blob: c769c0ae7c00f656b799fa06b281a3f2a35ce1ac [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: WALPlayer, enum: Counter">
<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 java.io.IOException;</span>
<span class="source-line-no">021</span><span id="line-21">import java.text.ParseException;</span>
<span class="source-line-no">022</span><span id="line-22">import java.text.SimpleDateFormat;</span>
<span class="source-line-no">023</span><span id="line-23">import java.util.ArrayList;</span>
<span class="source-line-no">024</span><span id="line-24">import java.util.Collections;</span>
<span class="source-line-no">025</span><span id="line-25">import java.util.HashSet;</span>
<span class="source-line-no">026</span><span id="line-26">import java.util.List;</span>
<span class="source-line-no">027</span><span id="line-27">import java.util.Map;</span>
<span class="source-line-no">028</span><span id="line-28">import java.util.Set;</span>
<span class="source-line-no">029</span><span id="line-29">import java.util.TreeMap;</span>
<span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.conf.Configuration;</span>
<span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.conf.Configured;</span>
<span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.fs.Path;</span>
<span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.Cell;</span>
<span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.CellUtil;</span>
<span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.HBaseConfiguration;</span>
<span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.PrivateCellUtil;</span>
<span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.TableName;</span>
<span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.client.Connection;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.hbase.client.ConnectionFactory;</span>
<span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.hbase.client.Delete;</span>
<span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.hbase.client.Mutation;</span>
<span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.hbase.client.Put;</span>
<span class="source-line-no">043</span><span id="line-43">import org.apache.hadoop.hbase.client.RegionLocator;</span>
<span class="source-line-no">044</span><span id="line-44">import org.apache.hadoop.hbase.client.Table;</span>
<span class="source-line-no">045</span><span id="line-45">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span>
<span class="source-line-no">046</span><span id="line-46">import org.apache.hadoop.hbase.mapreduce.HFileOutputFormat2.TableInfo;</span>
<span class="source-line-no">047</span><span id="line-47">import org.apache.hadoop.hbase.regionserver.wal.WALCellCodec;</span>
<span class="source-line-no">048</span><span id="line-48">import org.apache.hadoop.hbase.util.Bytes;</span>
<span class="source-line-no">049</span><span id="line-49">import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;</span>
<span class="source-line-no">050</span><span id="line-50">import org.apache.hadoop.hbase.util.MapReduceExtendedCell;</span>
<span class="source-line-no">051</span><span id="line-51">import org.apache.hadoop.hbase.wal.WALEdit;</span>
<span class="source-line-no">052</span><span id="line-52">import org.apache.hadoop.hbase.wal.WALKey;</span>
<span class="source-line-no">053</span><span id="line-53">import org.apache.hadoop.mapreduce.Job;</span>
<span class="source-line-no">054</span><span id="line-54">import org.apache.hadoop.mapreduce.Mapper;</span>
<span class="source-line-no">055</span><span id="line-55">import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;</span>
<span class="source-line-no">056</span><span id="line-56">import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;</span>
<span class="source-line-no">057</span><span id="line-57">import org.apache.hadoop.util.Tool;</span>
<span class="source-line-no">058</span><span id="line-58">import org.apache.hadoop.util.ToolRunner;</span>
<span class="source-line-no">059</span><span id="line-59">import org.apache.yetus.audience.InterfaceAudience;</span>
<span class="source-line-no">060</span><span id="line-60">import org.slf4j.Logger;</span>
<span class="source-line-no">061</span><span id="line-61">import org.slf4j.LoggerFactory;</span>
<span class="source-line-no">062</span><span id="line-62"></span>
<span class="source-line-no">063</span><span id="line-63">/**</span>
<span class="source-line-no">064</span><span id="line-64"> * A tool to replay WAL files as a M/R job. The WAL can be replayed for a set of tables or all</span>
<span class="source-line-no">065</span><span id="line-65"> * tables, and a time range can be provided (in milliseconds). The WAL is filtered to the passed set</span>
<span class="source-line-no">066</span><span id="line-66"> * of tables and the output can optionally be mapped to another set of tables. WAL replay can also</span>
<span class="source-line-no">067</span><span id="line-67"> * generate HFiles for later bulk importing, in that case the WAL is replayed for a single table</span>
<span class="source-line-no">068</span><span id="line-68"> * only.</span>
<span class="source-line-no">069</span><span id="line-69"> */</span>
<span class="source-line-no">070</span><span id="line-70">@InterfaceAudience.Public</span>
<span class="source-line-no">071</span><span id="line-71">public class WALPlayer extends Configured implements Tool {</span>
<span class="source-line-no">072</span><span id="line-72"> private static final Logger LOG = LoggerFactory.getLogger(WALPlayer.class);</span>
<span class="source-line-no">073</span><span id="line-73"> final static String NAME = "WALPlayer";</span>
<span class="source-line-no">074</span><span id="line-74"> public final static String BULK_OUTPUT_CONF_KEY = "wal.bulk.output";</span>
<span class="source-line-no">075</span><span id="line-75"> public final static String TABLES_KEY = "wal.input.tables";</span>
<span class="source-line-no">076</span><span id="line-76"> public final static String TABLE_MAP_KEY = "wal.input.tablesmap";</span>
<span class="source-line-no">077</span><span id="line-77"> public final static String INPUT_FILES_SEPARATOR_KEY = "wal.input.separator";</span>
<span class="source-line-no">078</span><span id="line-78"> public final static String IGNORE_MISSING_FILES = "wal.input.ignore.missing.files";</span>
<span class="source-line-no">079</span><span id="line-79"> public final static String MULTI_TABLES_SUPPORT = "wal.multi.tables.support";</span>
<span class="source-line-no">080</span><span id="line-80"></span>
<span class="source-line-no">081</span><span id="line-81"> protected static final String tableSeparator = ";";</span>
<span class="source-line-no">082</span><span id="line-82"></span>
<span class="source-line-no">083</span><span id="line-83"> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";</span>
<span class="source-line-no">084</span><span id="line-84"></span>
<span class="source-line-no">085</span><span id="line-85"> public WALPlayer() {</span>
<span class="source-line-no">086</span><span id="line-86"> }</span>
<span class="source-line-no">087</span><span id="line-87"></span>
<span class="source-line-no">088</span><span id="line-88"> protected WALPlayer(final Configuration c) {</span>
<span class="source-line-no">089</span><span id="line-89"> super(c);</span>
<span class="source-line-no">090</span><span id="line-90"> }</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"> * A mapper that just writes out KeyValues. This one can be used together with</span>
<span class="source-line-no">094</span><span id="line-94"> * {@link CellSortReducer}</span>
<span class="source-line-no">095</span><span id="line-95"> */</span>
<span class="source-line-no">096</span><span id="line-96"> static class WALKeyValueMapper extends Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Cell&gt; {</span>
<span class="source-line-no">097</span><span id="line-97"> private Set&lt;String&gt; tableSet = new HashSet&lt;String&gt;();</span>
<span class="source-line-no">098</span><span id="line-98"> private boolean multiTableSupport = false;</span>
<span class="source-line-no">099</span><span id="line-99"></span>
<span class="source-line-no">100</span><span id="line-100"> @Override</span>
<span class="source-line-no">101</span><span id="line-101"> public void map(WALKey key, WALEdit value, Context context) throws IOException {</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"> // skip all other tables</span>
<span class="source-line-no">104</span><span id="line-104"> TableName table = key.getTableName();</span>
<span class="source-line-no">105</span><span id="line-105"> if (tableSet.contains(table.getNameAsString())) {</span>
<span class="source-line-no">106</span><span id="line-106"> for (Cell cell : value.getCells()) {</span>
<span class="source-line-no">107</span><span id="line-107"> if (WALEdit.isMetaEditFamily(cell)) {</span>
<span class="source-line-no">108</span><span id="line-108"> continue;</span>
<span class="source-line-no">109</span><span id="line-109"> }</span>
<span class="source-line-no">110</span><span id="line-110"></span>
<span class="source-line-no">111</span><span id="line-111"> // Set sequenceId from WALKey, since it is not included by WALCellCodec. The sequenceId</span>
<span class="source-line-no">112</span><span id="line-112"> // on WALKey is the same value that was on the cells in the WALEdit. This enables</span>
<span class="source-line-no">113</span><span id="line-113"> // CellSortReducer to use sequenceId to disambiguate duplicate cell timestamps.</span>
<span class="source-line-no">114</span><span id="line-114"> // See HBASE-27649</span>
<span class="source-line-no">115</span><span id="line-115"> PrivateCellUtil.setSequenceId(cell, key.getSequenceId());</span>
<span class="source-line-no">116</span><span id="line-116"></span>
<span class="source-line-no">117</span><span id="line-117"> byte[] outKey = multiTableSupport</span>
<span class="source-line-no">118</span><span id="line-118"> ? Bytes.add(table.getName(), Bytes.toBytes(tableSeparator), CellUtil.cloneRow(cell))</span>
<span class="source-line-no">119</span><span id="line-119"> : CellUtil.cloneRow(cell);</span>
<span class="source-line-no">120</span><span id="line-120"> context.write(new ImmutableBytesWritable(outKey),</span>
<span class="source-line-no">121</span><span id="line-121"> new MapReduceExtendedCell(PrivateCellUtil.ensureExtendedCell(cell)));</span>
<span class="source-line-no">122</span><span id="line-122"> }</span>
<span class="source-line-no">123</span><span id="line-123"> }</span>
<span class="source-line-no">124</span><span id="line-124"> } catch (InterruptedException e) {</span>
<span class="source-line-no">125</span><span id="line-125"> LOG.error("Interrupted while emitting Cell", e);</span>
<span class="source-line-no">126</span><span id="line-126"> Thread.currentThread().interrupt();</span>
<span class="source-line-no">127</span><span id="line-127"> }</span>
<span class="source-line-no">128</span><span id="line-128"> }</span>
<span class="source-line-no">129</span><span id="line-129"></span>
<span class="source-line-no">130</span><span id="line-130"> @Override</span>
<span class="source-line-no">131</span><span id="line-131"> public void setup(Context context) throws IOException {</span>
<span class="source-line-no">132</span><span id="line-132"> Configuration conf = context.getConfiguration();</span>
<span class="source-line-no">133</span><span id="line-133"> String[] tables = conf.getStrings(TABLES_KEY);</span>
<span class="source-line-no">134</span><span id="line-134"> this.multiTableSupport = conf.getBoolean(MULTI_TABLES_SUPPORT, false);</span>
<span class="source-line-no">135</span><span id="line-135"> Collections.addAll(tableSet, tables);</span>
<span class="source-line-no">136</span><span id="line-136"> }</span>
<span class="source-line-no">137</span><span id="line-137"> }</span>
<span class="source-line-no">138</span><span id="line-138"></span>
<span class="source-line-no">139</span><span id="line-139"> /**</span>
<span class="source-line-no">140</span><span id="line-140"> * Enum for map metrics. Keep it out here rather than inside in the Map inner-class so we can find</span>
<span class="source-line-no">141</span><span id="line-141"> * associated properties.</span>
<span class="source-line-no">142</span><span id="line-142"> */</span>
<span class="source-line-no">143</span><span id="line-143"> protected static enum Counter {</span>
<span class="source-line-no">144</span><span id="line-144"> /** Number of aggregated writes */</span>
<span class="source-line-no">145</span><span id="line-145"> PUTS,</span>
<span class="source-line-no">146</span><span id="line-146"> /** Number of aggregated deletes */</span>
<span class="source-line-no">147</span><span id="line-147"> DELETES,</span>
<span class="source-line-no">148</span><span id="line-148"> CELLS_READ,</span>
<span class="source-line-no">149</span><span id="line-149"> CELLS_WRITTEN,</span>
<span class="source-line-no">150</span><span id="line-150"> WALEDITS</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"> * A mapper that writes out {@link Mutation} to be directly applied to a running HBase instance.</span>
<span class="source-line-no">155</span><span id="line-155"> */</span>
<span class="source-line-no">156</span><span id="line-156"> protected static class WALMapper</span>
<span class="source-line-no">157</span><span id="line-157"> extends Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Mutation&gt; {</span>
<span class="source-line-no">158</span><span id="line-158"> private Map&lt;TableName, TableName&gt; tables = new TreeMap&lt;&gt;();</span>
<span class="source-line-no">159</span><span id="line-159"></span>
<span class="source-line-no">160</span><span id="line-160"> @Override</span>
<span class="source-line-no">161</span><span id="line-161"> public void map(WALKey key, WALEdit value, Context context) throws IOException {</span>
<span class="source-line-no">162</span><span id="line-162"> context.getCounter(Counter.WALEDITS).increment(1);</span>
<span class="source-line-no">163</span><span id="line-163"> try {</span>
<span class="source-line-no">164</span><span id="line-164"> if (tables.isEmpty() || tables.containsKey(key.getTableName())) {</span>
<span class="source-line-no">165</span><span id="line-165"> TableName targetTable =</span>
<span class="source-line-no">166</span><span id="line-166"> tables.isEmpty() ? key.getTableName() : tables.get(key.getTableName());</span>
<span class="source-line-no">167</span><span id="line-167"> ImmutableBytesWritable tableOut = new ImmutableBytesWritable(targetTable.getName());</span>
<span class="source-line-no">168</span><span id="line-168"> Put put = null;</span>
<span class="source-line-no">169</span><span id="line-169"> Delete del = null;</span>
<span class="source-line-no">170</span><span id="line-170"> Cell lastCell = null;</span>
<span class="source-line-no">171</span><span id="line-171"> for (Cell cell : value.getCells()) {</span>
<span class="source-line-no">172</span><span id="line-172"> context.getCounter(Counter.CELLS_READ).increment(1);</span>
<span class="source-line-no">173</span><span id="line-173"> // Filtering WAL meta marker entries.</span>
<span class="source-line-no">174</span><span id="line-174"> if (WALEdit.isMetaEditFamily(cell)) {</span>
<span class="source-line-no">175</span><span id="line-175"> continue;</span>
<span class="source-line-no">176</span><span id="line-176"> }</span>
<span class="source-line-no">177</span><span id="line-177"> // Allow a subclass filter out this cell.</span>
<span class="source-line-no">178</span><span id="line-178"> if (filter(context, cell)) {</span>
<span class="source-line-no">179</span><span id="line-179"> // A WALEdit may contain multiple operations (HBASE-3584) and/or</span>
<span class="source-line-no">180</span><span id="line-180"> // multiple rows (HBASE-5229).</span>
<span class="source-line-no">181</span><span id="line-181"> // Aggregate as much as possible into a single Put/Delete</span>
<span class="source-line-no">182</span><span id="line-182"> // operation before writing to the context.</span>
<span class="source-line-no">183</span><span id="line-183"> if (</span>
<span class="source-line-no">184</span><span id="line-184"> lastCell == null || lastCell.getTypeByte() != cell.getTypeByte()</span>
<span class="source-line-no">185</span><span id="line-185"> || !CellUtil.matchingRows(lastCell, cell)</span>
<span class="source-line-no">186</span><span id="line-186"> ) {</span>
<span class="source-line-no">187</span><span id="line-187"> // row or type changed, write out aggregate KVs.</span>
<span class="source-line-no">188</span><span id="line-188"> if (put != null) {</span>
<span class="source-line-no">189</span><span id="line-189"> context.write(tableOut, put);</span>
<span class="source-line-no">190</span><span id="line-190"> context.getCounter(Counter.PUTS).increment(1);</span>
<span class="source-line-no">191</span><span id="line-191"> }</span>
<span class="source-line-no">192</span><span id="line-192"> if (del != null) {</span>
<span class="source-line-no">193</span><span id="line-193"> context.write(tableOut, del);</span>
<span class="source-line-no">194</span><span id="line-194"> context.getCounter(Counter.DELETES).increment(1);</span>
<span class="source-line-no">195</span><span id="line-195"> }</span>
<span class="source-line-no">196</span><span id="line-196"> if (CellUtil.isDelete(cell)) {</span>
<span class="source-line-no">197</span><span id="line-197"> del = new Delete(CellUtil.cloneRow(cell));</span>
<span class="source-line-no">198</span><span id="line-198"> } else {</span>
<span class="source-line-no">199</span><span id="line-199"> put = new Put(CellUtil.cloneRow(cell));</span>
<span class="source-line-no">200</span><span id="line-200"> }</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 (CellUtil.isDelete(cell)) {</span>
<span class="source-line-no">203</span><span id="line-203"> del.add(cell);</span>
<span class="source-line-no">204</span><span id="line-204"> } else {</span>
<span class="source-line-no">205</span><span id="line-205"> put.add(cell);</span>
<span class="source-line-no">206</span><span id="line-206"> }</span>
<span class="source-line-no">207</span><span id="line-207"> context.getCounter(Counter.CELLS_WRITTEN).increment(1);</span>
<span class="source-line-no">208</span><span id="line-208"> }</span>
<span class="source-line-no">209</span><span id="line-209"> lastCell = cell;</span>
<span class="source-line-no">210</span><span id="line-210"> }</span>
<span class="source-line-no">211</span><span id="line-211"> // write residual KVs</span>
<span class="source-line-no">212</span><span id="line-212"> if (put != null) {</span>
<span class="source-line-no">213</span><span id="line-213"> context.write(tableOut, put);</span>
<span class="source-line-no">214</span><span id="line-214"> context.getCounter(Counter.PUTS).increment(1);</span>
<span class="source-line-no">215</span><span id="line-215"> }</span>
<span class="source-line-no">216</span><span id="line-216"> if (del != null) {</span>
<span class="source-line-no">217</span><span id="line-217"> context.getCounter(Counter.DELETES).increment(1);</span>
<span class="source-line-no">218</span><span id="line-218"> context.write(tableOut, del);</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"> } catch (InterruptedException e) {</span>
<span class="source-line-no">222</span><span id="line-222"> LOG.error("Interrupted while writing results", e);</span>
<span class="source-line-no">223</span><span id="line-223"> Thread.currentThread().interrupt();</span>
<span class="source-line-no">224</span><span id="line-224"> }</span>
<span class="source-line-no">225</span><span id="line-225"> }</span>
<span class="source-line-no">226</span><span id="line-226"></span>
<span class="source-line-no">227</span><span id="line-227"> protected boolean filter(Context context, final Cell cell) {</span>
<span class="source-line-no">228</span><span id="line-228"> return true;</span>
<span class="source-line-no">229</span><span id="line-229"> }</span>
<span class="source-line-no">230</span><span id="line-230"></span>
<span class="source-line-no">231</span><span id="line-231"> @Override</span>
<span class="source-line-no">232</span><span id="line-232"> protected void</span>
<span class="source-line-no">233</span><span id="line-233"> cleanup(Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Mutation&gt;.Context context)</span>
<span class="source-line-no">234</span><span id="line-234"> throws IOException, InterruptedException {</span>
<span class="source-line-no">235</span><span id="line-235"> super.cleanup(context);</span>
<span class="source-line-no">236</span><span id="line-236"> }</span>
<span class="source-line-no">237</span><span id="line-237"></span>
<span class="source-line-no">238</span><span id="line-238"> @SuppressWarnings("checkstyle:EmptyBlock")</span>
<span class="source-line-no">239</span><span id="line-239"> @Override</span>
<span class="source-line-no">240</span><span id="line-240"> public void setup(Context context) throws IOException {</span>
<span class="source-line-no">241</span><span id="line-241"> String[] tableMap = context.getConfiguration().getStrings(TABLE_MAP_KEY);</span>
<span class="source-line-no">242</span><span id="line-242"> String[] tablesToUse = context.getConfiguration().getStrings(TABLES_KEY);</span>
<span class="source-line-no">243</span><span id="line-243"> if (tableMap == null) {</span>
<span class="source-line-no">244</span><span id="line-244"> tableMap = tablesToUse;</span>
<span class="source-line-no">245</span><span id="line-245"> }</span>
<span class="source-line-no">246</span><span id="line-246"> if (tablesToUse == null) {</span>
<span class="source-line-no">247</span><span id="line-247"> // Then user wants all tables.</span>
<span class="source-line-no">248</span><span id="line-248"> } else if (tablesToUse.length != tableMap.length) {</span>
<span class="source-line-no">249</span><span id="line-249"> // this can only happen when WALMapper is used directly by a class other than WALPlayer</span>
<span class="source-line-no">250</span><span id="line-250"> throw new IOException("Incorrect table mapping specified .");</span>
<span class="source-line-no">251</span><span id="line-251"> }</span>
<span class="source-line-no">252</span><span id="line-252"> int i = 0;</span>
<span class="source-line-no">253</span><span id="line-253"> if (tablesToUse != null) {</span>
<span class="source-line-no">254</span><span id="line-254"> for (String table : tablesToUse) {</span>
<span class="source-line-no">255</span><span id="line-255"> tables.put(TableName.valueOf(table), TableName.valueOf(tableMap[i++]));</span>
<span class="source-line-no">256</span><span id="line-256"> }</span>
<span class="source-line-no">257</span><span id="line-257"> }</span>
<span class="source-line-no">258</span><span id="line-258"> }</span>
<span class="source-line-no">259</span><span id="line-259"> }</span>
<span class="source-line-no">260</span><span id="line-260"></span>
<span class="source-line-no">261</span><span id="line-261"> void setupTime(Configuration conf, String option) throws IOException {</span>
<span class="source-line-no">262</span><span id="line-262"> String val = conf.get(option);</span>
<span class="source-line-no">263</span><span id="line-263"> if (null == val) {</span>
<span class="source-line-no">264</span><span id="line-264"> return;</span>
<span class="source-line-no">265</span><span id="line-265"> }</span>
<span class="source-line-no">266</span><span id="line-266"> long ms;</span>
<span class="source-line-no">267</span><span id="line-267"> try {</span>
<span class="source-line-no">268</span><span id="line-268"> // first try to parse in user friendly form</span>
<span class="source-line-no">269</span><span id="line-269"> ms = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss.SS").parse(val).getTime();</span>
<span class="source-line-no">270</span><span id="line-270"> } catch (ParseException pe) {</span>
<span class="source-line-no">271</span><span id="line-271"> try {</span>
<span class="source-line-no">272</span><span id="line-272"> // then see if just a number of ms's was specified</span>
<span class="source-line-no">273</span><span id="line-273"> ms = Long.parseLong(val);</span>
<span class="source-line-no">274</span><span id="line-274"> } catch (NumberFormatException nfe) {</span>
<span class="source-line-no">275</span><span id="line-275"> throw new IOException(</span>
<span class="source-line-no">276</span><span id="line-276"> option + " must be specified either in the form 2001-02-20T16:35:06.99 "</span>
<span class="source-line-no">277</span><span id="line-277"> + "or as number of milliseconds");</span>
<span class="source-line-no">278</span><span id="line-278"> }</span>
<span class="source-line-no">279</span><span id="line-279"> }</span>
<span class="source-line-no">280</span><span id="line-280"> conf.setLong(option, ms);</span>
<span class="source-line-no">281</span><span id="line-281"> }</span>
<span class="source-line-no">282</span><span id="line-282"></span>
<span class="source-line-no">283</span><span id="line-283"> /**</span>
<span class="source-line-no">284</span><span id="line-284"> * Sets up the actual job.</span>
<span class="source-line-no">285</span><span id="line-285"> * @param args The command line parameters.</span>
<span class="source-line-no">286</span><span id="line-286"> * @return The newly created job.</span>
<span class="source-line-no">287</span><span id="line-287"> * @throws IOException When setting up the job fails.</span>
<span class="source-line-no">288</span><span id="line-288"> */</span>
<span class="source-line-no">289</span><span id="line-289"> public Job createSubmittableJob(String[] args) throws IOException {</span>
<span class="source-line-no">290</span><span id="line-290"> Configuration conf = getConf();</span>
<span class="source-line-no">291</span><span id="line-291"> setupTime(conf, WALInputFormat.START_TIME_KEY);</span>
<span class="source-line-no">292</span><span id="line-292"> setupTime(conf, WALInputFormat.END_TIME_KEY);</span>
<span class="source-line-no">293</span><span id="line-293"> String inputDirs = args[0];</span>
<span class="source-line-no">294</span><span id="line-294"> String[] tables = args.length == 1 ? new String[] {} : args[1].split(",");</span>
<span class="source-line-no">295</span><span id="line-295"> String[] tableMap;</span>
<span class="source-line-no">296</span><span id="line-296"> if (args.length &gt; 2) {</span>
<span class="source-line-no">297</span><span id="line-297"> tableMap = args[2].split(",");</span>
<span class="source-line-no">298</span><span id="line-298"> if (tableMap.length != tables.length) {</span>
<span class="source-line-no">299</span><span id="line-299"> throw new IOException("The same number of tables and mapping must be provided.");</span>
<span class="source-line-no">300</span><span id="line-300"> }</span>
<span class="source-line-no">301</span><span id="line-301"> } else {</span>
<span class="source-line-no">302</span><span id="line-302"> // if no mapping is specified, map each table to itself</span>
<span class="source-line-no">303</span><span id="line-303"> tableMap = tables;</span>
<span class="source-line-no">304</span><span id="line-304"> }</span>
<span class="source-line-no">305</span><span id="line-305"> conf.setStrings(TABLES_KEY, tables);</span>
<span class="source-line-no">306</span><span id="line-306"> conf.setStrings(TABLE_MAP_KEY, tableMap);</span>
<span class="source-line-no">307</span><span id="line-307"> conf.set(FileInputFormat.INPUT_DIR, inputDirs);</span>
<span class="source-line-no">308</span><span id="line-308"> Job job = Job.getInstance(conf,</span>
<span class="source-line-no">309</span><span id="line-309"> conf.get(JOB_NAME_CONF_KEY, NAME + "_" + EnvironmentEdgeManager.currentTime()));</span>
<span class="source-line-no">310</span><span id="line-310"> job.setJarByClass(WALPlayer.class);</span>
<span class="source-line-no">311</span><span id="line-311"></span>
<span class="source-line-no">312</span><span id="line-312"> job.setInputFormatClass(WALInputFormat.class);</span>
<span class="source-line-no">313</span><span id="line-313"> job.setMapOutputKeyClass(ImmutableBytesWritable.class);</span>
<span class="source-line-no">314</span><span id="line-314"></span>
<span class="source-line-no">315</span><span id="line-315"> String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);</span>
<span class="source-line-no">316</span><span id="line-316"> if (hfileOutPath != null) {</span>
<span class="source-line-no">317</span><span id="line-317"> LOG.debug("add incremental job :" + hfileOutPath + " from " + inputDirs);</span>
<span class="source-line-no">318</span><span id="line-318"></span>
<span class="source-line-no">319</span><span id="line-319"> // WALPlayer needs ExtendedCellSerialization so that sequenceId can be propagated when</span>
<span class="source-line-no">320</span><span id="line-320"> // sorting cells in CellSortReducer</span>
<span class="source-line-no">321</span><span id="line-321"> job.getConfiguration().setBoolean(HFileOutputFormat2.EXTENDED_CELL_SERIALIZATION_ENABLED_KEY,</span>
<span class="source-line-no">322</span><span id="line-322"> true);</span>
<span class="source-line-no">323</span><span id="line-323"></span>
<span class="source-line-no">324</span><span id="line-324"> // the bulk HFile case</span>
<span class="source-line-no">325</span><span id="line-325"> List&lt;TableName&gt; tableNames = getTableNameList(tables);</span>
<span class="source-line-no">326</span><span id="line-326"></span>
<span class="source-line-no">327</span><span id="line-327"> job.setMapperClass(WALKeyValueMapper.class);</span>
<span class="source-line-no">328</span><span id="line-328"> job.setReducerClass(CellSortReducer.class);</span>
<span class="source-line-no">329</span><span id="line-329"> Path outputDir = new Path(hfileOutPath);</span>
<span class="source-line-no">330</span><span id="line-330"> FileOutputFormat.setOutputPath(job, outputDir);</span>
<span class="source-line-no">331</span><span id="line-331"> job.setMapOutputValueClass(MapReduceExtendedCell.class);</span>
<span class="source-line-no">332</span><span id="line-332"> try (Connection conn = ConnectionFactory.createConnection(conf);) {</span>
<span class="source-line-no">333</span><span id="line-333"> List&lt;TableInfo&gt; tableInfoList = new ArrayList&lt;TableInfo&gt;();</span>
<span class="source-line-no">334</span><span id="line-334"> for (TableName tableName : tableNames) {</span>
<span class="source-line-no">335</span><span id="line-335"> Table table = conn.getTable(tableName);</span>
<span class="source-line-no">336</span><span id="line-336"> RegionLocator regionLocator = conn.getRegionLocator(tableName);</span>
<span class="source-line-no">337</span><span id="line-337"> tableInfoList.add(new TableInfo(table.getDescriptor(), regionLocator));</span>
<span class="source-line-no">338</span><span id="line-338"> }</span>
<span class="source-line-no">339</span><span id="line-339"> MultiTableHFileOutputFormat.configureIncrementalLoad(job, tableInfoList);</span>
<span class="source-line-no">340</span><span id="line-340"> }</span>
<span class="source-line-no">341</span><span id="line-341"> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),</span>
<span class="source-line-no">342</span><span id="line-342"> org.apache.hbase.thirdparty.com.google.common.base.Preconditions.class);</span>
<span class="source-line-no">343</span><span id="line-343"> } else {</span>
<span class="source-line-no">344</span><span id="line-344"> // output to live cluster</span>
<span class="source-line-no">345</span><span id="line-345"> job.setMapperClass(WALMapper.class);</span>
<span class="source-line-no">346</span><span id="line-346"> job.setOutputFormatClass(MultiTableOutputFormat.class);</span>
<span class="source-line-no">347</span><span id="line-347"> TableMapReduceUtil.addDependencyJars(job);</span>
<span class="source-line-no">348</span><span id="line-348"> TableMapReduceUtil.initCredentials(job);</span>
<span class="source-line-no">349</span><span id="line-349"> // No reducers.</span>
<span class="source-line-no">350</span><span id="line-350"> job.setNumReduceTasks(0);</span>
<span class="source-line-no">351</span><span id="line-351"> }</span>
<span class="source-line-no">352</span><span id="line-352"> String codecCls = WALCellCodec.getWALCellCodecClass(conf).getName();</span>
<span class="source-line-no">353</span><span id="line-353"> try {</span>
<span class="source-line-no">354</span><span id="line-354"> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),</span>
<span class="source-line-no">355</span><span id="line-355"> Class.forName(codecCls));</span>
<span class="source-line-no">356</span><span id="line-356"> } catch (Exception e) {</span>
<span class="source-line-no">357</span><span id="line-357"> throw new IOException("Cannot determine wal codec class " + codecCls, e);</span>
<span class="source-line-no">358</span><span id="line-358"> }</span>
<span class="source-line-no">359</span><span id="line-359"> return job;</span>
<span class="source-line-no">360</span><span id="line-360"> }</span>
<span class="source-line-no">361</span><span id="line-361"></span>
<span class="source-line-no">362</span><span id="line-362"> private List&lt;TableName&gt; getTableNameList(String[] tables) {</span>
<span class="source-line-no">363</span><span id="line-363"> List&lt;TableName&gt; list = new ArrayList&lt;TableName&gt;();</span>
<span class="source-line-no">364</span><span id="line-364"> for (String name : tables) {</span>
<span class="source-line-no">365</span><span id="line-365"> list.add(TableName.valueOf(name));</span>
<span class="source-line-no">366</span><span id="line-366"> }</span>
<span class="source-line-no">367</span><span id="line-367"> return list;</span>
<span class="source-line-no">368</span><span id="line-368"> }</span>
<span class="source-line-no">369</span><span id="line-369"></span>
<span class="source-line-no">370</span><span id="line-370"> /**</span>
<span class="source-line-no">371</span><span id="line-371"> * Print usage</span>
<span class="source-line-no">372</span><span id="line-372"> * @param errorMsg Error message. Can be null.</span>
<span class="source-line-no">373</span><span id="line-373"> */</span>
<span class="source-line-no">374</span><span id="line-374"> private void usage(final String errorMsg) {</span>
<span class="source-line-no">375</span><span id="line-375"> if (errorMsg != null &amp;&amp; errorMsg.length() &gt; 0) {</span>
<span class="source-line-no">376</span><span id="line-376"> System.err.println("ERROR: " + errorMsg);</span>
<span class="source-line-no">377</span><span id="line-377"> }</span>
<span class="source-line-no">378</span><span id="line-378"> System.err.println("Usage: " + NAME + " [options] &lt;WAL inputdir&gt; [&lt;tables&gt; &lt;tableMappings&gt;]");</span>
<span class="source-line-no">379</span><span id="line-379"> System.err.println(" &lt;WAL inputdir&gt; directory of WALs to replay.");</span>
<span class="source-line-no">380</span><span id="line-380"> System.err.println(" &lt;tables&gt; comma separated list of tables. If no tables specified,");</span>
<span class="source-line-no">381</span><span id="line-381"> System.err.println(" all are imported (even hbase:meta if present).");</span>
<span class="source-line-no">382</span><span id="line-382"> System.err.println(</span>
<span class="source-line-no">383</span><span id="line-383"> " &lt;tableMappings&gt; WAL entries can be mapped to a new set of tables by " + "passing");</span>
<span class="source-line-no">384</span><span id="line-384"> System.err</span>
<span class="source-line-no">385</span><span id="line-385"> .println(" &lt;tableMappings&gt;, a comma separated list of target " + "tables.");</span>
<span class="source-line-no">386</span><span id="line-386"> System.err</span>
<span class="source-line-no">387</span><span id="line-387"> .println(" If specified, each table in &lt;tables&gt; must have a " + "mapping.");</span>
<span class="source-line-no">388</span><span id="line-388"> System.err.println("To generate HFiles to bulk load instead of loading HBase directly, pass:");</span>
<span class="source-line-no">389</span><span id="line-389"> System.err.println(" -D" + BULK_OUTPUT_CONF_KEY + "=/path/for/output");</span>
<span class="source-line-no">390</span><span id="line-390"> System.err.println(" Only one table can be specified, and no mapping allowed!");</span>
<span class="source-line-no">391</span><span id="line-391"> System.err.println("To specify a time range, pass:");</span>
<span class="source-line-no">392</span><span id="line-392"> System.err.println(" -D" + WALInputFormat.START_TIME_KEY + "=[date|ms]");</span>
<span class="source-line-no">393</span><span id="line-393"> System.err.println(" -D" + WALInputFormat.END_TIME_KEY + "=[date|ms]");</span>
<span class="source-line-no">394</span><span id="line-394"> System.err.println(" The start and the end date of timerange (inclusive). The dates can be");</span>
<span class="source-line-no">395</span><span id="line-395"> System.err</span>
<span class="source-line-no">396</span><span id="line-396"> .println(" expressed in milliseconds-since-epoch or yyyy-MM-dd'T'HH:mm:ss.SS " + "format.");</span>
<span class="source-line-no">397</span><span id="line-397"> System.err.println(" E.g. 1234567890120 or 2009-02-13T23:32:30.12");</span>
<span class="source-line-no">398</span><span id="line-398"> System.err.println("Other options:");</span>
<span class="source-line-no">399</span><span id="line-399"> System.err.println(" -D" + JOB_NAME_CONF_KEY + "=jobName");</span>
<span class="source-line-no">400</span><span id="line-400"> System.err.println(" Use the specified mapreduce job name for the wal player");</span>
<span class="source-line-no">401</span><span id="line-401"> System.err.println(" -Dwal.input.separator=' '");</span>
<span class="source-line-no">402</span><span id="line-402"> System.err.println(" Change WAL filename separator (WAL dir names use default ','.)");</span>
<span class="source-line-no">403</span><span id="line-403"> System.err.println("For performance also consider the following options:\n"</span>
<span class="source-line-no">404</span><span id="line-404"> + " -Dmapreduce.map.speculative=false\n" + " -Dmapreduce.reduce.speculative=false");</span>
<span class="source-line-no">405</span><span id="line-405"> }</span>
<span class="source-line-no">406</span><span id="line-406"></span>
<span class="source-line-no">407</span><span id="line-407"> /**</span>
<span class="source-line-no">408</span><span id="line-408"> * Main entry point.</span>
<span class="source-line-no">409</span><span id="line-409"> * @param args The command line parameters.</span>
<span class="source-line-no">410</span><span id="line-410"> * @throws Exception When running the job fails.</span>
<span class="source-line-no">411</span><span id="line-411"> */</span>
<span class="source-line-no">412</span><span id="line-412"> public static void main(String[] args) throws Exception {</span>
<span class="source-line-no">413</span><span id="line-413"> int ret = ToolRunner.run(new WALPlayer(HBaseConfiguration.create()), args);</span>
<span class="source-line-no">414</span><span id="line-414"> System.exit(ret);</span>
<span class="source-line-no">415</span><span id="line-415"> }</span>
<span class="source-line-no">416</span><span id="line-416"></span>
<span class="source-line-no">417</span><span id="line-417"> @Override</span>
<span class="source-line-no">418</span><span id="line-418"> public int run(String[] args) throws Exception {</span>
<span class="source-line-no">419</span><span id="line-419"> if (args.length &lt; 1) {</span>
<span class="source-line-no">420</span><span id="line-420"> usage("Wrong number of arguments: " + args.length);</span>
<span class="source-line-no">421</span><span id="line-421"> System.exit(-1);</span>
<span class="source-line-no">422</span><span id="line-422"> }</span>
<span class="source-line-no">423</span><span id="line-423"> Job job = createSubmittableJob(args);</span>
<span class="source-line-no">424</span><span id="line-424"> return job.waitForCompletion(true) ? 0 : 1;</span>
<span class="source-line-no">425</span><span id="line-425"> }</span>
<span class="source-line-no">426</span><span id="line-426">}</span>
</pre>
</div>
</main>
</body>
</html>