blob: dd3c0593805c3e81e93ba7824ee3ff87a80baf14 [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.text.ParseException;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.text.SimpleDateFormat;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.ArrayList;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.util.HashSet;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.List;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.Map;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.util.Set;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import java.util.TreeMap;<a name="line.28"></a>
<span class="sourceLineNo">029</span><a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configuration;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.conf.Configured;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.fs.Path;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.Cell;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.TableName;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.client.Delete;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.client.Mutation;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.client.Put;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.client.Table;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.mapreduce.HFileOutputFormat2.TableInfo;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.regionserver.wal.WALCellCodec;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.util.MapReduceExtendedCell;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.wal.WALEdit;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.wal.WALKey;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.hadoop.mapreduce.Job;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.54"></a>
<span class="sourceLineNo">055</span>import org.apache.hadoop.util.Tool;<a name="line.55"></a>
<span class="sourceLineNo">056</span>import org.apache.hadoop.util.ToolRunner;<a name="line.56"></a>
<span class="sourceLineNo">057</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.57"></a>
<span class="sourceLineNo">058</span>import org.slf4j.Logger;<a name="line.58"></a>
<span class="sourceLineNo">059</span>import org.slf4j.LoggerFactory;<a name="line.59"></a>
<span class="sourceLineNo">060</span><a name="line.60"></a>
<span class="sourceLineNo">061</span>/**<a name="line.61"></a>
<span class="sourceLineNo">062</span> * A tool to replay WAL files as a M/R job.<a name="line.62"></a>
<span class="sourceLineNo">063</span> * The WAL can be replayed for a set of tables or all tables,<a name="line.63"></a>
<span class="sourceLineNo">064</span> * and a time range can be provided (in milliseconds).<a name="line.64"></a>
<span class="sourceLineNo">065</span> * The WAL is filtered to the passed set of tables and the output<a name="line.65"></a>
<span class="sourceLineNo">066</span> * can optionally be mapped to another set of tables.<a name="line.66"></a>
<span class="sourceLineNo">067</span> *<a name="line.67"></a>
<span class="sourceLineNo">068</span> * WAL replay can also generate HFiles for later bulk importing,<a name="line.68"></a>
<span class="sourceLineNo">069</span> * in that case the WAL is replayed for a single table only.<a name="line.69"></a>
<span class="sourceLineNo">070</span> */<a name="line.70"></a>
<span class="sourceLineNo">071</span>@InterfaceAudience.Public<a name="line.71"></a>
<span class="sourceLineNo">072</span>public class WALPlayer extends Configured implements Tool {<a name="line.72"></a>
<span class="sourceLineNo">073</span> private static final Logger LOG = LoggerFactory.getLogger(WALPlayer.class);<a name="line.73"></a>
<span class="sourceLineNo">074</span> final static String NAME = "WALPlayer";<a name="line.74"></a>
<span class="sourceLineNo">075</span> public final static String BULK_OUTPUT_CONF_KEY = "wal.bulk.output";<a name="line.75"></a>
<span class="sourceLineNo">076</span> public final static String TABLES_KEY = "wal.input.tables";<a name="line.76"></a>
<span class="sourceLineNo">077</span> public final static String TABLE_MAP_KEY = "wal.input.tablesmap";<a name="line.77"></a>
<span class="sourceLineNo">078</span> public final static String INPUT_FILES_SEPARATOR_KEY = "wal.input.separator";<a name="line.78"></a>
<span class="sourceLineNo">079</span> public final static String IGNORE_MISSING_FILES = "wal.input.ignore.missing.files";<a name="line.79"></a>
<span class="sourceLineNo">080</span> public final static String MULTI_TABLES_SUPPORT = "wal.multi.tables.support";<a name="line.80"></a>
<span class="sourceLineNo">081</span><a name="line.81"></a>
<span class="sourceLineNo">082</span> protected static final String tableSeparator = ";";<a name="line.82"></a>
<span class="sourceLineNo">083</span><a name="line.83"></a>
<span class="sourceLineNo">084</span> // This relies on Hadoop Configuration to handle warning about deprecated configs and<a name="line.84"></a>
<span class="sourceLineNo">085</span> // to set the correct non-deprecated configs when an old one shows up.<a name="line.85"></a>
<span class="sourceLineNo">086</span> static {<a name="line.86"></a>
<span class="sourceLineNo">087</span> Configuration.addDeprecation("hlog.bulk.output", BULK_OUTPUT_CONF_KEY);<a name="line.87"></a>
<span class="sourceLineNo">088</span> Configuration.addDeprecation("hlog.input.tables", TABLES_KEY);<a name="line.88"></a>
<span class="sourceLineNo">089</span> Configuration.addDeprecation("hlog.input.tablesmap", TABLE_MAP_KEY);<a name="line.89"></a>
<span class="sourceLineNo">090</span> }<a name="line.90"></a>
<span class="sourceLineNo">091</span><a name="line.91"></a>
<span class="sourceLineNo">092</span> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.92"></a>
<span class="sourceLineNo">093</span><a name="line.93"></a>
<span class="sourceLineNo">094</span> public WALPlayer() {<a name="line.94"></a>
<span class="sourceLineNo">095</span> }<a name="line.95"></a>
<span class="sourceLineNo">096</span><a name="line.96"></a>
<span class="sourceLineNo">097</span> protected WALPlayer(final Configuration c) {<a name="line.97"></a>
<span class="sourceLineNo">098</span> super(c);<a name="line.98"></a>
<span class="sourceLineNo">099</span> }<a name="line.99"></a>
<span class="sourceLineNo">100</span><a name="line.100"></a>
<span class="sourceLineNo">101</span> /**<a name="line.101"></a>
<span class="sourceLineNo">102</span> * A mapper that just writes out KeyValues. This one can be used together with<a name="line.102"></a>
<span class="sourceLineNo">103</span> * {@link CellSortReducer}<a name="line.103"></a>
<span class="sourceLineNo">104</span> */<a name="line.104"></a>
<span class="sourceLineNo">105</span> static class WALKeyValueMapper extends Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Cell&gt; {<a name="line.105"></a>
<span class="sourceLineNo">106</span> private Set&lt;String&gt; tableSet = new HashSet&lt;String&gt;();<a name="line.106"></a>
<span class="sourceLineNo">107</span> private boolean multiTableSupport = false;<a name="line.107"></a>
<span class="sourceLineNo">108</span><a name="line.108"></a>
<span class="sourceLineNo">109</span> @Override<a name="line.109"></a>
<span class="sourceLineNo">110</span> public void map(WALKey key, WALEdit value, Context context) throws IOException {<a name="line.110"></a>
<span class="sourceLineNo">111</span> try {<a name="line.111"></a>
<span class="sourceLineNo">112</span> // skip all other tables<a name="line.112"></a>
<span class="sourceLineNo">113</span> TableName table = key.getTableName();<a name="line.113"></a>
<span class="sourceLineNo">114</span> if (tableSet.contains(table.getNameAsString())) {<a name="line.114"></a>
<span class="sourceLineNo">115</span> for (Cell cell : value.getCells()) {<a name="line.115"></a>
<span class="sourceLineNo">116</span> if (WALEdit.isMetaEditFamily(cell)) {<a name="line.116"></a>
<span class="sourceLineNo">117</span> continue;<a name="line.117"></a>
<span class="sourceLineNo">118</span> }<a name="line.118"></a>
<span class="sourceLineNo">119</span> byte[] outKey = multiTableSupport<a name="line.119"></a>
<span class="sourceLineNo">120</span> ? Bytes.add(table.getName(), Bytes.toBytes(tableSeparator), CellUtil.cloneRow(cell))<a name="line.120"></a>
<span class="sourceLineNo">121</span> : CellUtil.cloneRow(cell);<a name="line.121"></a>
<span class="sourceLineNo">122</span> context.write(new ImmutableBytesWritable(outKey), new MapReduceExtendedCell(cell));<a name="line.122"></a>
<span class="sourceLineNo">123</span> }<a name="line.123"></a>
<span class="sourceLineNo">124</span> }<a name="line.124"></a>
<span class="sourceLineNo">125</span> } catch (InterruptedException e) {<a name="line.125"></a>
<span class="sourceLineNo">126</span> LOG.error("Interrupted while emitting Cell", e);<a name="line.126"></a>
<span class="sourceLineNo">127</span> Thread.currentThread().interrupt();<a name="line.127"></a>
<span class="sourceLineNo">128</span> }<a name="line.128"></a>
<span class="sourceLineNo">129</span> }<a name="line.129"></a>
<span class="sourceLineNo">130</span><a name="line.130"></a>
<span class="sourceLineNo">131</span> @Override<a name="line.131"></a>
<span class="sourceLineNo">132</span> public void setup(Context context) throws IOException {<a name="line.132"></a>
<span class="sourceLineNo">133</span> Configuration conf = context.getConfiguration();<a name="line.133"></a>
<span class="sourceLineNo">134</span> String[] tables = conf.getStrings(TABLES_KEY);<a name="line.134"></a>
<span class="sourceLineNo">135</span> this.multiTableSupport = conf.getBoolean(MULTI_TABLES_SUPPORT, false);<a name="line.135"></a>
<span class="sourceLineNo">136</span> for (String table : tables) {<a name="line.136"></a>
<span class="sourceLineNo">137</span> tableSet.add(table);<a name="line.137"></a>
<span class="sourceLineNo">138</span> }<a name="line.138"></a>
<span class="sourceLineNo">139</span> }<a name="line.139"></a>
<span class="sourceLineNo">140</span> }<a name="line.140"></a>
<span class="sourceLineNo">141</span><a name="line.141"></a>
<span class="sourceLineNo">142</span> /**<a name="line.142"></a>
<span class="sourceLineNo">143</span> * A mapper that writes out {@link Mutation} to be directly applied to a running HBase instance.<a name="line.143"></a>
<span class="sourceLineNo">144</span> */<a name="line.144"></a>
<span class="sourceLineNo">145</span> protected static class WALMapper<a name="line.145"></a>
<span class="sourceLineNo">146</span> extends Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Mutation&gt; {<a name="line.146"></a>
<span class="sourceLineNo">147</span> private Map&lt;TableName, TableName&gt; tables = new TreeMap&lt;&gt;();<a name="line.147"></a>
<span class="sourceLineNo">148</span><a name="line.148"></a>
<span class="sourceLineNo">149</span> @Override<a name="line.149"></a>
<span class="sourceLineNo">150</span> public void map(WALKey key, WALEdit value, Context context) throws IOException {<a name="line.150"></a>
<span class="sourceLineNo">151</span> try {<a name="line.151"></a>
<span class="sourceLineNo">152</span> if (tables.isEmpty() || tables.containsKey(key.getTableName())) {<a name="line.152"></a>
<span class="sourceLineNo">153</span> TableName targetTable =<a name="line.153"></a>
<span class="sourceLineNo">154</span> tables.isEmpty() ? key.getTableName() : tables.get(key.getTableName());<a name="line.154"></a>
<span class="sourceLineNo">155</span> ImmutableBytesWritable tableOut = new ImmutableBytesWritable(targetTable.getName());<a name="line.155"></a>
<span class="sourceLineNo">156</span> Put put = null;<a name="line.156"></a>
<span class="sourceLineNo">157</span> Delete del = null;<a name="line.157"></a>
<span class="sourceLineNo">158</span> Cell lastCell = null;<a name="line.158"></a>
<span class="sourceLineNo">159</span> for (Cell cell : value.getCells()) {<a name="line.159"></a>
<span class="sourceLineNo">160</span> // Filtering WAL meta marker entries.<a name="line.160"></a>
<span class="sourceLineNo">161</span> if (WALEdit.isMetaEditFamily(cell)) {<a name="line.161"></a>
<span class="sourceLineNo">162</span> continue;<a name="line.162"></a>
<span class="sourceLineNo">163</span> }<a name="line.163"></a>
<span class="sourceLineNo">164</span> // Allow a subclass filter out this cell.<a name="line.164"></a>
<span class="sourceLineNo">165</span> if (filter(context, cell)) {<a name="line.165"></a>
<span class="sourceLineNo">166</span> // A WALEdit may contain multiple operations (HBASE-3584) and/or<a name="line.166"></a>
<span class="sourceLineNo">167</span> // multiple rows (HBASE-5229).<a name="line.167"></a>
<span class="sourceLineNo">168</span> // Aggregate as much as possible into a single Put/Delete<a name="line.168"></a>
<span class="sourceLineNo">169</span> // operation before writing to the context.<a name="line.169"></a>
<span class="sourceLineNo">170</span> if (lastCell == null || lastCell.getTypeByte() != cell.getTypeByte()<a name="line.170"></a>
<span class="sourceLineNo">171</span> || !CellUtil.matchingRows(lastCell, cell)) {<a name="line.171"></a>
<span class="sourceLineNo">172</span> // row or type changed, write out aggregate KVs.<a name="line.172"></a>
<span class="sourceLineNo">173</span> if (put != null) {<a name="line.173"></a>
<span class="sourceLineNo">174</span> context.write(tableOut, put);<a name="line.174"></a>
<span class="sourceLineNo">175</span> }<a name="line.175"></a>
<span class="sourceLineNo">176</span> if (del != null) {<a name="line.176"></a>
<span class="sourceLineNo">177</span> context.write(tableOut, del);<a name="line.177"></a>
<span class="sourceLineNo">178</span> }<a name="line.178"></a>
<span class="sourceLineNo">179</span> if (CellUtil.isDelete(cell)) {<a name="line.179"></a>
<span class="sourceLineNo">180</span> del = new Delete(CellUtil.cloneRow(cell));<a name="line.180"></a>
<span class="sourceLineNo">181</span> } else {<a name="line.181"></a>
<span class="sourceLineNo">182</span> put = new Put(CellUtil.cloneRow(cell));<a name="line.182"></a>
<span class="sourceLineNo">183</span> }<a name="line.183"></a>
<span class="sourceLineNo">184</span> }<a name="line.184"></a>
<span class="sourceLineNo">185</span> if (CellUtil.isDelete(cell)) {<a name="line.185"></a>
<span class="sourceLineNo">186</span> del.add(cell);<a name="line.186"></a>
<span class="sourceLineNo">187</span> } else {<a name="line.187"></a>
<span class="sourceLineNo">188</span> put.add(cell);<a name="line.188"></a>
<span class="sourceLineNo">189</span> }<a name="line.189"></a>
<span class="sourceLineNo">190</span> }<a name="line.190"></a>
<span class="sourceLineNo">191</span> lastCell = cell;<a name="line.191"></a>
<span class="sourceLineNo">192</span> }<a name="line.192"></a>
<span class="sourceLineNo">193</span> // write residual KVs<a name="line.193"></a>
<span class="sourceLineNo">194</span> if (put != null) {<a name="line.194"></a>
<span class="sourceLineNo">195</span> context.write(tableOut, put);<a name="line.195"></a>
<span class="sourceLineNo">196</span> }<a name="line.196"></a>
<span class="sourceLineNo">197</span> if (del != null) {<a name="line.197"></a>
<span class="sourceLineNo">198</span> context.write(tableOut, del);<a name="line.198"></a>
<span class="sourceLineNo">199</span> }<a name="line.199"></a>
<span class="sourceLineNo">200</span> }<a name="line.200"></a>
<span class="sourceLineNo">201</span> } catch (InterruptedException e) {<a name="line.201"></a>
<span class="sourceLineNo">202</span> LOG.error("Interrupted while writing results", e);<a name="line.202"></a>
<span class="sourceLineNo">203</span> Thread.currentThread().interrupt();<a name="line.203"></a>
<span class="sourceLineNo">204</span> }<a name="line.204"></a>
<span class="sourceLineNo">205</span> }<a name="line.205"></a>
<span class="sourceLineNo">206</span><a name="line.206"></a>
<span class="sourceLineNo">207</span> protected boolean filter(Context context, final Cell cell) {<a name="line.207"></a>
<span class="sourceLineNo">208</span> return true;<a name="line.208"></a>
<span class="sourceLineNo">209</span> }<a name="line.209"></a>
<span class="sourceLineNo">210</span><a name="line.210"></a>
<span class="sourceLineNo">211</span> @Override<a name="line.211"></a>
<span class="sourceLineNo">212</span> protected void<a name="line.212"></a>
<span class="sourceLineNo">213</span> cleanup(Mapper&lt;WALKey, WALEdit, ImmutableBytesWritable, Mutation&gt;.Context context)<a name="line.213"></a>
<span class="sourceLineNo">214</span> throws IOException, InterruptedException {<a name="line.214"></a>
<span class="sourceLineNo">215</span> super.cleanup(context);<a name="line.215"></a>
<span class="sourceLineNo">216</span> }<a name="line.216"></a>
<span class="sourceLineNo">217</span><a name="line.217"></a>
<span class="sourceLineNo">218</span> @Override<a name="line.218"></a>
<span class="sourceLineNo">219</span> public void setup(Context context) throws IOException {<a name="line.219"></a>
<span class="sourceLineNo">220</span> String[] tableMap = context.getConfiguration().getStrings(TABLE_MAP_KEY);<a name="line.220"></a>
<span class="sourceLineNo">221</span> String[] tablesToUse = context.getConfiguration().getStrings(TABLES_KEY);<a name="line.221"></a>
<span class="sourceLineNo">222</span> if (tableMap == null) {<a name="line.222"></a>
<span class="sourceLineNo">223</span> tableMap = tablesToUse;<a name="line.223"></a>
<span class="sourceLineNo">224</span> }<a name="line.224"></a>
<span class="sourceLineNo">225</span> if (tablesToUse == null) {<a name="line.225"></a>
<span class="sourceLineNo">226</span> // Then user wants all tables.<a name="line.226"></a>
<span class="sourceLineNo">227</span> } else if (tablesToUse.length != tableMap.length) {<a name="line.227"></a>
<span class="sourceLineNo">228</span> // this can only happen when WALMapper is used directly by a class other than WALPlayer<a name="line.228"></a>
<span class="sourceLineNo">229</span> throw new IOException("Incorrect table mapping specified .");<a name="line.229"></a>
<span class="sourceLineNo">230</span> }<a name="line.230"></a>
<span class="sourceLineNo">231</span> int i = 0;<a name="line.231"></a>
<span class="sourceLineNo">232</span> if (tablesToUse != null) {<a name="line.232"></a>
<span class="sourceLineNo">233</span> for (String table : tablesToUse) {<a name="line.233"></a>
<span class="sourceLineNo">234</span> tables.put(TableName.valueOf(table), TableName.valueOf(tableMap[i++]));<a name="line.234"></a>
<span class="sourceLineNo">235</span> }<a name="line.235"></a>
<span class="sourceLineNo">236</span> }<a name="line.236"></a>
<span class="sourceLineNo">237</span> }<a name="line.237"></a>
<span class="sourceLineNo">238</span> }<a name="line.238"></a>
<span class="sourceLineNo">239</span><a name="line.239"></a>
<span class="sourceLineNo">240</span> void setupTime(Configuration conf, String option) throws IOException {<a name="line.240"></a>
<span class="sourceLineNo">241</span> String val = conf.get(option);<a name="line.241"></a>
<span class="sourceLineNo">242</span> if (null == val) {<a name="line.242"></a>
<span class="sourceLineNo">243</span> return;<a name="line.243"></a>
<span class="sourceLineNo">244</span> }<a name="line.244"></a>
<span class="sourceLineNo">245</span> long ms;<a name="line.245"></a>
<span class="sourceLineNo">246</span> try {<a name="line.246"></a>
<span class="sourceLineNo">247</span> // first try to parse in user friendly form<a name="line.247"></a>
<span class="sourceLineNo">248</span> ms = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss.SS").parse(val).getTime();<a name="line.248"></a>
<span class="sourceLineNo">249</span> } catch (ParseException pe) {<a name="line.249"></a>
<span class="sourceLineNo">250</span> try {<a name="line.250"></a>
<span class="sourceLineNo">251</span> // then see if just a number of ms's was specified<a name="line.251"></a>
<span class="sourceLineNo">252</span> ms = Long.parseLong(val);<a name="line.252"></a>
<span class="sourceLineNo">253</span> } catch (NumberFormatException nfe) {<a name="line.253"></a>
<span class="sourceLineNo">254</span> throw new IOException(<a name="line.254"></a>
<span class="sourceLineNo">255</span> option + " must be specified either in the form 2001-02-20T16:35:06.99 "<a name="line.255"></a>
<span class="sourceLineNo">256</span> + "or as number of milliseconds");<a name="line.256"></a>
<span class="sourceLineNo">257</span> }<a name="line.257"></a>
<span class="sourceLineNo">258</span> }<a name="line.258"></a>
<span class="sourceLineNo">259</span> conf.setLong(option, ms);<a name="line.259"></a>
<span class="sourceLineNo">260</span> }<a name="line.260"></a>
<span class="sourceLineNo">261</span><a name="line.261"></a>
<span class="sourceLineNo">262</span> /**<a name="line.262"></a>
<span class="sourceLineNo">263</span> * Sets up the actual job.<a name="line.263"></a>
<span class="sourceLineNo">264</span> * @param args The command line parameters.<a name="line.264"></a>
<span class="sourceLineNo">265</span> * @return The newly created job.<a name="line.265"></a>
<span class="sourceLineNo">266</span> * @throws IOException When setting up the job fails.<a name="line.266"></a>
<span class="sourceLineNo">267</span> */<a name="line.267"></a>
<span class="sourceLineNo">268</span> public Job createSubmittableJob(String[] args) throws IOException {<a name="line.268"></a>
<span class="sourceLineNo">269</span> Configuration conf = getConf();<a name="line.269"></a>
<span class="sourceLineNo">270</span> setupTime(conf, WALInputFormat.START_TIME_KEY);<a name="line.270"></a>
<span class="sourceLineNo">271</span> setupTime(conf, WALInputFormat.END_TIME_KEY);<a name="line.271"></a>
<span class="sourceLineNo">272</span> String inputDirs = args[0];<a name="line.272"></a>
<span class="sourceLineNo">273</span> String[] tables = args[1].split(",");<a name="line.273"></a>
<span class="sourceLineNo">274</span> String[] tableMap;<a name="line.274"></a>
<span class="sourceLineNo">275</span> if (args.length &gt; 2) {<a name="line.275"></a>
<span class="sourceLineNo">276</span> tableMap = args[2].split(",");<a name="line.276"></a>
<span class="sourceLineNo">277</span> if (tableMap.length != tables.length) {<a name="line.277"></a>
<span class="sourceLineNo">278</span> throw new IOException("The same number of tables and mapping must be provided.");<a name="line.278"></a>
<span class="sourceLineNo">279</span> }<a name="line.279"></a>
<span class="sourceLineNo">280</span> } else {<a name="line.280"></a>
<span class="sourceLineNo">281</span> // if not mapping is specified map each table to itself<a name="line.281"></a>
<span class="sourceLineNo">282</span> tableMap = tables;<a name="line.282"></a>
<span class="sourceLineNo">283</span> }<a name="line.283"></a>
<span class="sourceLineNo">284</span> conf.setStrings(TABLES_KEY, tables);<a name="line.284"></a>
<span class="sourceLineNo">285</span> conf.setStrings(TABLE_MAP_KEY, tableMap);<a name="line.285"></a>
<span class="sourceLineNo">286</span> conf.set(FileInputFormat.INPUT_DIR, inputDirs);<a name="line.286"></a>
<span class="sourceLineNo">287</span> Job job =<a name="line.287"></a>
<span class="sourceLineNo">288</span> Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + System.currentTimeMillis()));<a name="line.288"></a>
<span class="sourceLineNo">289</span> job.setJarByClass(WALPlayer.class);<a name="line.289"></a>
<span class="sourceLineNo">290</span><a name="line.290"></a>
<span class="sourceLineNo">291</span> job.setInputFormatClass(WALInputFormat.class);<a name="line.291"></a>
<span class="sourceLineNo">292</span> job.setMapOutputKeyClass(ImmutableBytesWritable.class);<a name="line.292"></a>
<span class="sourceLineNo">293</span><a name="line.293"></a>
<span class="sourceLineNo">294</span> String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);<a name="line.294"></a>
<span class="sourceLineNo">295</span> if (hfileOutPath != null) {<a name="line.295"></a>
<span class="sourceLineNo">296</span> LOG.debug("add incremental job :" + hfileOutPath + " from " + inputDirs);<a name="line.296"></a>
<span class="sourceLineNo">297</span><a name="line.297"></a>
<span class="sourceLineNo">298</span> // the bulk HFile case<a name="line.298"></a>
<span class="sourceLineNo">299</span> List&lt;TableName&gt; tableNames = getTableNameList(tables);<a name="line.299"></a>
<span class="sourceLineNo">300</span><a name="line.300"></a>
<span class="sourceLineNo">301</span> job.setMapperClass(WALKeyValueMapper.class);<a name="line.301"></a>
<span class="sourceLineNo">302</span> job.setReducerClass(CellSortReducer.class);<a name="line.302"></a>
<span class="sourceLineNo">303</span> Path outputDir = new Path(hfileOutPath);<a name="line.303"></a>
<span class="sourceLineNo">304</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.304"></a>
<span class="sourceLineNo">305</span> job.setMapOutputValueClass(MapReduceExtendedCell.class);<a name="line.305"></a>
<span class="sourceLineNo">306</span> try (Connection conn = ConnectionFactory.createConnection(conf);) {<a name="line.306"></a>
<span class="sourceLineNo">307</span> List&lt;TableInfo&gt; tableInfoList = new ArrayList&lt;TableInfo&gt;();<a name="line.307"></a>
<span class="sourceLineNo">308</span> for (TableName tableName : tableNames) {<a name="line.308"></a>
<span class="sourceLineNo">309</span> Table table = conn.getTable(tableName);<a name="line.309"></a>
<span class="sourceLineNo">310</span> RegionLocator regionLocator = conn.getRegionLocator(tableName);<a name="line.310"></a>
<span class="sourceLineNo">311</span> tableInfoList.add(new TableInfo(table.getDescriptor(), regionLocator));<a name="line.311"></a>
<span class="sourceLineNo">312</span> }<a name="line.312"></a>
<span class="sourceLineNo">313</span> MultiTableHFileOutputFormat.configureIncrementalLoad(job, tableInfoList);<a name="line.313"></a>
<span class="sourceLineNo">314</span> }<a name="line.314"></a>
<span class="sourceLineNo">315</span> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),<a name="line.315"></a>
<span class="sourceLineNo">316</span> org.apache.hbase.thirdparty.com.google.common.base.Preconditions.class);<a name="line.316"></a>
<span class="sourceLineNo">317</span> } else {<a name="line.317"></a>
<span class="sourceLineNo">318</span> // output to live cluster<a name="line.318"></a>
<span class="sourceLineNo">319</span> job.setMapperClass(WALMapper.class);<a name="line.319"></a>
<span class="sourceLineNo">320</span> job.setOutputFormatClass(MultiTableOutputFormat.class);<a name="line.320"></a>
<span class="sourceLineNo">321</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.321"></a>
<span class="sourceLineNo">322</span> TableMapReduceUtil.initCredentials(job);<a name="line.322"></a>
<span class="sourceLineNo">323</span> // No reducers.<a name="line.323"></a>
<span class="sourceLineNo">324</span> job.setNumReduceTasks(0);<a name="line.324"></a>
<span class="sourceLineNo">325</span> }<a name="line.325"></a>
<span class="sourceLineNo">326</span> String codecCls = WALCellCodec.getWALCellCodecClass(conf).getName();<a name="line.326"></a>
<span class="sourceLineNo">327</span> try {<a name="line.327"></a>
<span class="sourceLineNo">328</span> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),<a name="line.328"></a>
<span class="sourceLineNo">329</span> Class.forName(codecCls));<a name="line.329"></a>
<span class="sourceLineNo">330</span> } catch (Exception e) {<a name="line.330"></a>
<span class="sourceLineNo">331</span> throw new IOException("Cannot determine wal codec class " + codecCls, e);<a name="line.331"></a>
<span class="sourceLineNo">332</span> }<a name="line.332"></a>
<span class="sourceLineNo">333</span> return job;<a name="line.333"></a>
<span class="sourceLineNo">334</span> }<a name="line.334"></a>
<span class="sourceLineNo">335</span><a name="line.335"></a>
<span class="sourceLineNo">336</span> private List&lt;TableName&gt; getTableNameList(String[] tables) {<a name="line.336"></a>
<span class="sourceLineNo">337</span> List&lt;TableName&gt; list = new ArrayList&lt;TableName&gt;();<a name="line.337"></a>
<span class="sourceLineNo">338</span> for (String name : tables) {<a name="line.338"></a>
<span class="sourceLineNo">339</span> list.add(TableName.valueOf(name));<a name="line.339"></a>
<span class="sourceLineNo">340</span> }<a name="line.340"></a>
<span class="sourceLineNo">341</span> return list;<a name="line.341"></a>
<span class="sourceLineNo">342</span> }<a name="line.342"></a>
<span class="sourceLineNo">343</span><a name="line.343"></a>
<span class="sourceLineNo">344</span> /**<a name="line.344"></a>
<span class="sourceLineNo">345</span> * Print usage<a name="line.345"></a>
<span class="sourceLineNo">346</span> * @param errorMsg Error message. Can be null.<a name="line.346"></a>
<span class="sourceLineNo">347</span> */<a name="line.347"></a>
<span class="sourceLineNo">348</span> private void usage(final String errorMsg) {<a name="line.348"></a>
<span class="sourceLineNo">349</span> if (errorMsg != null &amp;&amp; errorMsg.length() &gt; 0) {<a name="line.349"></a>
<span class="sourceLineNo">350</span> System.err.println("ERROR: " + errorMsg);<a name="line.350"></a>
<span class="sourceLineNo">351</span> }<a name="line.351"></a>
<span class="sourceLineNo">352</span> System.err.println("Usage: " + NAME + " [options] &lt;wal inputdir&gt; &lt;tables&gt; [&lt;tableMappings&gt;]");<a name="line.352"></a>
<span class="sourceLineNo">353</span> System.err.println("Replay all WAL files into HBase.");<a name="line.353"></a>
<span class="sourceLineNo">354</span> System.err.println("&lt;tables&gt; is a comma separated list of tables.");<a name="line.354"></a>
<span class="sourceLineNo">355</span> System.err.println("If no tables (\"\") are specified, all tables are imported.");<a name="line.355"></a>
<span class="sourceLineNo">356</span> System.err.println("(Be careful, hbase:meta entries will be imported in this case.)\n");<a name="line.356"></a>
<span class="sourceLineNo">357</span> System.err.println("WAL entries can be mapped to new set of tables via &lt;tableMappings&gt;.");<a name="line.357"></a>
<span class="sourceLineNo">358</span> System.err.println("&lt;tableMappings&gt; is a comma separated list of target tables.");<a name="line.358"></a>
<span class="sourceLineNo">359</span> System.err.println("If specified, each table in &lt;tables&gt; must have a mapping.\n");<a name="line.359"></a>
<span class="sourceLineNo">360</span> System.err.println("By default " + NAME + " will load data directly into HBase.");<a name="line.360"></a>
<span class="sourceLineNo">361</span> System.err.println("To generate HFiles for a bulk data load instead, pass the following option:");<a name="line.361"></a>
<span class="sourceLineNo">362</span> System.err.println(" -D" + BULK_OUTPUT_CONF_KEY + "=/path/for/output");<a name="line.362"></a>
<span class="sourceLineNo">363</span> System.err.println(" (Only one table can be specified, and no mapping is allowed!)");<a name="line.363"></a>
<span class="sourceLineNo">364</span> System.err.println("Time range options:");<a name="line.364"></a>
<span class="sourceLineNo">365</span> System.err.println(" -D" + WALInputFormat.START_TIME_KEY + "=[date|ms]");<a name="line.365"></a>
<span class="sourceLineNo">366</span> System.err.println(" -D" + WALInputFormat.END_TIME_KEY + "=[date|ms]");<a name="line.366"></a>
<span class="sourceLineNo">367</span> System.err.println(" (The start and the end date of timerange. The dates can be expressed");<a name="line.367"></a>
<span class="sourceLineNo">368</span> System.err.println(" in milliseconds since epoch or in yyyy-MM-dd'T'HH:mm:ss.SS format.");<a name="line.368"></a>
<span class="sourceLineNo">369</span> System.err.println(" E.g. 1234567890120 or 2009-02-13T23:32:30.12)");<a name="line.369"></a>
<span class="sourceLineNo">370</span> System.err.println("Other options:");<a name="line.370"></a>
<span class="sourceLineNo">371</span> System.err.println(" -D" + JOB_NAME_CONF_KEY + "=jobName");<a name="line.371"></a>
<span class="sourceLineNo">372</span> System.err.println(" Use the specified mapreduce job name for the wal player");<a name="line.372"></a>
<span class="sourceLineNo">373</span> System.err.println("For performance also consider the following options:\n"<a name="line.373"></a>
<span class="sourceLineNo">374</span> + " -Dmapreduce.map.speculative=false\n"<a name="line.374"></a>
<span class="sourceLineNo">375</span> + " -Dmapreduce.reduce.speculative=false");<a name="line.375"></a>
<span class="sourceLineNo">376</span> }<a name="line.376"></a>
<span class="sourceLineNo">377</span><a name="line.377"></a>
<span class="sourceLineNo">378</span> /**<a name="line.378"></a>
<span class="sourceLineNo">379</span> * Main entry point.<a name="line.379"></a>
<span class="sourceLineNo">380</span> * @param args The command line parameters.<a name="line.380"></a>
<span class="sourceLineNo">381</span> * @throws Exception When running the job fails.<a name="line.381"></a>
<span class="sourceLineNo">382</span> */<a name="line.382"></a>
<span class="sourceLineNo">383</span> public static void main(String[] args) throws Exception {<a name="line.383"></a>
<span class="sourceLineNo">384</span> int ret = ToolRunner.run(new WALPlayer(HBaseConfiguration.create()), args);<a name="line.384"></a>
<span class="sourceLineNo">385</span> System.exit(ret);<a name="line.385"></a>
<span class="sourceLineNo">386</span> }<a name="line.386"></a>
<span class="sourceLineNo">387</span><a name="line.387"></a>
<span class="sourceLineNo">388</span> @Override<a name="line.388"></a>
<span class="sourceLineNo">389</span> public int run(String[] args) throws Exception {<a name="line.389"></a>
<span class="sourceLineNo">390</span> if (args.length &lt; 2) {<a name="line.390"></a>
<span class="sourceLineNo">391</span> usage("Wrong number of arguments: " + args.length);<a name="line.391"></a>
<span class="sourceLineNo">392</span> System.exit(-1);<a name="line.392"></a>
<span class="sourceLineNo">393</span> }<a name="line.393"></a>
<span class="sourceLineNo">394</span> Job job = createSubmittableJob(args);<a name="line.394"></a>
<span class="sourceLineNo">395</span> return job.waitForCompletion(true) ? 0 : 1;<a name="line.395"></a>
<span class="sourceLineNo">396</span> }<a name="line.396"></a>
<span class="sourceLineNo">397</span>}<a name="line.397"></a>
</pre>
</div>
</body>
</html>