blob: 9124a20e2e8a5d2b730a906f307f8f231bd7dde8 [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.test, class: IntegrationTestLoadAndVerify, class: LoadMapper">
<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.test;</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.assertEquals;</span>
<span class="source-line-no">021</span><span id="line-21">import static org.junit.Assert.assertTrue;</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.BufferedReader;</span>
<span class="source-line-no">024</span><span id="line-24">import java.io.FileNotFoundException;</span>
<span class="source-line-no">025</span><span id="line-25">import java.io.IOException;</span>
<span class="source-line-no">026</span><span id="line-26">import java.io.InputStream;</span>
<span class="source-line-no">027</span><span id="line-27">import java.io.InputStreamReader;</span>
<span class="source-line-no">028</span><span id="line-28">import java.io.InterruptedIOException;</span>
<span class="source-line-no">029</span><span id="line-29">import java.nio.charset.StandardCharsets;</span>
<span class="source-line-no">030</span><span id="line-30">import java.util.List;</span>
<span class="source-line-no">031</span><span id="line-31">import java.util.Random;</span>
<span class="source-line-no">032</span><span id="line-32">import java.util.Set;</span>
<span class="source-line-no">033</span><span id="line-33">import java.util.SortedSet;</span>
<span class="source-line-no">034</span><span id="line-34">import java.util.TreeSet;</span>
<span class="source-line-no">035</span><span id="line-35">import java.util.concurrent.ThreadLocalRandom;</span>
<span class="source-line-no">036</span><span id="line-36">import java.util.concurrent.atomic.AtomicInteger;</span>
<span class="source-line-no">037</span><span id="line-37">import java.util.regex.Matcher;</span>
<span class="source-line-no">038</span><span id="line-38">import java.util.regex.Pattern;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.conf.Configuration;</span>
<span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.fs.FileStatus;</span>
<span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.fs.FileSystem;</span>
<span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.fs.LocatedFileStatus;</span>
<span class="source-line-no">043</span><span id="line-43">import org.apache.hadoop.fs.Path;</span>
<span class="source-line-no">044</span><span id="line-44">import org.apache.hadoop.fs.RemoteIterator;</span>
<span class="source-line-no">045</span><span id="line-45">import org.apache.hadoop.hbase.Cell;</span>
<span class="source-line-no">046</span><span id="line-46">import org.apache.hadoop.hbase.HBaseConfiguration;</span>
<span class="source-line-no">047</span><span id="line-47">import org.apache.hadoop.hbase.HConstants;</span>
<span class="source-line-no">048</span><span id="line-48">import org.apache.hadoop.hbase.IntegrationTestBase;</span>
<span class="source-line-no">049</span><span id="line-49">import org.apache.hadoop.hbase.IntegrationTestingUtility;</span>
<span class="source-line-no">050</span><span id="line-50">import org.apache.hadoop.hbase.TableName;</span>
<span class="source-line-no">051</span><span id="line-51">import org.apache.hadoop.hbase.client.Admin;</span>
<span class="source-line-no">052</span><span id="line-52">import org.apache.hadoop.hbase.client.BufferedMutator;</span>
<span class="source-line-no">053</span><span id="line-53">import org.apache.hadoop.hbase.client.BufferedMutatorParams;</span>
<span class="source-line-no">054</span><span id="line-54">import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;</span>
<span class="source-line-no">055</span><span id="line-55">import org.apache.hadoop.hbase.client.Connection;</span>
<span class="source-line-no">056</span><span id="line-56">import org.apache.hadoop.hbase.client.ConnectionFactory;</span>
<span class="source-line-no">057</span><span id="line-57">import org.apache.hadoop.hbase.client.Mutation;</span>
<span class="source-line-no">058</span><span id="line-58">import org.apache.hadoop.hbase.client.Put;</span>
<span class="source-line-no">059</span><span id="line-59">import org.apache.hadoop.hbase.client.Result;</span>
<span class="source-line-no">060</span><span id="line-60">import org.apache.hadoop.hbase.client.Scan;</span>
<span class="source-line-no">061</span><span id="line-61">import org.apache.hadoop.hbase.client.TableDescriptor;</span>
<span class="source-line-no">062</span><span id="line-62">import org.apache.hadoop.hbase.client.TableDescriptorBuilder;</span>
<span class="source-line-no">063</span><span id="line-63">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span>
<span class="source-line-no">064</span><span id="line-64">import org.apache.hadoop.hbase.mapreduce.NMapInputFormat;</span>
<span class="source-line-no">065</span><span id="line-65">import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;</span>
<span class="source-line-no">066</span><span id="line-66">import org.apache.hadoop.hbase.mapreduce.TableMapper;</span>
<span class="source-line-no">067</span><span id="line-67">import org.apache.hadoop.hbase.mapreduce.TableRecordReaderImpl;</span>
<span class="source-line-no">068</span><span id="line-68">import org.apache.hadoop.hbase.mapreduce.WALPlayer;</span>
<span class="source-line-no">069</span><span id="line-69">import org.apache.hadoop.hbase.testclassification.IntegrationTests;</span>
<span class="source-line-no">070</span><span id="line-70">import org.apache.hadoop.hbase.util.AbstractHBaseTool;</span>
<span class="source-line-no">071</span><span id="line-71">import org.apache.hadoop.hbase.util.Bytes;</span>
<span class="source-line-no">072</span><span id="line-72">import org.apache.hadoop.hbase.util.CommonFSUtils;</span>
<span class="source-line-no">073</span><span id="line-73">import org.apache.hadoop.hbase.wal.WALEdit;</span>
<span class="source-line-no">074</span><span id="line-74">import org.apache.hadoop.hbase.wal.WALKey;</span>
<span class="source-line-no">075</span><span id="line-75">import org.apache.hadoop.io.BytesWritable;</span>
<span class="source-line-no">076</span><span id="line-76">import org.apache.hadoop.io.NullWritable;</span>
<span class="source-line-no">077</span><span id="line-77">import org.apache.hadoop.io.Text;</span>
<span class="source-line-no">078</span><span id="line-78">import org.apache.hadoop.mapreduce.Counter;</span>
<span class="source-line-no">079</span><span id="line-79">import org.apache.hadoop.mapreduce.Job;</span>
<span class="source-line-no">080</span><span id="line-80">import org.apache.hadoop.mapreduce.Mapper;</span>
<span class="source-line-no">081</span><span id="line-81">import org.apache.hadoop.mapreduce.Reducer;</span>
<span class="source-line-no">082</span><span id="line-82">import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;</span>
<span class="source-line-no">083</span><span id="line-83">import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;</span>
<span class="source-line-no">084</span><span id="line-84">import org.apache.hadoop.util.ToolRunner;</span>
<span class="source-line-no">085</span><span id="line-85">import org.junit.Test;</span>
<span class="source-line-no">086</span><span id="line-86">import org.junit.experimental.categories.Category;</span>
<span class="source-line-no">087</span><span id="line-87">import org.slf4j.Logger;</span>
<span class="source-line-no">088</span><span id="line-88">import org.slf4j.LoggerFactory;</span>
<span class="source-line-no">089</span><span id="line-89"></span>
<span class="source-line-no">090</span><span id="line-90">import org.apache.hbase.thirdparty.com.google.common.base.Splitter;</span>
<span class="source-line-no">091</span><span id="line-91">import org.apache.hbase.thirdparty.com.google.common.collect.Iterables;</span>
<span class="source-line-no">092</span><span id="line-92">import org.apache.hbase.thirdparty.com.google.common.collect.Sets;</span>
<span class="source-line-no">093</span><span id="line-93">import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;</span>
<span class="source-line-no">094</span><span id="line-94"></span>
<span class="source-line-no">095</span><span id="line-95">/**</span>
<span class="source-line-no">096</span><span id="line-96"> * A large test which loads a lot of data that has internal references, and verifies the data. In</span>
<span class="source-line-no">097</span><span id="line-97"> * load step, 200 map tasks are launched, which in turn write loadmapper.num_to_write (default 100K)</span>
<span class="source-line-no">098</span><span id="line-98"> * rows to an hbase table. Rows are written in blocks, for a total of 100 blocks. Each row in a</span>
<span class="source-line-no">099</span><span id="line-99"> * block, contains loadmapper.backrefs (default 50) references to random rows in the prev block.</span>
<span class="source-line-no">100</span><span id="line-100"> * Verify step is scans the table, and verifies that for every referenced row, the row is actually</span>
<span class="source-line-no">101</span><span id="line-101"> * there (no data loss). Failed rows are output from reduce to be saved in the job output dir in</span>
<span class="source-line-no">102</span><span id="line-102"> * hdfs and inspected later. This class can be run as a unit test, as an integration test, or from</span>
<span class="source-line-no">103</span><span id="line-103"> * the command line Originally taken from Apache Bigtop.</span>
<span class="source-line-no">104</span><span id="line-104"> */</span>
<span class="source-line-no">105</span><span id="line-105">@Category(IntegrationTests.class)</span>
<span class="source-line-no">106</span><span id="line-106">public class IntegrationTestLoadAndVerify extends IntegrationTestBase {</span>
<span class="source-line-no">107</span><span id="line-107"></span>
<span class="source-line-no">108</span><span id="line-108"> private static final Logger LOG = LoggerFactory.getLogger(IntegrationTestLoadAndVerify.class);</span>
<span class="source-line-no">109</span><span id="line-109"></span>
<span class="source-line-no">110</span><span id="line-110"> private static final String TEST_NAME = "IntegrationTestLoadAndVerify";</span>
<span class="source-line-no">111</span><span id="line-111"> private static final byte[] TEST_FAMILY = Bytes.toBytes("f1");</span>
<span class="source-line-no">112</span><span id="line-112"> private static final byte[] TEST_QUALIFIER = Bytes.toBytes("q1");</span>
<span class="source-line-no">113</span><span id="line-113"></span>
<span class="source-line-no">114</span><span id="line-114"> private static final String NUM_TO_WRITE_KEY = "loadmapper.num_to_write";</span>
<span class="source-line-no">115</span><span id="line-115"> private static final long NUM_TO_WRITE_DEFAULT = 100 * 1000;</span>
<span class="source-line-no">116</span><span id="line-116"></span>
<span class="source-line-no">117</span><span id="line-117"> private static final String TABLE_NAME_KEY = "loadmapper.table";</span>
<span class="source-line-no">118</span><span id="line-118"> private static final String TABLE_NAME_DEFAULT = "table";</span>
<span class="source-line-no">119</span><span id="line-119"></span>
<span class="source-line-no">120</span><span id="line-120"> private static final String NUM_BACKREFS_KEY = "loadmapper.backrefs";</span>
<span class="source-line-no">121</span><span id="line-121"> private static final int NUM_BACKREFS_DEFAULT = 50;</span>
<span class="source-line-no">122</span><span id="line-122"></span>
<span class="source-line-no">123</span><span id="line-123"> private static final String NUM_MAP_TASKS_KEY = "loadmapper.map.tasks";</span>
<span class="source-line-no">124</span><span id="line-124"> private static final String NUM_REDUCE_TASKS_KEY = "verify.reduce.tasks";</span>
<span class="source-line-no">125</span><span id="line-125"> private static final int NUM_MAP_TASKS_DEFAULT = 200;</span>
<span class="source-line-no">126</span><span id="line-126"> private static final int NUM_REDUCE_TASKS_DEFAULT = 35;</span>
<span class="source-line-no">127</span><span id="line-127"></span>
<span class="source-line-no">128</span><span id="line-128"> private static final int SCANNER_CACHING = 500;</span>
<span class="source-line-no">129</span><span id="line-129"></span>
<span class="source-line-no">130</span><span id="line-130"> private static final int MISSING_ROWS_TO_LOG = 10; // YARN complains when too many counters</span>
<span class="source-line-no">131</span><span id="line-131"></span>
<span class="source-line-no">132</span><span id="line-132"> private String toRun = null;</span>
<span class="source-line-no">133</span><span id="line-133"> private String keysDir = null;</span>
<span class="source-line-no">134</span><span id="line-134"></span>
<span class="source-line-no">135</span><span id="line-135"> private enum Counters {</span>
<span class="source-line-no">136</span><span id="line-136"> ROWS_WRITTEN,</span>
<span class="source-line-no">137</span><span id="line-137"> REFERENCES_WRITTEN,</span>
<span class="source-line-no">138</span><span id="line-138"> REFERENCES_CHECKED</span>
<span class="source-line-no">139</span><span id="line-139"> }</span>
<span class="source-line-no">140</span><span id="line-140"></span>
<span class="source-line-no">141</span><span id="line-141"> @Override</span>
<span class="source-line-no">142</span><span id="line-142"> public void setUpCluster() throws Exception {</span>
<span class="source-line-no">143</span><span id="line-143"> util = getTestingUtil(getConf());</span>
<span class="source-line-no">144</span><span id="line-144"> util.initializeCluster(3);</span>
<span class="source-line-no">145</span><span id="line-145"> this.setConf(util.getConfiguration());</span>
<span class="source-line-no">146</span><span id="line-146"> if (!util.isDistributedCluster()) {</span>
<span class="source-line-no">147</span><span id="line-147"> getConf().setLong(NUM_TO_WRITE_KEY, NUM_TO_WRITE_DEFAULT / 100);</span>
<span class="source-line-no">148</span><span id="line-148"> getConf().setInt(NUM_MAP_TASKS_KEY, NUM_MAP_TASKS_DEFAULT / 100);</span>
<span class="source-line-no">149</span><span id="line-149"> getConf().setInt(NUM_REDUCE_TASKS_KEY, NUM_REDUCE_TASKS_DEFAULT / 10);</span>
<span class="source-line-no">150</span><span id="line-150"> util.startMiniMapReduceCluster();</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"></span>
<span class="source-line-no">154</span><span id="line-154"> @Override</span>
<span class="source-line-no">155</span><span id="line-155"> public void cleanUpCluster() throws Exception {</span>
<span class="source-line-no">156</span><span id="line-156"> super.cleanUpCluster();</span>
<span class="source-line-no">157</span><span id="line-157"> if (!util.isDistributedCluster()) {</span>
<span class="source-line-no">158</span><span id="line-158"> util.shutdownMiniMapReduceCluster();</span>
<span class="source-line-no">159</span><span id="line-159"> }</span>
<span class="source-line-no">160</span><span id="line-160"> }</span>
<span class="source-line-no">161</span><span id="line-161"></span>
<span class="source-line-no">162</span><span id="line-162"> /**</span>
<span class="source-line-no">163</span><span id="line-163"> * Converts a "long" value between endian systems. Borrowed from Apache Commons IO</span>
<span class="source-line-no">164</span><span id="line-164"> * @param value value to convert</span>
<span class="source-line-no">165</span><span id="line-165"> * @return the converted value</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 long swapLong(long value) {</span>
<span class="source-line-no">168</span><span id="line-168"> return (((value &gt;&gt; 0) &amp; 0xff) &lt;&lt; 56) + (((value &gt;&gt; 8) &amp; 0xff) &lt;&lt; 48)</span>
<span class="source-line-no">169</span><span id="line-169"> + (((value &gt;&gt; 16) &amp; 0xff) &lt;&lt; 40) + (((value &gt;&gt; 24) &amp; 0xff) &lt;&lt; 32)</span>
<span class="source-line-no">170</span><span id="line-170"> + (((value &gt;&gt; 32) &amp; 0xff) &lt;&lt; 24) + (((value &gt;&gt; 40) &amp; 0xff) &lt;&lt; 16)</span>
<span class="source-line-no">171</span><span id="line-171"> + (((value &gt;&gt; 48) &amp; 0xff) &lt;&lt; 8) + (((value &gt;&gt; 56) &amp; 0xff) &lt;&lt; 0);</span>
<span class="source-line-no">172</span><span id="line-172"> }</span>
<span class="source-line-no">173</span><span id="line-173"></span>
<span class="source-line-no">174</span><span id="line-174"> public static class LoadMapper</span>
<span class="source-line-no">175</span><span id="line-175"> extends Mapper&lt;NullWritable, NullWritable, NullWritable, NullWritable&gt; {</span>
<span class="source-line-no">176</span><span id="line-176"> protected long recordsToWrite;</span>
<span class="source-line-no">177</span><span id="line-177"> protected Connection connection;</span>
<span class="source-line-no">178</span><span id="line-178"> protected BufferedMutator mutator;</span>
<span class="source-line-no">179</span><span id="line-179"> protected Configuration conf;</span>
<span class="source-line-no">180</span><span id="line-180"> protected int numBackReferencesPerRow;</span>
<span class="source-line-no">181</span><span id="line-181"> protected String shortTaskId;</span>
<span class="source-line-no">182</span><span id="line-182"> protected Counter rowsWritten, refsWritten;</span>
<span class="source-line-no">183</span><span id="line-183"></span>
<span class="source-line-no">184</span><span id="line-184"> @Override</span>
<span class="source-line-no">185</span><span id="line-185"> public void setup(Context context) throws IOException {</span>
<span class="source-line-no">186</span><span id="line-186"> conf = context.getConfiguration();</span>
<span class="source-line-no">187</span><span id="line-187"> recordsToWrite = conf.getLong(NUM_TO_WRITE_KEY, NUM_TO_WRITE_DEFAULT);</span>
<span class="source-line-no">188</span><span id="line-188"> String tableName = conf.get(TABLE_NAME_KEY, TABLE_NAME_DEFAULT);</span>
<span class="source-line-no">189</span><span id="line-189"> numBackReferencesPerRow = conf.getInt(NUM_BACKREFS_KEY, NUM_BACKREFS_DEFAULT);</span>
<span class="source-line-no">190</span><span id="line-190"> this.connection = ConnectionFactory.createConnection(conf);</span>
<span class="source-line-no">191</span><span id="line-191"> mutator = connection.getBufferedMutator(</span>
<span class="source-line-no">192</span><span id="line-192"> new BufferedMutatorParams(TableName.valueOf(tableName)).writeBufferSize(4 * 1024 * 1024));</span>
<span class="source-line-no">193</span><span id="line-193"></span>
<span class="source-line-no">194</span><span id="line-194"> String taskId = conf.get("mapreduce.task.attempt.id");</span>
<span class="source-line-no">195</span><span id="line-195"> Matcher matcher = Pattern.compile(".+_m_(\\d+_\\d+)").matcher(taskId);</span>
<span class="source-line-no">196</span><span id="line-196"> if (!matcher.matches()) {</span>
<span class="source-line-no">197</span><span id="line-197"> throw new RuntimeException("Strange task ID: " + taskId);</span>
<span class="source-line-no">198</span><span id="line-198"> }</span>
<span class="source-line-no">199</span><span id="line-199"> shortTaskId = matcher.group(1);</span>
<span class="source-line-no">200</span><span id="line-200"></span>
<span class="source-line-no">201</span><span id="line-201"> rowsWritten = context.getCounter(Counters.ROWS_WRITTEN);</span>
<span class="source-line-no">202</span><span id="line-202"> refsWritten = context.getCounter(Counters.REFERENCES_WRITTEN);</span>
<span class="source-line-no">203</span><span id="line-203"> }</span>
<span class="source-line-no">204</span><span id="line-204"></span>
<span class="source-line-no">205</span><span id="line-205"> @Override</span>
<span class="source-line-no">206</span><span id="line-206"> public void cleanup(Context context) throws IOException {</span>
<span class="source-line-no">207</span><span id="line-207"> mutator.close();</span>
<span class="source-line-no">208</span><span id="line-208"> connection.close();</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"> @Override</span>
<span class="source-line-no">212</span><span id="line-212"> protected void map(NullWritable key, NullWritable value, Context context)</span>
<span class="source-line-no">213</span><span id="line-213"> throws IOException, InterruptedException {</span>
<span class="source-line-no">214</span><span id="line-214"></span>
<span class="source-line-no">215</span><span id="line-215"> String suffix = "/" + shortTaskId;</span>
<span class="source-line-no">216</span><span id="line-216"> byte[] row = Bytes.add(new byte[8], Bytes.toBytes(suffix));</span>
<span class="source-line-no">217</span><span id="line-217"> int BLOCK_SIZE = (int) (recordsToWrite / 100);</span>
<span class="source-line-no">218</span><span id="line-218"> Random rand = ThreadLocalRandom.current();</span>
<span class="source-line-no">219</span><span id="line-219"></span>
<span class="source-line-no">220</span><span id="line-220"> for (long i = 0; i &lt; recordsToWrite;) {</span>
<span class="source-line-no">221</span><span id="line-221"> long blockStart = i;</span>
<span class="source-line-no">222</span><span id="line-222"> for (long idxInBlock = 0; idxInBlock &lt; BLOCK_SIZE</span>
<span class="source-line-no">223</span><span id="line-223"> &amp;&amp; i &lt; recordsToWrite; idxInBlock++, i++) {</span>
<span class="source-line-no">224</span><span id="line-224"></span>
<span class="source-line-no">225</span><span id="line-225"> long byteSwapped = swapLong(i);</span>
<span class="source-line-no">226</span><span id="line-226"> Bytes.putLong(row, 0, byteSwapped);</span>
<span class="source-line-no">227</span><span id="line-227"></span>
<span class="source-line-no">228</span><span id="line-228"> Put p = new Put(row);</span>
<span class="source-line-no">229</span><span id="line-229"> p.addColumn(TEST_FAMILY, TEST_QUALIFIER, HConstants.EMPTY_BYTE_ARRAY);</span>
<span class="source-line-no">230</span><span id="line-230"> if (blockStart &gt; 0) {</span>
<span class="source-line-no">231</span><span id="line-231"> for (int j = 0; j &lt; numBackReferencesPerRow; j++) {</span>
<span class="source-line-no">232</span><span id="line-232"> long referredRow = blockStart - BLOCK_SIZE + rand.nextInt(BLOCK_SIZE);</span>
<span class="source-line-no">233</span><span id="line-233"> Bytes.putLong(row, 0, swapLong(referredRow));</span>
<span class="source-line-no">234</span><span id="line-234"> p.addColumn(TEST_FAMILY, row, HConstants.EMPTY_BYTE_ARRAY);</span>
<span class="source-line-no">235</span><span id="line-235"> }</span>
<span class="source-line-no">236</span><span id="line-236"> refsWritten.increment(1);</span>
<span class="source-line-no">237</span><span id="line-237"> }</span>
<span class="source-line-no">238</span><span id="line-238"> rowsWritten.increment(1);</span>
<span class="source-line-no">239</span><span id="line-239"> mutator.mutate(p);</span>
<span class="source-line-no">240</span><span id="line-240"></span>
<span class="source-line-no">241</span><span id="line-241"> if (i % 100 == 0) {</span>
<span class="source-line-no">242</span><span id="line-242"> context.setStatus("Written " + i + "/" + recordsToWrite + " records");</span>
<span class="source-line-no">243</span><span id="line-243"> context.progress();</span>
<span class="source-line-no">244</span><span id="line-244"> }</span>
<span class="source-line-no">245</span><span id="line-245"> }</span>
<span class="source-line-no">246</span><span id="line-246"> // End of block, flush all of them before we start writing anything</span>
<span class="source-line-no">247</span><span id="line-247"> // pointing to these!</span>
<span class="source-line-no">248</span><span id="line-248"> mutator.flush();</span>
<span class="source-line-no">249</span><span id="line-249"> }</span>
<span class="source-line-no">250</span><span id="line-250"> }</span>
<span class="source-line-no">251</span><span id="line-251"> }</span>
<span class="source-line-no">252</span><span id="line-252"></span>
<span class="source-line-no">253</span><span id="line-253"> public static class VerifyMapper extends TableMapper&lt;BytesWritable, BytesWritable&gt; {</span>
<span class="source-line-no">254</span><span id="line-254"> static final BytesWritable EMPTY = new BytesWritable(HConstants.EMPTY_BYTE_ARRAY);</span>
<span class="source-line-no">255</span><span id="line-255"></span>
<span class="source-line-no">256</span><span id="line-256"> @Override</span>
<span class="source-line-no">257</span><span id="line-257"> protected void map(ImmutableBytesWritable key, Result value, Context context)</span>
<span class="source-line-no">258</span><span id="line-258"> throws IOException, InterruptedException {</span>
<span class="source-line-no">259</span><span id="line-259"> BytesWritable bwKey = new BytesWritable(key.get());</span>
<span class="source-line-no">260</span><span id="line-260"> BytesWritable bwVal = new BytesWritable();</span>
<span class="source-line-no">261</span><span id="line-261"> for (Cell kv : value.listCells()) {</span>
<span class="source-line-no">262</span><span id="line-262"> if (</span>
<span class="source-line-no">263</span><span id="line-263"> Bytes.compareTo(TEST_QUALIFIER, 0, TEST_QUALIFIER.length, kv.getQualifierArray(),</span>
<span class="source-line-no">264</span><span id="line-264"> kv.getQualifierOffset(), kv.getQualifierLength()) == 0</span>
<span class="source-line-no">265</span><span id="line-265"> ) {</span>
<span class="source-line-no">266</span><span id="line-266"> context.write(bwKey, EMPTY);</span>
<span class="source-line-no">267</span><span id="line-267"> } else {</span>
<span class="source-line-no">268</span><span id="line-268"> bwVal.set(kv.getQualifierArray(), kv.getQualifierOffset(), kv.getQualifierLength());</span>
<span class="source-line-no">269</span><span id="line-269"> context.write(bwVal, bwKey);</span>
<span class="source-line-no">270</span><span id="line-270"> }</span>
<span class="source-line-no">271</span><span id="line-271"> }</span>
<span class="source-line-no">272</span><span id="line-272"> }</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"> public static class VerifyReducer extends Reducer&lt;BytesWritable, BytesWritable, Text, Text&gt; {</span>
<span class="source-line-no">276</span><span id="line-276"> private Counter refsChecked;</span>
<span class="source-line-no">277</span><span id="line-277"> private Counter rowsWritten;</span>
<span class="source-line-no">278</span><span id="line-278"></span>
<span class="source-line-no">279</span><span id="line-279"> @Override</span>
<span class="source-line-no">280</span><span id="line-280"> public void setup(Context context) throws IOException {</span>
<span class="source-line-no">281</span><span id="line-281"> refsChecked = context.getCounter(Counters.REFERENCES_CHECKED);</span>
<span class="source-line-no">282</span><span id="line-282"> rowsWritten = context.getCounter(Counters.ROWS_WRITTEN);</span>
<span class="source-line-no">283</span><span id="line-283"> }</span>
<span class="source-line-no">284</span><span id="line-284"></span>
<span class="source-line-no">285</span><span id="line-285"> @Override</span>
<span class="source-line-no">286</span><span id="line-286"> protected void reduce(BytesWritable referredRow, Iterable&lt;BytesWritable&gt; referrers,</span>
<span class="source-line-no">287</span><span id="line-287"> VerifyReducer.Context ctx) throws IOException, InterruptedException {</span>
<span class="source-line-no">288</span><span id="line-288"> boolean gotOriginalRow = false;</span>
<span class="source-line-no">289</span><span id="line-289"> int refCount = 0;</span>
<span class="source-line-no">290</span><span id="line-290"></span>
<span class="source-line-no">291</span><span id="line-291"> for (BytesWritable ref : referrers) {</span>
<span class="source-line-no">292</span><span id="line-292"> if (ref.getLength() == 0) {</span>
<span class="source-line-no">293</span><span id="line-293"> assert !gotOriginalRow;</span>
<span class="source-line-no">294</span><span id="line-294"> gotOriginalRow = true;</span>
<span class="source-line-no">295</span><span id="line-295"> } else {</span>
<span class="source-line-no">296</span><span id="line-296"> refCount++;</span>
<span class="source-line-no">297</span><span id="line-297"> }</span>
<span class="source-line-no">298</span><span id="line-298"> }</span>
<span class="source-line-no">299</span><span id="line-299"> refsChecked.increment(refCount);</span>
<span class="source-line-no">300</span><span id="line-300"></span>
<span class="source-line-no">301</span><span id="line-301"> if (!gotOriginalRow) {</span>
<span class="source-line-no">302</span><span id="line-302"> String parsedRow = makeRowReadable(referredRow.getBytes(), referredRow.getLength());</span>
<span class="source-line-no">303</span><span id="line-303"> String binRow = Bytes.toStringBinary(referredRow.getBytes(), 0, referredRow.getLength());</span>
<span class="source-line-no">304</span><span id="line-304"> LOG.error("Reference error row " + parsedRow);</span>
<span class="source-line-no">305</span><span id="line-305"> ctx.write(new Text(binRow), new Text(parsedRow));</span>
<span class="source-line-no">306</span><span id="line-306"> rowsWritten.increment(1);</span>
<span class="source-line-no">307</span><span id="line-307"> }</span>
<span class="source-line-no">308</span><span id="line-308"> }</span>
<span class="source-line-no">309</span><span id="line-309"></span>
<span class="source-line-no">310</span><span id="line-310"> private String makeRowReadable(byte[] bytes, int length) {</span>
<span class="source-line-no">311</span><span id="line-311"> long rowIdx = swapLong(Bytes.toLong(bytes, 0));</span>
<span class="source-line-no">312</span><span id="line-312"> String suffix = Bytes.toString(bytes, 8, length - 8);</span>
<span class="source-line-no">313</span><span id="line-313"></span>
<span class="source-line-no">314</span><span id="line-314"> return "Row #" + rowIdx + " suffix " + suffix;</span>
<span class="source-line-no">315</span><span id="line-315"> }</span>
<span class="source-line-no">316</span><span id="line-316"> }</span>
<span class="source-line-no">317</span><span id="line-317"></span>
<span class="source-line-no">318</span><span id="line-318"> protected Job doLoad(Configuration conf, TableDescriptor tableDescriptor) throws Exception {</span>
<span class="source-line-no">319</span><span id="line-319"> Path outputDir = getTestDir(TEST_NAME, "load-output");</span>
<span class="source-line-no">320</span><span id="line-320"> LOG.info("Load output dir: " + outputDir);</span>
<span class="source-line-no">321</span><span id="line-321"></span>
<span class="source-line-no">322</span><span id="line-322"> NMapInputFormat.setNumMapTasks(conf, conf.getInt(NUM_MAP_TASKS_KEY, NUM_MAP_TASKS_DEFAULT));</span>
<span class="source-line-no">323</span><span id="line-323"> conf.set(TABLE_NAME_KEY, tableDescriptor.getTableName().getNameAsString());</span>
<span class="source-line-no">324</span><span id="line-324"></span>
<span class="source-line-no">325</span><span id="line-325"> Job job = Job.getInstance(conf);</span>
<span class="source-line-no">326</span><span id="line-326"> job.setJobName(TEST_NAME + " Load for " + tableDescriptor.getTableName());</span>
<span class="source-line-no">327</span><span id="line-327"> job.setJarByClass(this.getClass());</span>
<span class="source-line-no">328</span><span id="line-328"> setMapperClass(job);</span>
<span class="source-line-no">329</span><span id="line-329"> job.setInputFormatClass(NMapInputFormat.class);</span>
<span class="source-line-no">330</span><span id="line-330"> job.setNumReduceTasks(0);</span>
<span class="source-line-no">331</span><span id="line-331"> setJobScannerConf(job);</span>
<span class="source-line-no">332</span><span id="line-332"> FileOutputFormat.setOutputPath(job, outputDir);</span>
<span class="source-line-no">333</span><span id="line-333"></span>
<span class="source-line-no">334</span><span id="line-334"> TableMapReduceUtil.addDependencyJars(job);</span>
<span class="source-line-no">335</span><span id="line-335"></span>
<span class="source-line-no">336</span><span id="line-336"> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(), AbstractHBaseTool.class);</span>
<span class="source-line-no">337</span><span id="line-337"> TableMapReduceUtil.initCredentials(job);</span>
<span class="source-line-no">338</span><span id="line-338"> assertTrue(job.waitForCompletion(true));</span>
<span class="source-line-no">339</span><span id="line-339"> return job;</span>
<span class="source-line-no">340</span><span id="line-340"> }</span>
<span class="source-line-no">341</span><span id="line-341"></span>
<span class="source-line-no">342</span><span id="line-342"> protected void setMapperClass(Job job) {</span>
<span class="source-line-no">343</span><span id="line-343"> job.setMapperClass(LoadMapper.class);</span>
<span class="source-line-no">344</span><span id="line-344"> }</span>
<span class="source-line-no">345</span><span id="line-345"></span>
<span class="source-line-no">346</span><span id="line-346"> protected void doVerify(Configuration conf, TableDescriptor tableDescriptor) throws Exception {</span>
<span class="source-line-no">347</span><span id="line-347"> Path outputDir = getTestDir(TEST_NAME, "verify-output");</span>
<span class="source-line-no">348</span><span id="line-348"> LOG.info("Verify output dir: " + outputDir);</span>
<span class="source-line-no">349</span><span id="line-349"></span>
<span class="source-line-no">350</span><span id="line-350"> Job job = Job.getInstance(conf);</span>
<span class="source-line-no">351</span><span id="line-351"> job.setJarByClass(this.getClass());</span>
<span class="source-line-no">352</span><span id="line-352"> job.setJobName(TEST_NAME + " Verification for " + tableDescriptor.getTableName());</span>
<span class="source-line-no">353</span><span id="line-353"> setJobScannerConf(job);</span>
<span class="source-line-no">354</span><span id="line-354"></span>
<span class="source-line-no">355</span><span id="line-355"> Scan scan = new Scan();</span>
<span class="source-line-no">356</span><span id="line-356"></span>
<span class="source-line-no">357</span><span id="line-357"> TableMapReduceUtil.initTableMapperJob(tableDescriptor.getTableName().getNameAsString(), scan,</span>
<span class="source-line-no">358</span><span id="line-358"> VerifyMapper.class, BytesWritable.class, BytesWritable.class, job);</span>
<span class="source-line-no">359</span><span id="line-359"> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(), AbstractHBaseTool.class);</span>
<span class="source-line-no">360</span><span id="line-360"> int scannerCaching = conf.getInt("verify.scannercaching", SCANNER_CACHING);</span>
<span class="source-line-no">361</span><span id="line-361"> TableMapReduceUtil.setScannerCaching(job, scannerCaching);</span>
<span class="source-line-no">362</span><span id="line-362"></span>
<span class="source-line-no">363</span><span id="line-363"> job.setReducerClass(VerifyReducer.class);</span>
<span class="source-line-no">364</span><span id="line-364"> job.setNumReduceTasks(conf.getInt(NUM_REDUCE_TASKS_KEY, NUM_REDUCE_TASKS_DEFAULT));</span>
<span class="source-line-no">365</span><span id="line-365"> FileOutputFormat.setOutputPath(job, outputDir);</span>
<span class="source-line-no">366</span><span id="line-366"> assertTrue(job.waitForCompletion(true));</span>
<span class="source-line-no">367</span><span id="line-367"></span>
<span class="source-line-no">368</span><span id="line-368"> long numOutputRecords = job.getCounters().findCounter(Counters.ROWS_WRITTEN).getValue();</span>
<span class="source-line-no">369</span><span id="line-369"> assertEquals(0, numOutputRecords);</span>
<span class="source-line-no">370</span><span id="line-370"> }</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"> * Tool to search missing rows in WALs and hfiles. Pass in file or dir of keys to search for. Key</span>
<span class="source-line-no">374</span><span id="line-374"> * file must have been written by Verify step (we depend on the format it writes out. We'll read</span>
<span class="source-line-no">375</span><span id="line-375"> * them in and then search in hbase WALs and oldWALs dirs (Some of this is TODO).</span>
<span class="source-line-no">376</span><span id="line-376"> */</span>
<span class="source-line-no">377</span><span id="line-377"> public static class WALSearcher extends WALPlayer {</span>
<span class="source-line-no">378</span><span id="line-378"> public WALSearcher(Configuration conf) {</span>
<span class="source-line-no">379</span><span id="line-379"> super(conf);</span>
<span class="source-line-no">380</span><span id="line-380"> }</span>
<span class="source-line-no">381</span><span id="line-381"></span>
<span class="source-line-no">382</span><span id="line-382"> /**</span>
<span class="source-line-no">383</span><span id="line-383"> * The actual searcher mapper.</span>
<span class="source-line-no">384</span><span id="line-384"> */</span>
<span class="source-line-no">385</span><span id="line-385"> public static class WALMapperSearcher extends WALMapper {</span>
<span class="source-line-no">386</span><span id="line-386"> private SortedSet&lt;byte[]&gt; keysToFind;</span>
<span class="source-line-no">387</span><span id="line-387"> private AtomicInteger rows = new AtomicInteger(0);</span>
<span class="source-line-no">388</span><span id="line-388"></span>
<span class="source-line-no">389</span><span id="line-389"> @Override</span>
<span class="source-line-no">390</span><span id="line-390"> public void setup(Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Mutation&gt;.Context context)</span>
<span class="source-line-no">391</span><span id="line-391"> throws IOException {</span>
<span class="source-line-no">392</span><span id="line-392"> super.setup(context);</span>
<span class="source-line-no">393</span><span id="line-393"> try {</span>
<span class="source-line-no">394</span><span id="line-394"> this.keysToFind = readKeysToSearch(context.getConfiguration());</span>
<span class="source-line-no">395</span><span id="line-395"> LOG.info("Loaded keys to find: count=" + this.keysToFind.size());</span>
<span class="source-line-no">396</span><span id="line-396"> } catch (InterruptedException e) {</span>
<span class="source-line-no">397</span><span id="line-397"> throw new InterruptedIOException(e.toString());</span>
<span class="source-line-no">398</span><span id="line-398"> }</span>
<span class="source-line-no">399</span><span id="line-399"> }</span>
<span class="source-line-no">400</span><span id="line-400"></span>
<span class="source-line-no">401</span><span id="line-401"> @Override</span>
<span class="source-line-no">402</span><span id="line-402"> protected boolean filter(Context context, Cell cell) {</span>
<span class="source-line-no">403</span><span id="line-403"> // TODO: Can I do a better compare than this copying out key?</span>
<span class="source-line-no">404</span><span id="line-404"> byte[] row = new byte[cell.getRowLength()];</span>
<span class="source-line-no">405</span><span id="line-405"> System.arraycopy(cell.getRowArray(), cell.getRowOffset(), row, 0, cell.getRowLength());</span>
<span class="source-line-no">406</span><span id="line-406"> boolean b = this.keysToFind.contains(row);</span>
<span class="source-line-no">407</span><span id="line-407"> if (b) {</span>
<span class="source-line-no">408</span><span id="line-408"> String keyStr = Bytes.toStringBinary(row);</span>
<span class="source-line-no">409</span><span id="line-409"> try {</span>
<span class="source-line-no">410</span><span id="line-410"> LOG.info("Found cell=" + cell + " , walKey=" + context.getCurrentKey());</span>
<span class="source-line-no">411</span><span id="line-411"> } catch (IOException | InterruptedException e) {</span>
<span class="source-line-no">412</span><span id="line-412"> LOG.warn(e.toString(), e);</span>
<span class="source-line-no">413</span><span id="line-413"> }</span>
<span class="source-line-no">414</span><span id="line-414"> if (rows.addAndGet(1) &lt; MISSING_ROWS_TO_LOG) {</span>
<span class="source-line-no">415</span><span id="line-415"> context.getCounter(FOUND_GROUP_KEY, keyStr).increment(1);</span>
<span class="source-line-no">416</span><span id="line-416"> }</span>
<span class="source-line-no">417</span><span id="line-417"> context.getCounter(FOUND_GROUP_KEY, "CELL_WITH_MISSING_ROW").increment(1);</span>
<span class="source-line-no">418</span><span id="line-418"> }</span>
<span class="source-line-no">419</span><span id="line-419"> return b;</span>
<span class="source-line-no">420</span><span id="line-420"> }</span>
<span class="source-line-no">421</span><span id="line-421"> }</span>
<span class="source-line-no">422</span><span id="line-422"></span>
<span class="source-line-no">423</span><span id="line-423"> // Put in place the above WALMapperSearcher.</span>
<span class="source-line-no">424</span><span id="line-424"> @Override</span>
<span class="source-line-no">425</span><span id="line-425"> public Job createSubmittableJob(String[] args) throws IOException {</span>
<span class="source-line-no">426</span><span id="line-426"> Job job = super.createSubmittableJob(args);</span>
<span class="source-line-no">427</span><span id="line-427"> // Call my class instead.</span>
<span class="source-line-no">428</span><span id="line-428"> job.setJarByClass(WALMapperSearcher.class);</span>
<span class="source-line-no">429</span><span id="line-429"> job.setMapperClass(WALMapperSearcher.class);</span>
<span class="source-line-no">430</span><span id="line-430"> job.setOutputFormatClass(NullOutputFormat.class);</span>
<span class="source-line-no">431</span><span id="line-431"> return job;</span>
<span class="source-line-no">432</span><span id="line-432"> }</span>
<span class="source-line-no">433</span><span id="line-433"> }</span>
<span class="source-line-no">434</span><span id="line-434"></span>
<span class="source-line-no">435</span><span id="line-435"> static final String FOUND_GROUP_KEY = "Found";</span>
<span class="source-line-no">436</span><span id="line-436"> static final String SEARCHER_INPUTDIR_KEY = "searcher.keys.inputdir";</span>
<span class="source-line-no">437</span><span id="line-437"></span>
<span class="source-line-no">438</span><span id="line-438"> static SortedSet&lt;byte[]&gt; readKeysToSearch(final Configuration conf)</span>
<span class="source-line-no">439</span><span id="line-439"> throws IOException, InterruptedException {</span>
<span class="source-line-no">440</span><span id="line-440"> Path keysInputDir = new Path(conf.get(SEARCHER_INPUTDIR_KEY));</span>
<span class="source-line-no">441</span><span id="line-441"> FileSystem fs = FileSystem.get(conf);</span>
<span class="source-line-no">442</span><span id="line-442"> SortedSet&lt;byte[]&gt; result = new TreeSet&lt;&gt;(Bytes.BYTES_COMPARATOR);</span>
<span class="source-line-no">443</span><span id="line-443"> if (!fs.exists(keysInputDir)) {</span>
<span class="source-line-no">444</span><span id="line-444"> throw new FileNotFoundException(keysInputDir.toString());</span>
<span class="source-line-no">445</span><span id="line-445"> }</span>
<span class="source-line-no">446</span><span id="line-446"> if (!fs.isDirectory(keysInputDir)) {</span>
<span class="source-line-no">447</span><span id="line-447"> FileStatus keyFileStatus = fs.getFileStatus(keysInputDir);</span>
<span class="source-line-no">448</span><span id="line-448"> readFileToSearch(fs, keyFileStatus, result);</span>
<span class="source-line-no">449</span><span id="line-449"> } else {</span>
<span class="source-line-no">450</span><span id="line-450"> RemoteIterator&lt;LocatedFileStatus&gt; iterator = fs.listFiles(keysInputDir, false);</span>
<span class="source-line-no">451</span><span id="line-451"> while (iterator.hasNext()) {</span>
<span class="source-line-no">452</span><span id="line-452"> LocatedFileStatus keyFileStatus = iterator.next();</span>
<span class="source-line-no">453</span><span id="line-453"> // Skip "_SUCCESS" file.</span>
<span class="source-line-no">454</span><span id="line-454"> if (keyFileStatus.getPath().getName().startsWith("_")) continue;</span>
<span class="source-line-no">455</span><span id="line-455"> readFileToSearch(fs, keyFileStatus, result);</span>
<span class="source-line-no">456</span><span id="line-456"> }</span>
<span class="source-line-no">457</span><span id="line-457"> }</span>
<span class="source-line-no">458</span><span id="line-458"> return result;</span>
<span class="source-line-no">459</span><span id="line-459"> }</span>
<span class="source-line-no">460</span><span id="line-460"></span>
<span class="source-line-no">461</span><span id="line-461"> private static SortedSet&lt;byte[]&gt; readFileToSearch(final FileSystem fs,</span>
<span class="source-line-no">462</span><span id="line-462"> final FileStatus keyFileStatus, SortedSet&lt;byte[]&gt; result)</span>
<span class="source-line-no">463</span><span id="line-463"> throws IOException, InterruptedException {</span>
<span class="source-line-no">464</span><span id="line-464"> // verify uses file output format and writes &lt;Text, Text&gt;. We can read it as a text file</span>
<span class="source-line-no">465</span><span id="line-465"> try (InputStream in = fs.open(keyFileStatus.getPath()); BufferedReader reader =</span>
<span class="source-line-no">466</span><span id="line-466"> new BufferedReader(new InputStreamReader(in, StandardCharsets.UTF_8))) {</span>
<span class="source-line-no">467</span><span id="line-467"> // extract out the key and return that missing as a missing key</span>
<span class="source-line-no">468</span><span id="line-468"> String line;</span>
<span class="source-line-no">469</span><span id="line-469"> while ((line = reader.readLine()) != null) {</span>
<span class="source-line-no">470</span><span id="line-470"> if (line.isEmpty()) continue;</span>
<span class="source-line-no">471</span><span id="line-471"> List&lt;String&gt; parts = Splitter.onPattern("\\s+").splitToList(line);</span>
<span class="source-line-no">472</span><span id="line-472"> if (parts.size() &gt;= 1) {</span>
<span class="source-line-no">473</span><span id="line-473"> result.add(Bytes.toBytesBinary(Iterables.get(parts, 0)));</span>
<span class="source-line-no">474</span><span id="line-474"> } else {</span>
<span class="source-line-no">475</span><span id="line-475"> LOG.info("Cannot parse key from: " + line);</span>
<span class="source-line-no">476</span><span id="line-476"> }</span>
<span class="source-line-no">477</span><span id="line-477"> }</span>
<span class="source-line-no">478</span><span id="line-478"> }</span>
<span class="source-line-no">479</span><span id="line-479"> return result;</span>
<span class="source-line-no">480</span><span id="line-480"> }</span>
<span class="source-line-no">481</span><span id="line-481"></span>
<span class="source-line-no">482</span><span id="line-482"> private int doSearch(String keysDir) throws Exception {</span>
<span class="source-line-no">483</span><span id="line-483"> Path inputDir = new Path(keysDir);</span>
<span class="source-line-no">484</span><span id="line-484"></span>
<span class="source-line-no">485</span><span id="line-485"> getConf().set(SEARCHER_INPUTDIR_KEY, inputDir.toString());</span>
<span class="source-line-no">486</span><span id="line-486"> SortedSet&lt;byte[]&gt; keys = readKeysToSearch(getConf());</span>
<span class="source-line-no">487</span><span id="line-487"> if (keys.isEmpty()) throw new RuntimeException("No keys to find");</span>
<span class="source-line-no">488</span><span id="line-488"> LOG.info("Count of keys to find: " + keys.size());</span>
<span class="source-line-no">489</span><span id="line-489"> for (byte[] key : keys)</span>
<span class="source-line-no">490</span><span id="line-490"> LOG.info("Key: " + Bytes.toStringBinary(key));</span>
<span class="source-line-no">491</span><span id="line-491"> // Now read all WALs. In two dirs. Presumes certain layout.</span>
<span class="source-line-no">492</span><span id="line-492"> Path walsDir = new Path(CommonFSUtils.getWALRootDir(getConf()), HConstants.HREGION_LOGDIR_NAME);</span>
<span class="source-line-no">493</span><span id="line-493"> Path oldWalsDir =</span>
<span class="source-line-no">494</span><span id="line-494"> new Path(CommonFSUtils.getWALRootDir(getConf()), HConstants.HREGION_OLDLOGDIR_NAME);</span>
<span class="source-line-no">495</span><span id="line-495"> LOG.info("Running Search with keys inputDir=" + inputDir + " against "</span>
<span class="source-line-no">496</span><span id="line-496"> + getConf().get(HConstants.HBASE_DIR));</span>
<span class="source-line-no">497</span><span id="line-497"> int ret = ToolRunner.run(new WALSearcher(getConf()), new String[] { walsDir.toString(), "" });</span>
<span class="source-line-no">498</span><span id="line-498"> if (ret != 0) return ret;</span>
<span class="source-line-no">499</span><span id="line-499"> return ToolRunner.run(new WALSearcher(getConf()), new String[] { oldWalsDir.toString(), "" });</span>
<span class="source-line-no">500</span><span id="line-500"> }</span>
<span class="source-line-no">501</span><span id="line-501"></span>
<span class="source-line-no">502</span><span id="line-502"> private static void setJobScannerConf(Job job) {</span>
<span class="source-line-no">503</span><span id="line-503"> long lpr = job.getConfiguration().getLong(NUM_TO_WRITE_KEY, NUM_TO_WRITE_DEFAULT) / 100;</span>
<span class="source-line-no">504</span><span id="line-504"> job.getConfiguration().setInt(TableRecordReaderImpl.LOG_PER_ROW_COUNT, (int) lpr);</span>
<span class="source-line-no">505</span><span id="line-505"> }</span>
<span class="source-line-no">506</span><span id="line-506"></span>
<span class="source-line-no">507</span><span id="line-507"> public Path getTestDir(String testName, String subdir) throws IOException {</span>
<span class="source-line-no">508</span><span id="line-508"> Path testDir = util.getDataTestDirOnTestFS(testName);</span>
<span class="source-line-no">509</span><span id="line-509"> FileSystem fs = FileSystem.get(getConf());</span>
<span class="source-line-no">510</span><span id="line-510"> fs.deleteOnExit(testDir);</span>
<span class="source-line-no">511</span><span id="line-511"></span>
<span class="source-line-no">512</span><span id="line-512"> return new Path(new Path(testDir, testName), subdir);</span>
<span class="source-line-no">513</span><span id="line-513"> }</span>
<span class="source-line-no">514</span><span id="line-514"></span>
<span class="source-line-no">515</span><span id="line-515"> @Test</span>
<span class="source-line-no">516</span><span id="line-516"> public void testLoadAndVerify() throws Exception {</span>
<span class="source-line-no">517</span><span id="line-517"> TableDescriptor tableDescriptor =</span>
<span class="source-line-no">518</span><span id="line-518"> TableDescriptorBuilder.newBuilder(TableName.valueOf(TEST_NAME))</span>
<span class="source-line-no">519</span><span id="line-519"> .setColumnFamily(ColumnFamilyDescriptorBuilder.of(TEST_FAMILY)).build();</span>
<span class="source-line-no">520</span><span id="line-520"></span>
<span class="source-line-no">521</span><span id="line-521"> Admin admin = getTestingUtil(getConf()).getAdmin();</span>
<span class="source-line-no">522</span><span id="line-522"> admin.createTable(tableDescriptor, Bytes.toBytes(0L), Bytes.toBytes(-1L), 40);</span>
<span class="source-line-no">523</span><span id="line-523"></span>
<span class="source-line-no">524</span><span id="line-524"> doLoad(getConf(), tableDescriptor);</span>
<span class="source-line-no">525</span><span id="line-525"> doVerify(getConf(), tableDescriptor);</span>
<span class="source-line-no">526</span><span id="line-526"></span>
<span class="source-line-no">527</span><span id="line-527"> // Only disable and drop if we succeeded to verify - otherwise it's useful</span>
<span class="source-line-no">528</span><span id="line-528"> // to leave it around for post-mortem</span>
<span class="source-line-no">529</span><span id="line-529"> getTestingUtil(getConf()).deleteTable(tableDescriptor.getTableName());</span>
<span class="source-line-no">530</span><span id="line-530"> }</span>
<span class="source-line-no">531</span><span id="line-531"></span>
<span class="source-line-no">532</span><span id="line-532"> @Override</span>
<span class="source-line-no">533</span><span id="line-533"> public void printUsage() {</span>
<span class="source-line-no">534</span><span id="line-534"> printUsage(this.getClass().getSimpleName() + " &lt;options&gt;"</span>
<span class="source-line-no">535</span><span id="line-535"> + " [-Doptions] &lt;load|verify|loadAndVerify|search&gt;", "Options", "");</span>
<span class="source-line-no">536</span><span id="line-536"> System.err.println("");</span>
<span class="source-line-no">537</span><span id="line-537"> System.err.println(" Loads a table with row dependencies and verifies the dependency chains");</span>
<span class="source-line-no">538</span><span id="line-538"> System.err.println("Options");</span>
<span class="source-line-no">539</span><span id="line-539"> System.err</span>
<span class="source-line-no">540</span><span id="line-540"> .println(" -Dloadmapper.table=&lt;name&gt; Table to write/verify (default autogen)");</span>
<span class="source-line-no">541</span><span id="line-541"> System.err</span>
<span class="source-line-no">542</span><span id="line-542"> .println(" -Dloadmapper.backrefs=&lt;n&gt; Number of backreferences per row (default 50)");</span>
<span class="source-line-no">543</span><span id="line-543"> System.err.println(</span>
<span class="source-line-no">544</span><span id="line-544"> " -Dloadmapper.num_to_write=&lt;n&gt; Number of rows per mapper (default 100,000 per mapper)");</span>
<span class="source-line-no">545</span><span id="line-545"> System.err.println(</span>
<span class="source-line-no">546</span><span id="line-546"> " -Dloadmapper.deleteAfter=&lt;bool&gt; Delete after a successful verify (default true)");</span>
<span class="source-line-no">547</span><span id="line-547"> System.err.println(</span>
<span class="source-line-no">548</span><span id="line-548"> " -Dloadmapper.numPresplits=&lt;n&gt; Number of presplit regions to start with (default 40)");</span>
<span class="source-line-no">549</span><span id="line-549"> System.err</span>
<span class="source-line-no">550</span><span id="line-550"> .println(" -Dloadmapper.map.tasks=&lt;n&gt; Number of map tasks for load (default 200)");</span>
<span class="source-line-no">551</span><span id="line-551"> System.err</span>
<span class="source-line-no">552</span><span id="line-552"> .println(" -Dverify.reduce.tasks=&lt;n&gt; Number of reduce tasks for verify (default 35)");</span>
<span class="source-line-no">553</span><span id="line-553"> System.err.println(</span>
<span class="source-line-no">554</span><span id="line-554"> " -Dverify.scannercaching=&lt;n&gt; Number hbase scanner caching rows to read (default 50)");</span>
<span class="source-line-no">555</span><span id="line-555"> }</span>
<span class="source-line-no">556</span><span id="line-556"></span>
<span class="source-line-no">557</span><span id="line-557"> @Override</span>
<span class="source-line-no">558</span><span id="line-558"> protected void processOptions(CommandLine cmd) {</span>
<span class="source-line-no">559</span><span id="line-559"> super.processOptions(cmd);</span>
<span class="source-line-no">560</span><span id="line-560"></span>
<span class="source-line-no">561</span><span id="line-561"> String[] args = cmd.getArgs();</span>
<span class="source-line-no">562</span><span id="line-562"> if (args == null || args.length &lt; 1) {</span>
<span class="source-line-no">563</span><span id="line-563"> printUsage();</span>
<span class="source-line-no">564</span><span id="line-564"> throw new RuntimeException("Incorrect Number of args.");</span>
<span class="source-line-no">565</span><span id="line-565"> }</span>
<span class="source-line-no">566</span><span id="line-566"> toRun = args[0];</span>
<span class="source-line-no">567</span><span id="line-567"> if (toRun.equalsIgnoreCase("search")) {</span>
<span class="source-line-no">568</span><span id="line-568"> if (args.length &gt; 1) {</span>
<span class="source-line-no">569</span><span id="line-569"> keysDir = args[1];</span>
<span class="source-line-no">570</span><span id="line-570"> }</span>
<span class="source-line-no">571</span><span id="line-571"> }</span>
<span class="source-line-no">572</span><span id="line-572"> }</span>
<span class="source-line-no">573</span><span id="line-573"></span>
<span class="source-line-no">574</span><span id="line-574"> @Override</span>
<span class="source-line-no">575</span><span id="line-575"> public int runTestFromCommandLine() throws Exception {</span>
<span class="source-line-no">576</span><span id="line-576"> IntegrationTestingUtility.setUseDistributedCluster(getConf());</span>
<span class="source-line-no">577</span><span id="line-577"> boolean doLoad = false;</span>
<span class="source-line-no">578</span><span id="line-578"> boolean doVerify = false;</span>
<span class="source-line-no">579</span><span id="line-579"> boolean doSearch = false;</span>
<span class="source-line-no">580</span><span id="line-580"> boolean doDelete = getConf().getBoolean("loadmapper.deleteAfter", true);</span>
<span class="source-line-no">581</span><span id="line-581"> int numPresplits = getConf().getInt("loadmapper.numPresplits", 40);</span>
<span class="source-line-no">582</span><span id="line-582"></span>
<span class="source-line-no">583</span><span id="line-583"> if (toRun.equalsIgnoreCase("load")) {</span>
<span class="source-line-no">584</span><span id="line-584"> doLoad = true;</span>
<span class="source-line-no">585</span><span id="line-585"> } else if (toRun.equalsIgnoreCase("verify")) {</span>
<span class="source-line-no">586</span><span id="line-586"> doVerify = true;</span>
<span class="source-line-no">587</span><span id="line-587"> } else if (toRun.equalsIgnoreCase("loadAndVerify")) {</span>
<span class="source-line-no">588</span><span id="line-588"> doLoad = true;</span>
<span class="source-line-no">589</span><span id="line-589"> doVerify = true;</span>
<span class="source-line-no">590</span><span id="line-590"> } else if (toRun.equalsIgnoreCase("search")) {</span>
<span class="source-line-no">591</span><span id="line-591"> doLoad = false;</span>
<span class="source-line-no">592</span><span id="line-592"> doVerify = false;</span>
<span class="source-line-no">593</span><span id="line-593"> doSearch = true;</span>
<span class="source-line-no">594</span><span id="line-594"> if (keysDir == null) {</span>
<span class="source-line-no">595</span><span id="line-595"> System.err.println("Usage: search &lt;KEYS_DIR&gt;]");</span>
<span class="source-line-no">596</span><span id="line-596"> return 1;</span>
<span class="source-line-no">597</span><span id="line-597"> }</span>
<span class="source-line-no">598</span><span id="line-598"> } else {</span>
<span class="source-line-no">599</span><span id="line-599"> System.err.println("Invalid argument " + toRun);</span>
<span class="source-line-no">600</span><span id="line-600"> printUsage();</span>
<span class="source-line-no">601</span><span id="line-601"> return 1;</span>
<span class="source-line-no">602</span><span id="line-602"> }</span>
<span class="source-line-no">603</span><span id="line-603"></span>
<span class="source-line-no">604</span><span id="line-604"> // create HTableDescriptor for specified table</span>
<span class="source-line-no">605</span><span id="line-605"> TableName table = getTablename();</span>
<span class="source-line-no">606</span><span id="line-606"> TableDescriptor tableDescriptor = TableDescriptorBuilder.newBuilder(table)</span>
<span class="source-line-no">607</span><span id="line-607"> .setColumnFamily(ColumnFamilyDescriptorBuilder.of(TEST_FAMILY)).build();</span>
<span class="source-line-no">608</span><span id="line-608"></span>
<span class="source-line-no">609</span><span id="line-609"> if (doLoad) {</span>
<span class="source-line-no">610</span><span id="line-610"> try (Connection conn = ConnectionFactory.createConnection(getConf());</span>
<span class="source-line-no">611</span><span id="line-611"> Admin admin = conn.getAdmin()) {</span>
<span class="source-line-no">612</span><span id="line-612"> admin.createTable(tableDescriptor, Bytes.toBytes(0L), Bytes.toBytes(-1L), numPresplits);</span>
<span class="source-line-no">613</span><span id="line-613"> doLoad(getConf(), tableDescriptor);</span>
<span class="source-line-no">614</span><span id="line-614"> }</span>
<span class="source-line-no">615</span><span id="line-615"> }</span>
<span class="source-line-no">616</span><span id="line-616"> if (doVerify) {</span>
<span class="source-line-no">617</span><span id="line-617"> doVerify(getConf(), tableDescriptor);</span>
<span class="source-line-no">618</span><span id="line-618"> if (doDelete) {</span>
<span class="source-line-no">619</span><span id="line-619"> getTestingUtil(getConf()).deleteTable(tableDescriptor.getTableName());</span>
<span class="source-line-no">620</span><span id="line-620"> }</span>
<span class="source-line-no">621</span><span id="line-621"> }</span>
<span class="source-line-no">622</span><span id="line-622"> if (doSearch) {</span>
<span class="source-line-no">623</span><span id="line-623"> return doSearch(keysDir);</span>
<span class="source-line-no">624</span><span id="line-624"> }</span>
<span class="source-line-no">625</span><span id="line-625"> return 0;</span>
<span class="source-line-no">626</span><span id="line-626"> }</span>
<span class="source-line-no">627</span><span id="line-627"></span>
<span class="source-line-no">628</span><span id="line-628"> @Override</span>
<span class="source-line-no">629</span><span id="line-629"> public TableName getTablename() {</span>
<span class="source-line-no">630</span><span id="line-630"> return TableName.valueOf(getConf().get(TABLE_NAME_KEY, TEST_NAME));</span>
<span class="source-line-no">631</span><span id="line-631"> }</span>
<span class="source-line-no">632</span><span id="line-632"></span>
<span class="source-line-no">633</span><span id="line-633"> @Override</span>
<span class="source-line-no">634</span><span id="line-634"> protected Set&lt;String&gt; getColumnFamilies() {</span>
<span class="source-line-no">635</span><span id="line-635"> return Sets.newHashSet(Bytes.toString(TEST_FAMILY));</span>
<span class="source-line-no">636</span><span id="line-636"> }</span>
<span class="source-line-no">637</span><span id="line-637"></span>
<span class="source-line-no">638</span><span id="line-638"> public static void main(String argv[]) throws Exception {</span>
<span class="source-line-no">639</span><span id="line-639"> Configuration conf = HBaseConfiguration.create();</span>
<span class="source-line-no">640</span><span id="line-640"> IntegrationTestingUtility.setUseDistributedCluster(conf);</span>
<span class="source-line-no">641</span><span id="line-641"> int ret = ToolRunner.run(conf, new IntegrationTestLoadAndVerify(), argv);</span>
<span class="source-line-no">642</span><span id="line-642"> System.exit(ret);</span>
<span class="source-line-no">643</span><span id="line-643"> }</span>
<span class="source-line-no">644</span><span id="line-644">}</span>
</pre>
</div>
</main>
</body>
</html>