blob: 26e1692af167db25153197be7b32b32d8c85a38f [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.mapreduce, class: TestTableMapReduceBase">
<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.mapreduce;</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.junit.Assert.assertTrue;</span>
<span class="source-line-no">021</span><span id="line-21">import static org.junit.Assert.fail;</span>
<span class="source-line-no">022</span><span id="line-22"></span>
<span class="source-line-no">023</span><span id="line-23">import java.io.IOException;</span>
<span class="source-line-no">024</span><span id="line-24">import java.util.Iterator;</span>
<span class="source-line-no">025</span><span id="line-25">import java.util.Map;</span>
<span class="source-line-no">026</span><span id="line-26">import java.util.NavigableMap;</span>
<span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.conf.Configuration;</span>
<span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.hbase.Cell;</span>
<span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.hbase.CellUtil;</span>
<span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.hbase.HBaseTestingUtil;</span>
<span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.hbase.HConstants;</span>
<span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.TableName;</span>
<span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.client.Put;</span>
<span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.client.Result;</span>
<span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.client.ResultScanner;</span>
<span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.client.Scan;</span>
<span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.client.Table;</span>
<span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.hbase.util.Bytes;</span>
<span class="source-line-no">040</span><span id="line-40">import org.junit.AfterClass;</span>
<span class="source-line-no">041</span><span id="line-41">import org.junit.BeforeClass;</span>
<span class="source-line-no">042</span><span id="line-42">import org.junit.Test;</span>
<span class="source-line-no">043</span><span id="line-43">import org.slf4j.Logger;</span>
<span class="source-line-no">044</span><span id="line-44"></span>
<span class="source-line-no">045</span><span id="line-45">/**</span>
<span class="source-line-no">046</span><span id="line-46"> * A base class for a test Map/Reduce job over HBase tables. The map/reduce process we're testing on</span>
<span class="source-line-no">047</span><span id="line-47"> * our tables is simple - take every row in the table, reverse the value of a particular cell, and</span>
<span class="source-line-no">048</span><span id="line-48"> * write it back to the table. Implements common components between mapred and mapreduce</span>
<span class="source-line-no">049</span><span id="line-49"> * implementations.</span>
<span class="source-line-no">050</span><span id="line-50"> */</span>
<span class="source-line-no">051</span><span id="line-51">public abstract class TestTableMapReduceBase {</span>
<span class="source-line-no">052</span><span id="line-52"> protected static final HBaseTestingUtil UTIL = new HBaseTestingUtil();</span>
<span class="source-line-no">053</span><span id="line-53"> protected static final TableName MULTI_REGION_TABLE_NAME = TableName.valueOf("mrtest");</span>
<span class="source-line-no">054</span><span id="line-54"> protected static final TableName TABLE_FOR_NEGATIVE_TESTS = TableName.valueOf("testfailuretable");</span>
<span class="source-line-no">055</span><span id="line-55"> protected static final byte[] INPUT_FAMILY = Bytes.toBytes("contents");</span>
<span class="source-line-no">056</span><span id="line-56"> protected static final byte[] OUTPUT_FAMILY = Bytes.toBytes("text");</span>
<span class="source-line-no">057</span><span id="line-57"></span>
<span class="source-line-no">058</span><span id="line-58"> protected static final byte[][] columns = new byte[][] { INPUT_FAMILY, OUTPUT_FAMILY };</span>
<span class="source-line-no">059</span><span id="line-59"></span>
<span class="source-line-no">060</span><span id="line-60"> /**</span>
<span class="source-line-no">061</span><span id="line-61"> * Retrieve my logger instance.</span>
<span class="source-line-no">062</span><span id="line-62"> */</span>
<span class="source-line-no">063</span><span id="line-63"> protected abstract Logger getLog();</span>
<span class="source-line-no">064</span><span id="line-64"></span>
<span class="source-line-no">065</span><span id="line-65"> /**</span>
<span class="source-line-no">066</span><span id="line-66"> * Handles API-specifics for setting up and executing the job.</span>
<span class="source-line-no">067</span><span id="line-67"> */</span>
<span class="source-line-no">068</span><span id="line-68"> protected abstract void runTestOnTable(Table table) throws IOException;</span>
<span class="source-line-no">069</span><span id="line-69"></span>
<span class="source-line-no">070</span><span id="line-70"> @BeforeClass</span>
<span class="source-line-no">071</span><span id="line-71"> public static void beforeClass() throws Exception {</span>
<span class="source-line-no">072</span><span id="line-72"> UTIL.startMiniCluster();</span>
<span class="source-line-no">073</span><span id="line-73"> Table table = UTIL.createMultiRegionTable(MULTI_REGION_TABLE_NAME,</span>
<span class="source-line-no">074</span><span id="line-74"> new byte[][] { INPUT_FAMILY, OUTPUT_FAMILY });</span>
<span class="source-line-no">075</span><span id="line-75"> UTIL.loadTable(table, INPUT_FAMILY, false);</span>
<span class="source-line-no">076</span><span id="line-76"> UTIL.createTable(TABLE_FOR_NEGATIVE_TESTS, new byte[][] { INPUT_FAMILY, OUTPUT_FAMILY });</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"> @AfterClass</span>
<span class="source-line-no">080</span><span id="line-80"> public static void afterClass() throws Exception {</span>
<span class="source-line-no">081</span><span id="line-81"> UTIL.deleteTable(TABLE_FOR_NEGATIVE_TESTS);</span>
<span class="source-line-no">082</span><span id="line-82"> UTIL.shutdownMiniCluster();</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"> /**</span>
<span class="source-line-no">086</span><span id="line-86"> * Test a map/reduce against a multi-region table</span>
<span class="source-line-no">087</span><span id="line-87"> */</span>
<span class="source-line-no">088</span><span id="line-88"> @Test</span>
<span class="source-line-no">089</span><span id="line-89"> public void testMultiRegionTable() throws IOException {</span>
<span class="source-line-no">090</span><span id="line-90"> runTestOnTable(UTIL.getConnection().getTable(MULTI_REGION_TABLE_NAME));</span>
<span class="source-line-no">091</span><span id="line-91"> }</span>
<span class="source-line-no">092</span><span id="line-92"></span>
<span class="source-line-no">093</span><span id="line-93"> @Test</span>
<span class="source-line-no">094</span><span id="line-94"> public void testCombiner() throws IOException {</span>
<span class="source-line-no">095</span><span id="line-95"> Configuration conf = new Configuration(UTIL.getConfiguration());</span>
<span class="source-line-no">096</span><span id="line-96"> // force use of combiner for testing purposes</span>
<span class="source-line-no">097</span><span id="line-97"> conf.setInt("mapreduce.map.combine.minspills", 1);</span>
<span class="source-line-no">098</span><span id="line-98"> runTestOnTable(UTIL.getConnection().getTable(MULTI_REGION_TABLE_NAME));</span>
<span class="source-line-no">099</span><span id="line-99"> }</span>
<span class="source-line-no">100</span><span id="line-100"></span>
<span class="source-line-no">101</span><span id="line-101"> /**</span>
<span class="source-line-no">102</span><span id="line-102"> * Implements mapper logic for use across APIs.</span>
<span class="source-line-no">103</span><span id="line-103"> */</span>
<span class="source-line-no">104</span><span id="line-104"> protected static Put map(ImmutableBytesWritable key, Result value) throws IOException {</span>
<span class="source-line-no">105</span><span id="line-105"> if (value.size() != 1) {</span>
<span class="source-line-no">106</span><span id="line-106"> throw new IOException("There should only be one input column");</span>
<span class="source-line-no">107</span><span id="line-107"> }</span>
<span class="source-line-no">108</span><span id="line-108"> Map&lt;byte[], NavigableMap&lt;byte[], NavigableMap&lt;Long, byte[]&gt;&gt;&gt; cf = value.getMap();</span>
<span class="source-line-no">109</span><span id="line-109"> if (!cf.containsKey(INPUT_FAMILY)) {</span>
<span class="source-line-no">110</span><span id="line-110"> throw new IOException(</span>
<span class="source-line-no">111</span><span id="line-111"> "Wrong input columns. Missing: '" + Bytes.toString(INPUT_FAMILY) + "'.");</span>
<span class="source-line-no">112</span><span id="line-112"> }</span>
<span class="source-line-no">113</span><span id="line-113"></span>
<span class="source-line-no">114</span><span id="line-114"> // Get the original value and reverse it</span>
<span class="source-line-no">115</span><span id="line-115"></span>
<span class="source-line-no">116</span><span id="line-116"> String originalValue = Bytes.toString(value.getValue(INPUT_FAMILY, INPUT_FAMILY));</span>
<span class="source-line-no">117</span><span id="line-117"> StringBuilder newValue = new StringBuilder(originalValue);</span>
<span class="source-line-no">118</span><span id="line-118"> newValue.reverse();</span>
<span class="source-line-no">119</span><span id="line-119"></span>
<span class="source-line-no">120</span><span id="line-120"> // Now set the value to be collected</span>
<span class="source-line-no">121</span><span id="line-121"></span>
<span class="source-line-no">122</span><span id="line-122"> Put outval = new Put(key.get());</span>
<span class="source-line-no">123</span><span id="line-123"> outval.addColumn(OUTPUT_FAMILY, null, Bytes.toBytes(newValue.toString()));</span>
<span class="source-line-no">124</span><span id="line-124"> return outval;</span>
<span class="source-line-no">125</span><span id="line-125"> }</span>
<span class="source-line-no">126</span><span id="line-126"></span>
<span class="source-line-no">127</span><span id="line-127"> protected void verify(TableName tableName) throws IOException {</span>
<span class="source-line-no">128</span><span id="line-128"> Table table = UTIL.getConnection().getTable(tableName);</span>
<span class="source-line-no">129</span><span id="line-129"> boolean verified = false;</span>
<span class="source-line-no">130</span><span id="line-130"> long pause = UTIL.getConfiguration().getLong("hbase.client.pause", 5 * 1000);</span>
<span class="source-line-no">131</span><span id="line-131"> int numRetries = UTIL.getConfiguration().getInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER, 5);</span>
<span class="source-line-no">132</span><span id="line-132"> for (int i = 0; i &lt; numRetries; i++) {</span>
<span class="source-line-no">133</span><span id="line-133"> try {</span>
<span class="source-line-no">134</span><span id="line-134"> getLog().info("Verification attempt #" + i);</span>
<span class="source-line-no">135</span><span id="line-135"> verifyAttempt(table);</span>
<span class="source-line-no">136</span><span id="line-136"> verified = true;</span>
<span class="source-line-no">137</span><span id="line-137"> break;</span>
<span class="source-line-no">138</span><span id="line-138"> } catch (NullPointerException e) {</span>
<span class="source-line-no">139</span><span id="line-139"> // If here, a cell was empty. Presume its because updates came in</span>
<span class="source-line-no">140</span><span id="line-140"> // after the scanner had been opened. Wait a while and retry.</span>
<span class="source-line-no">141</span><span id="line-141"> getLog().debug("Verification attempt failed: " + e.getMessage());</span>
<span class="source-line-no">142</span><span id="line-142"> }</span>
<span class="source-line-no">143</span><span id="line-143"> try {</span>
<span class="source-line-no">144</span><span id="line-144"> Thread.sleep(pause);</span>
<span class="source-line-no">145</span><span id="line-145"> } catch (InterruptedException e) {</span>
<span class="source-line-no">146</span><span id="line-146"> // continue</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"> assertTrue(verified);</span>
<span class="source-line-no">150</span><span id="line-150"> }</span>
<span class="source-line-no">151</span><span id="line-151"></span>
<span class="source-line-no">152</span><span id="line-152"> /**</span>
<span class="source-line-no">153</span><span id="line-153"> * Looks at every value of the mapreduce output and verifies that indeed the values have been</span>
<span class="source-line-no">154</span><span id="line-154"> * reversed.</span>
<span class="source-line-no">155</span><span id="line-155"> * @param table Table to scan.</span>
<span class="source-line-no">156</span><span id="line-156"> * @throws NullPointerException if we failed to find a cell value</span>
<span class="source-line-no">157</span><span id="line-157"> */</span>
<span class="source-line-no">158</span><span id="line-158"> private void verifyAttempt(final Table table) throws IOException, NullPointerException {</span>
<span class="source-line-no">159</span><span id="line-159"> Scan scan = new Scan();</span>
<span class="source-line-no">160</span><span id="line-160"> TableInputFormat.addColumns(scan, columns);</span>
<span class="source-line-no">161</span><span id="line-161"> ResultScanner scanner = table.getScanner(scan);</span>
<span class="source-line-no">162</span><span id="line-162"> try {</span>
<span class="source-line-no">163</span><span id="line-163"> Iterator&lt;Result&gt; itr = scanner.iterator();</span>
<span class="source-line-no">164</span><span id="line-164"> assertTrue(itr.hasNext());</span>
<span class="source-line-no">165</span><span id="line-165"> while (itr.hasNext()) {</span>
<span class="source-line-no">166</span><span id="line-166"> Result r = itr.next();</span>
<span class="source-line-no">167</span><span id="line-167"> if (getLog().isDebugEnabled()) {</span>
<span class="source-line-no">168</span><span id="line-168"> if (r.size() &gt; 2) {</span>
<span class="source-line-no">169</span><span id="line-169"> throw new IOException("Too many results, expected 2 got " + r.size());</span>
<span class="source-line-no">170</span><span id="line-170"> }</span>
<span class="source-line-no">171</span><span id="line-171"> }</span>
<span class="source-line-no">172</span><span id="line-172"> byte[] firstValue = null;</span>
<span class="source-line-no">173</span><span id="line-173"> byte[] secondValue = null;</span>
<span class="source-line-no">174</span><span id="line-174"> int count = 0;</span>
<span class="source-line-no">175</span><span id="line-175"> for (Cell kv : r.listCells()) {</span>
<span class="source-line-no">176</span><span id="line-176"> if (count == 0) {</span>
<span class="source-line-no">177</span><span id="line-177"> firstValue = CellUtil.cloneValue(kv);</span>
<span class="source-line-no">178</span><span id="line-178"> }</span>
<span class="source-line-no">179</span><span id="line-179"> if (count == 1) {</span>
<span class="source-line-no">180</span><span id="line-180"> secondValue = CellUtil.cloneValue(kv);</span>
<span class="source-line-no">181</span><span id="line-181"> }</span>
<span class="source-line-no">182</span><span id="line-182"> count++;</span>
<span class="source-line-no">183</span><span id="line-183"> if (count == 2) {</span>
<span class="source-line-no">184</span><span id="line-184"> break;</span>
<span class="source-line-no">185</span><span id="line-185"> }</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"> if (firstValue == null) {</span>
<span class="source-line-no">189</span><span id="line-189"> throw new NullPointerException(Bytes.toString(r.getRow()) + ": first value is null");</span>
<span class="source-line-no">190</span><span id="line-190"> }</span>
<span class="source-line-no">191</span><span id="line-191"> String first = Bytes.toString(firstValue);</span>
<span class="source-line-no">192</span><span id="line-192"></span>
<span class="source-line-no">193</span><span id="line-193"> if (secondValue == null) {</span>
<span class="source-line-no">194</span><span id="line-194"> throw new NullPointerException(Bytes.toString(r.getRow()) + ": second value is null");</span>
<span class="source-line-no">195</span><span id="line-195"> }</span>
<span class="source-line-no">196</span><span id="line-196"> byte[] secondReversed = new byte[secondValue.length];</span>
<span class="source-line-no">197</span><span id="line-197"> for (int i = 0, j = secondValue.length - 1; j &gt;= 0; j--, i++) {</span>
<span class="source-line-no">198</span><span id="line-198"> secondReversed[i] = secondValue[j];</span>
<span class="source-line-no">199</span><span id="line-199"> }</span>
<span class="source-line-no">200</span><span id="line-200"> String second = Bytes.toString(secondReversed);</span>
<span class="source-line-no">201</span><span id="line-201"></span>
<span class="source-line-no">202</span><span id="line-202"> if (first.compareTo(second) != 0) {</span>
<span class="source-line-no">203</span><span id="line-203"> if (getLog().isDebugEnabled()) {</span>
<span class="source-line-no">204</span><span id="line-204"> getLog().debug(</span>
<span class="source-line-no">205</span><span id="line-205"> "second key is not the reverse of first. row=" + Bytes.toStringBinary(r.getRow())</span>
<span class="source-line-no">206</span><span id="line-206"> + ", first value=" + first + ", second value=" + second);</span>
<span class="source-line-no">207</span><span id="line-207"> }</span>
<span class="source-line-no">208</span><span id="line-208"> fail();</span>
<span class="source-line-no">209</span><span id="line-209"> }</span>
<span class="source-line-no">210</span><span id="line-210"> }</span>
<span class="source-line-no">211</span><span id="line-211"> } finally {</span>
<span class="source-line-no">212</span><span id="line-212"> scanner.close();</span>
<span class="source-line-no">213</span><span id="line-213"> }</span>
<span class="source-line-no">214</span><span id="line-214"> }</span>
<span class="source-line-no">215</span><span id="line-215">}</span>
</pre>
</div>
</main>
</body>
</html>