expense of requiring more heap to contain the cached</span> |
| <span class="source-line-no">346</span><span id="line-346"> * rows.</span> |
| <span class="source-line-no">347</span><span id="line-347"> * @param job The current job configuration to adjust.</span> |
| <span class="source-line-no">348</span><span id="line-348"> * @param batchSize The number of rows to return in batch with each scanner iteration.</span> |
| <span class="source-line-no">349</span><span id="line-349"> */</span> |
| <span class="source-line-no">350</span><span id="line-350"> public static void setScannerCaching(JobConf job, int batchSize) {</span> |
| <span class="source-line-no">351</span><span id="line-351"> job.setInt("hbase.client.scanner.caching", batchSize);</span> |
| <span class="source-line-no">352</span><span id="line-352"> }</span> |
| <span class="source-line-no">353</span><span id="line-353"></span> |
| <span class="source-line-no">354</span><span id="line-354"> /**</span> |
| <span class="source-line-no">355</span><span id="line-355"> * @see org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil#addDependencyJars(org.apache.hadoop.mapreduce.Job)</span> |
| <span class="source-line-no">356</span><span id="line-356"> */</span> |
| <span class="source-line-no">357</span><span id="line-357"> public static void addDependencyJars(JobConf job) throws IOException {</span> |
| <span class="source-line-no">358</span><span id="line-358"> org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.addHBaseDependencyJars(job);</span> |
| <span class="source-line-no">359</span><span id="line-359"> org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil.addDependencyJarsForClasses(job,</span> |
| <span class="source-line-no">360</span><span id="line-360"> job.getMapOutputKeyClass(), job.getMapOutputValueClass(), job.getOutputKeyClass(),</span> |
| <span class="source-line-no">361</span><span id="line-361"> job.getOutputValueClass(), job.getPartitionerClass(),</span> |
| <span class="source-line-no">362</span><span id="line-362"> job.getClass("mapred.input.format.class", TextInputFormat.class, InputFormat.class),</span> |
| <span class="source-line-no">363</span><span id="line-363"> job.getClass("mapred.output.format.class", TextOutputFormat.class, OutputFormat.class),</span> |
| <span class="source-line-no">364</span><span id="line-364"> job.getCombinerClass());</span> |
| <span class="source-line-no">365</span><span id="line-365"> }</span> |
| <span class="source-line-no">366</span><span id="line-366"></span> |
| <span class="source-line-no">367</span><span id="line-367"> private static int getRegionCount(Configuration conf, TableName tableName) throws IOException {</span> |
| <span class="source-line-no">368</span><span id="line-368"> try (Connection conn = ConnectionFactory.createConnection(conf);</span> |
| <span class="source-line-no">369</span><span id="line-369"> RegionLocator locator = conn.getRegionLocator(tableName)) {</span> |
| <span class="source-line-no">370</span><span id="line-370"> return locator.getAllRegionLocations().size();</span> |
| <span class="source-line-no">371</span><span id="line-371"> }</span> |
| <span class="source-line-no">372</span><span id="line-372"> }</span> |
| <span class="source-line-no">373</span><span id="line-373">}</span> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </main> |
| </body> |
| </html> |
