blob: c00855e6a5fbdd3345dc2a980190cac867f22286 [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.backup.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.lang.reflect.Field;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.lang.reflect.InvocationTargetException;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.lang.reflect.Method;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.math.BigDecimal;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.Arrays;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.List;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.util.Objects;<a name="line.27"></a>
<span class="sourceLineNo">028</span><a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.conf.Configuration;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.fs.FileStatus;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.FileSystem;<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.TableName;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.backup.BackupCopyJob;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.backup.BackupInfo;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.backup.BackupType;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.backup.impl.BackupManager;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.backup.util.BackupUtils;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.snapshot.ExportSnapshot;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.io.SequenceFile;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.io.Text;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.Cluster;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.mapreduce.Job;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.mapreduce.JobID;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.tools.CopyListingFileStatus;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.tools.DistCp;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.tools.DistCpConstants;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.tools.DistCpOptions;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.zookeeper.KeeperException.NoNodeException;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.slf4j.Logger;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.slf4j.LoggerFactory;<a name="line.53"></a>
<span class="sourceLineNo">054</span><a name="line.54"></a>
<span class="sourceLineNo">055</span>/**<a name="line.55"></a>
<span class="sourceLineNo">056</span> * Map-Reduce implementation of {@link BackupCopyJob}. Basically, there are 2 types of copy<a name="line.56"></a>
<span class="sourceLineNo">057</span> * operation: one is copying from snapshot, which bases on extending ExportSnapshot's function, the<a name="line.57"></a>
<span class="sourceLineNo">058</span> * other is copying for incremental log files, which bases on extending DistCp's function.<a name="line.58"></a>
<span class="sourceLineNo">059</span> */<a name="line.59"></a>
<span class="sourceLineNo">060</span>@InterfaceAudience.Private<a name="line.60"></a>
<span class="sourceLineNo">061</span>public class MapReduceBackupCopyJob implements BackupCopyJob {<a name="line.61"></a>
<span class="sourceLineNo">062</span> public static final String NUMBER_OF_LEVELS_TO_PRESERVE_KEY = "num.levels.preserve";<a name="line.62"></a>
<span class="sourceLineNo">063</span> private static final Logger LOG = LoggerFactory.getLogger(MapReduceBackupCopyJob.class);<a name="line.63"></a>
<span class="sourceLineNo">064</span><a name="line.64"></a>
<span class="sourceLineNo">065</span> private Configuration conf;<a name="line.65"></a>
<span class="sourceLineNo">066</span><a name="line.66"></a>
<span class="sourceLineNo">067</span> // Accumulated progress within the whole backup process for the copy operation<a name="line.67"></a>
<span class="sourceLineNo">068</span> private float progressDone = 0.1f;<a name="line.68"></a>
<span class="sourceLineNo">069</span> private long bytesCopied = 0;<a name="line.69"></a>
<span class="sourceLineNo">070</span> private static float INIT_PROGRESS = 0.1f;<a name="line.70"></a>
<span class="sourceLineNo">071</span><a name="line.71"></a>
<span class="sourceLineNo">072</span> // The percentage of the current copy task within the whole task if multiple time copies are<a name="line.72"></a>
<span class="sourceLineNo">073</span> // needed. The default value is 100%, which means only 1 copy task for the whole.<a name="line.73"></a>
<span class="sourceLineNo">074</span> private float subTaskPercntgInWholeTask = 1f;<a name="line.74"></a>
<span class="sourceLineNo">075</span><a name="line.75"></a>
<span class="sourceLineNo">076</span> public MapReduceBackupCopyJob() {<a name="line.76"></a>
<span class="sourceLineNo">077</span> }<a name="line.77"></a>
<span class="sourceLineNo">078</span><a name="line.78"></a>
<span class="sourceLineNo">079</span> @Override<a name="line.79"></a>
<span class="sourceLineNo">080</span> public Configuration getConf() {<a name="line.80"></a>
<span class="sourceLineNo">081</span> return conf;<a name="line.81"></a>
<span class="sourceLineNo">082</span> }<a name="line.82"></a>
<span class="sourceLineNo">083</span><a name="line.83"></a>
<span class="sourceLineNo">084</span> @Override<a name="line.84"></a>
<span class="sourceLineNo">085</span> public void setConf(Configuration conf) {<a name="line.85"></a>
<span class="sourceLineNo">086</span> this.conf = conf;<a name="line.86"></a>
<span class="sourceLineNo">087</span> }<a name="line.87"></a>
<span class="sourceLineNo">088</span><a name="line.88"></a>
<span class="sourceLineNo">089</span> /**<a name="line.89"></a>
<span class="sourceLineNo">090</span> * Get the current copy task percentage within the whole task if multiple copies are needed.<a name="line.90"></a>
<span class="sourceLineNo">091</span> * @return the current copy task percentage<a name="line.91"></a>
<span class="sourceLineNo">092</span> */<a name="line.92"></a>
<span class="sourceLineNo">093</span> public float getSubTaskPercntgInWholeTask() {<a name="line.93"></a>
<span class="sourceLineNo">094</span> return subTaskPercntgInWholeTask;<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> /**<a name="line.97"></a>
<span class="sourceLineNo">098</span> * Set the current copy task percentage within the whole task if multiple copies are needed. Must<a name="line.98"></a>
<span class="sourceLineNo">099</span> * be called before calling<a name="line.99"></a>
<span class="sourceLineNo">100</span> * {@link #copy(BackupInfo, BackupManager, Configuration, BackupType, String[])}<a name="line.100"></a>
<span class="sourceLineNo">101</span> * @param subTaskPercntgInWholeTask The percentage of the copy subtask<a name="line.101"></a>
<span class="sourceLineNo">102</span> */<a name="line.102"></a>
<span class="sourceLineNo">103</span> public void setSubTaskPercntgInWholeTask(float subTaskPercntgInWholeTask) {<a name="line.103"></a>
<span class="sourceLineNo">104</span> this.subTaskPercntgInWholeTask = subTaskPercntgInWholeTask;<a name="line.104"></a>
<span class="sourceLineNo">105</span> }<a name="line.105"></a>
<span class="sourceLineNo">106</span><a name="line.106"></a>
<span class="sourceLineNo">107</span> static class SnapshotCopy extends ExportSnapshot {<a name="line.107"></a>
<span class="sourceLineNo">108</span> private BackupInfo backupInfo;<a name="line.108"></a>
<span class="sourceLineNo">109</span> private TableName table;<a name="line.109"></a>
<span class="sourceLineNo">110</span><a name="line.110"></a>
<span class="sourceLineNo">111</span> public SnapshotCopy(BackupInfo backupInfo, TableName table) {<a name="line.111"></a>
<span class="sourceLineNo">112</span> super();<a name="line.112"></a>
<span class="sourceLineNo">113</span> this.backupInfo = backupInfo;<a name="line.113"></a>
<span class="sourceLineNo">114</span> this.table = table;<a name="line.114"></a>
<span class="sourceLineNo">115</span> }<a name="line.115"></a>
<span class="sourceLineNo">116</span><a name="line.116"></a>
<span class="sourceLineNo">117</span> public TableName getTable() {<a name="line.117"></a>
<span class="sourceLineNo">118</span> return this.table;<a name="line.118"></a>
<span class="sourceLineNo">119</span> }<a name="line.119"></a>
<span class="sourceLineNo">120</span><a name="line.120"></a>
<span class="sourceLineNo">121</span> public BackupInfo getBackupInfo() {<a name="line.121"></a>
<span class="sourceLineNo">122</span> return this.backupInfo;<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><a name="line.125"></a>
<span class="sourceLineNo">126</span> /**<a name="line.126"></a>
<span class="sourceLineNo">127</span> * Update the ongoing backup with new progress.<a name="line.127"></a>
<span class="sourceLineNo">128</span> * @param backupInfo backup info<a name="line.128"></a>
<span class="sourceLineNo">129</span> * @param newProgress progress<a name="line.129"></a>
<span class="sourceLineNo">130</span> * @param bytesCopied bytes copied<a name="line.130"></a>
<span class="sourceLineNo">131</span> * @throws NoNodeException exception<a name="line.131"></a>
<span class="sourceLineNo">132</span> */<a name="line.132"></a>
<span class="sourceLineNo">133</span> static void updateProgress(BackupInfo backupInfo, BackupManager backupManager,<a name="line.133"></a>
<span class="sourceLineNo">134</span> int newProgress, long bytesCopied) throws IOException {<a name="line.134"></a>
<span class="sourceLineNo">135</span> // compose the new backup progress data, using fake number for now<a name="line.135"></a>
<span class="sourceLineNo">136</span> String backupProgressData = newProgress + "%";<a name="line.136"></a>
<span class="sourceLineNo">137</span><a name="line.137"></a>
<span class="sourceLineNo">138</span> backupInfo.setProgress(newProgress);<a name="line.138"></a>
<span class="sourceLineNo">139</span> backupManager.updateBackupInfo(backupInfo);<a name="line.139"></a>
<span class="sourceLineNo">140</span> LOG.debug("Backup progress data \"" + backupProgressData<a name="line.140"></a>
<span class="sourceLineNo">141</span> + "\" has been updated to backup system table for " + backupInfo.getBackupId());<a name="line.141"></a>
<span class="sourceLineNo">142</span> }<a name="line.142"></a>
<span class="sourceLineNo">143</span><a name="line.143"></a>
<span class="sourceLineNo">144</span> /**<a name="line.144"></a>
<span class="sourceLineNo">145</span> * Extends DistCp for progress updating to backup system table<a name="line.145"></a>
<span class="sourceLineNo">146</span> * during backup. Using DistCpV2 (MAPREDUCE-2765).<a name="line.146"></a>
<span class="sourceLineNo">147</span> * Simply extend it and override execute() method to get the<a name="line.147"></a>
<span class="sourceLineNo">148</span> * Job reference for progress updating.<a name="line.148"></a>
<span class="sourceLineNo">149</span> * Only the argument "src1, [src2, [...]] dst" is supported,<a name="line.149"></a>
<span class="sourceLineNo">150</span> * no more DistCp options.<a name="line.150"></a>
<span class="sourceLineNo">151</span> */<a name="line.151"></a>
<span class="sourceLineNo">152</span><a name="line.152"></a>
<span class="sourceLineNo">153</span> class BackupDistCp extends DistCp {<a name="line.153"></a>
<span class="sourceLineNo">154</span><a name="line.154"></a>
<span class="sourceLineNo">155</span> private BackupInfo backupInfo;<a name="line.155"></a>
<span class="sourceLineNo">156</span> private BackupManager backupManager;<a name="line.156"></a>
<span class="sourceLineNo">157</span><a name="line.157"></a>
<span class="sourceLineNo">158</span> public BackupDistCp(Configuration conf, DistCpOptions options, BackupInfo backupInfo,<a name="line.158"></a>
<span class="sourceLineNo">159</span> BackupManager backupManager) throws Exception {<a name="line.159"></a>
<span class="sourceLineNo">160</span> super(conf, options);<a name="line.160"></a>
<span class="sourceLineNo">161</span> this.backupInfo = backupInfo;<a name="line.161"></a>
<span class="sourceLineNo">162</span> this.backupManager = backupManager;<a name="line.162"></a>
<span class="sourceLineNo">163</span> }<a name="line.163"></a>
<span class="sourceLineNo">164</span><a name="line.164"></a>
<span class="sourceLineNo">165</span><a name="line.165"></a>
<span class="sourceLineNo">166</span><a name="line.166"></a>
<span class="sourceLineNo">167</span> @Override<a name="line.167"></a>
<span class="sourceLineNo">168</span> public Job execute() throws Exception {<a name="line.168"></a>
<span class="sourceLineNo">169</span><a name="line.169"></a>
<span class="sourceLineNo">170</span> // reflection preparation for private methods and fields<a name="line.170"></a>
<span class="sourceLineNo">171</span> Class&lt;?&gt; classDistCp = org.apache.hadoop.tools.DistCp.class;<a name="line.171"></a>
<span class="sourceLineNo">172</span> Method methodCleanup = classDistCp.getDeclaredMethod("cleanup");<a name="line.172"></a>
<span class="sourceLineNo">173</span><a name="line.173"></a>
<span class="sourceLineNo">174</span> Field fieldInputOptions = getInputOptionsField(classDistCp);<a name="line.174"></a>
<span class="sourceLineNo">175</span> Field fieldSubmitted = classDistCp.getDeclaredField("submitted");<a name="line.175"></a>
<span class="sourceLineNo">176</span><a name="line.176"></a>
<span class="sourceLineNo">177</span> methodCleanup.setAccessible(true);<a name="line.177"></a>
<span class="sourceLineNo">178</span> fieldInputOptions.setAccessible(true);<a name="line.178"></a>
<span class="sourceLineNo">179</span> fieldSubmitted.setAccessible(true);<a name="line.179"></a>
<span class="sourceLineNo">180</span><a name="line.180"></a>
<span class="sourceLineNo">181</span> // execute() logic starts here<a name="line.181"></a>
<span class="sourceLineNo">182</span> assert fieldInputOptions.get(this) != null;<a name="line.182"></a>
<span class="sourceLineNo">183</span><a name="line.183"></a>
<span class="sourceLineNo">184</span> Job job = null;<a name="line.184"></a>
<span class="sourceLineNo">185</span> try {<a name="line.185"></a>
<span class="sourceLineNo">186</span><a name="line.186"></a>
<span class="sourceLineNo">187</span> List&lt;Path&gt; srcs = getSourcePaths(fieldInputOptions);<a name="line.187"></a>
<span class="sourceLineNo">188</span><a name="line.188"></a>
<span class="sourceLineNo">189</span> long totalSrcLgth = 0;<a name="line.189"></a>
<span class="sourceLineNo">190</span> for (Path aSrc : srcs) {<a name="line.190"></a>
<span class="sourceLineNo">191</span> totalSrcLgth +=<a name="line.191"></a>
<span class="sourceLineNo">192</span> BackupUtils.getFilesLength(aSrc.getFileSystem(super.getConf()), aSrc);<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
<span class="sourceLineNo">194</span><a name="line.194"></a>
<span class="sourceLineNo">195</span> // Async call<a name="line.195"></a>
<span class="sourceLineNo">196</span> job = super.execute();<a name="line.196"></a>
<span class="sourceLineNo">197</span> // Update the copy progress to system table every 0.5s if progress value changed<a name="line.197"></a>
<span class="sourceLineNo">198</span> int progressReportFreq =<a name="line.198"></a>
<span class="sourceLineNo">199</span> MapReduceBackupCopyJob.this.getConf().getInt("hbase.backup.progressreport.frequency",<a name="line.199"></a>
<span class="sourceLineNo">200</span> 500);<a name="line.200"></a>
<span class="sourceLineNo">201</span> float lastProgress = progressDone;<a name="line.201"></a>
<span class="sourceLineNo">202</span> while (!job.isComplete()) {<a name="line.202"></a>
<span class="sourceLineNo">203</span> float newProgress =<a name="line.203"></a>
<span class="sourceLineNo">204</span> progressDone + job.mapProgress() * subTaskPercntgInWholeTask * (1 - INIT_PROGRESS);<a name="line.204"></a>
<span class="sourceLineNo">205</span><a name="line.205"></a>
<span class="sourceLineNo">206</span> if (newProgress &gt; lastProgress) {<a name="line.206"></a>
<span class="sourceLineNo">207</span><a name="line.207"></a>
<span class="sourceLineNo">208</span> BigDecimal progressData =<a name="line.208"></a>
<span class="sourceLineNo">209</span> new BigDecimal(newProgress * 100).setScale(1, BigDecimal.ROUND_HALF_UP);<a name="line.209"></a>
<span class="sourceLineNo">210</span> String newProgressStr = progressData + "%";<a name="line.210"></a>
<span class="sourceLineNo">211</span> LOG.info("Progress: " + newProgressStr);<a name="line.211"></a>
<span class="sourceLineNo">212</span> updateProgress(backupInfo, backupManager, progressData.intValue(), bytesCopied);<a name="line.212"></a>
<span class="sourceLineNo">213</span> LOG.debug("Backup progress data updated to backup system table: \"Progress: "<a name="line.213"></a>
<span class="sourceLineNo">214</span> + newProgressStr + ".\"");<a name="line.214"></a>
<span class="sourceLineNo">215</span> lastProgress = newProgress;<a name="line.215"></a>
<span class="sourceLineNo">216</span> }<a name="line.216"></a>
<span class="sourceLineNo">217</span> Thread.sleep(progressReportFreq);<a name="line.217"></a>
<span class="sourceLineNo">218</span> }<a name="line.218"></a>
<span class="sourceLineNo">219</span> // update the progress data after copy job complete<a name="line.219"></a>
<span class="sourceLineNo">220</span> float newProgress =<a name="line.220"></a>
<span class="sourceLineNo">221</span> progressDone + job.mapProgress() * subTaskPercntgInWholeTask * (1 - INIT_PROGRESS);<a name="line.221"></a>
<span class="sourceLineNo">222</span> BigDecimal progressData =<a name="line.222"></a>
<span class="sourceLineNo">223</span> new BigDecimal(newProgress * 100).setScale(1, BigDecimal.ROUND_HALF_UP);<a name="line.223"></a>
<span class="sourceLineNo">224</span><a name="line.224"></a>
<span class="sourceLineNo">225</span> String newProgressStr = progressData + "%";<a name="line.225"></a>
<span class="sourceLineNo">226</span> LOG.info("Progress: " + newProgressStr + " subTask: " + subTaskPercntgInWholeTask<a name="line.226"></a>
<span class="sourceLineNo">227</span> + " mapProgress: " + job.mapProgress());<a name="line.227"></a>
<span class="sourceLineNo">228</span><a name="line.228"></a>
<span class="sourceLineNo">229</span> // accumulate the overall backup progress<a name="line.229"></a>
<span class="sourceLineNo">230</span> progressDone = newProgress;<a name="line.230"></a>
<span class="sourceLineNo">231</span> bytesCopied += totalSrcLgth;<a name="line.231"></a>
<span class="sourceLineNo">232</span><a name="line.232"></a>
<span class="sourceLineNo">233</span> updateProgress(backupInfo, backupManager, progressData.intValue(), bytesCopied);<a name="line.233"></a>
<span class="sourceLineNo">234</span> LOG.debug("Backup progress data updated to backup system table: \"Progress: "<a name="line.234"></a>
<span class="sourceLineNo">235</span> + newProgressStr + " - " + bytesCopied + " bytes copied.\"");<a name="line.235"></a>
<span class="sourceLineNo">236</span> } catch (Throwable t) {<a name="line.236"></a>
<span class="sourceLineNo">237</span> LOG.error(t.toString(), t);<a name="line.237"></a>
<span class="sourceLineNo">238</span> throw t;<a name="line.238"></a>
<span class="sourceLineNo">239</span> }<a name="line.239"></a>
<span class="sourceLineNo">240</span><a name="line.240"></a>
<span class="sourceLineNo">241</span> String jobID = job.getJobID().toString();<a name="line.241"></a>
<span class="sourceLineNo">242</span> job.getConfiguration().set(DistCpConstants.CONF_LABEL_DISTCP_JOB_ID, jobID);<a name="line.242"></a>
<span class="sourceLineNo">243</span><a name="line.243"></a>
<span class="sourceLineNo">244</span> LOG.debug("DistCp job-id: " + jobID + " completed: " + job.isComplete() + " "<a name="line.244"></a>
<span class="sourceLineNo">245</span> + job.isSuccessful());<a name="line.245"></a>
<span class="sourceLineNo">246</span> Counters ctrs = job.getCounters();<a name="line.246"></a>
<span class="sourceLineNo">247</span> LOG.debug(Objects.toString(ctrs));<a name="line.247"></a>
<span class="sourceLineNo">248</span> if (job.isComplete() &amp;&amp; !job.isSuccessful()) {<a name="line.248"></a>
<span class="sourceLineNo">249</span> throw new Exception("DistCp job-id: " + jobID + " failed");<a name="line.249"></a>
<span class="sourceLineNo">250</span> }<a name="line.250"></a>
<span class="sourceLineNo">251</span><a name="line.251"></a>
<span class="sourceLineNo">252</span> return job;<a name="line.252"></a>
<span class="sourceLineNo">253</span> }<a name="line.253"></a>
<span class="sourceLineNo">254</span><a name="line.254"></a>
<span class="sourceLineNo">255</span> private Field getInputOptionsField(Class&lt;?&gt; classDistCp) throws IOException{<a name="line.255"></a>
<span class="sourceLineNo">256</span> Field f = null;<a name="line.256"></a>
<span class="sourceLineNo">257</span> try {<a name="line.257"></a>
<span class="sourceLineNo">258</span> f = classDistCp.getDeclaredField("inputOptions");<a name="line.258"></a>
<span class="sourceLineNo">259</span> } catch(Exception e) {<a name="line.259"></a>
<span class="sourceLineNo">260</span> // Haddop 3<a name="line.260"></a>
<span class="sourceLineNo">261</span> try {<a name="line.261"></a>
<span class="sourceLineNo">262</span> f = classDistCp.getDeclaredField("context");<a name="line.262"></a>
<span class="sourceLineNo">263</span> } catch (NoSuchFieldException | SecurityException e1) {<a name="line.263"></a>
<span class="sourceLineNo">264</span> throw new IOException(e1);<a name="line.264"></a>
<span class="sourceLineNo">265</span> }<a name="line.265"></a>
<span class="sourceLineNo">266</span> }<a name="line.266"></a>
<span class="sourceLineNo">267</span> return f;<a name="line.267"></a>
<span class="sourceLineNo">268</span> }<a name="line.268"></a>
<span class="sourceLineNo">269</span><a name="line.269"></a>
<span class="sourceLineNo">270</span> @SuppressWarnings("unchecked")<a name="line.270"></a>
<span class="sourceLineNo">271</span> private List&lt;Path&gt; getSourcePaths(Field fieldInputOptions) throws IOException{<a name="line.271"></a>
<span class="sourceLineNo">272</span> Object options;<a name="line.272"></a>
<span class="sourceLineNo">273</span> try {<a name="line.273"></a>
<span class="sourceLineNo">274</span> options = fieldInputOptions.get(this);<a name="line.274"></a>
<span class="sourceLineNo">275</span> if (options instanceof DistCpOptions) {<a name="line.275"></a>
<span class="sourceLineNo">276</span> return ((DistCpOptions) options).getSourcePaths();<a name="line.276"></a>
<span class="sourceLineNo">277</span> } else {<a name="line.277"></a>
<span class="sourceLineNo">278</span> // Hadoop 3<a name="line.278"></a>
<span class="sourceLineNo">279</span> Class&lt;?&gt; classContext = Class.forName("org.apache.hadoop.tools.DistCpContext");<a name="line.279"></a>
<span class="sourceLineNo">280</span> Method methodGetSourcePaths = classContext.getDeclaredMethod("getSourcePaths");<a name="line.280"></a>
<span class="sourceLineNo">281</span> methodGetSourcePaths.setAccessible(true);<a name="line.281"></a>
<span class="sourceLineNo">282</span><a name="line.282"></a>
<span class="sourceLineNo">283</span> return (List&lt;Path&gt;) methodGetSourcePaths.invoke(options);<a name="line.283"></a>
<span class="sourceLineNo">284</span> }<a name="line.284"></a>
<span class="sourceLineNo">285</span> } catch (IllegalArgumentException | IllegalAccessException |<a name="line.285"></a>
<span class="sourceLineNo">286</span> ClassNotFoundException | NoSuchMethodException |<a name="line.286"></a>
<span class="sourceLineNo">287</span> SecurityException | InvocationTargetException e) {<a name="line.287"></a>
<span class="sourceLineNo">288</span> throw new IOException(e);<a name="line.288"></a>
<span class="sourceLineNo">289</span> }<a name="line.289"></a>
<span class="sourceLineNo">290</span><a name="line.290"></a>
<span class="sourceLineNo">291</span> }<a name="line.291"></a>
<span class="sourceLineNo">292</span><a name="line.292"></a>
<span class="sourceLineNo">293</span> @Override<a name="line.293"></a>
<span class="sourceLineNo">294</span> protected Path createInputFileListing(Job job) throws IOException {<a name="line.294"></a>
<span class="sourceLineNo">295</span><a name="line.295"></a>
<span class="sourceLineNo">296</span> if (conf.get(NUMBER_OF_LEVELS_TO_PRESERVE_KEY) == null) {<a name="line.296"></a>
<span class="sourceLineNo">297</span> return super.createInputFileListing(job);<a name="line.297"></a>
<span class="sourceLineNo">298</span> }<a name="line.298"></a>
<span class="sourceLineNo">299</span> long totalBytesExpected = 0;<a name="line.299"></a>
<span class="sourceLineNo">300</span> int totalRecords = 0;<a name="line.300"></a>
<span class="sourceLineNo">301</span> Path fileListingPath = getFileListingPath();<a name="line.301"></a>
<span class="sourceLineNo">302</span> try (SequenceFile.Writer writer = getWriter(fileListingPath)) {<a name="line.302"></a>
<span class="sourceLineNo">303</span> List&lt;Path&gt; srcFiles = getSourceFiles();<a name="line.303"></a>
<span class="sourceLineNo">304</span> if (srcFiles.size() == 0) {<a name="line.304"></a>
<span class="sourceLineNo">305</span> return fileListingPath;<a name="line.305"></a>
<span class="sourceLineNo">306</span> }<a name="line.306"></a>
<span class="sourceLineNo">307</span> totalRecords = srcFiles.size();<a name="line.307"></a>
<span class="sourceLineNo">308</span> FileSystem fs = srcFiles.get(0).getFileSystem(conf);<a name="line.308"></a>
<span class="sourceLineNo">309</span> for (Path path : srcFiles) {<a name="line.309"></a>
<span class="sourceLineNo">310</span> FileStatus fst = fs.getFileStatus(path);<a name="line.310"></a>
<span class="sourceLineNo">311</span> totalBytesExpected += fst.getLen();<a name="line.311"></a>
<span class="sourceLineNo">312</span> Text key = getKey(path);<a name="line.312"></a>
<span class="sourceLineNo">313</span> writer.append(key, new CopyListingFileStatus(fst));<a name="line.313"></a>
<span class="sourceLineNo">314</span> }<a name="line.314"></a>
<span class="sourceLineNo">315</span> writer.close();<a name="line.315"></a>
<span class="sourceLineNo">316</span><a name="line.316"></a>
<span class="sourceLineNo">317</span> // update jobs configuration<a name="line.317"></a>
<span class="sourceLineNo">318</span><a name="line.318"></a>
<span class="sourceLineNo">319</span> Configuration cfg = job.getConfiguration();<a name="line.319"></a>
<span class="sourceLineNo">320</span> cfg.setLong(DistCpConstants.CONF_LABEL_TOTAL_BYTES_TO_BE_COPIED, totalBytesExpected);<a name="line.320"></a>
<span class="sourceLineNo">321</span> cfg.set(DistCpConstants.CONF_LABEL_LISTING_FILE_PATH, fileListingPath.toString());<a name="line.321"></a>
<span class="sourceLineNo">322</span> cfg.setLong(DistCpConstants.CONF_LABEL_TOTAL_NUMBER_OF_RECORDS, totalRecords);<a name="line.322"></a>
<span class="sourceLineNo">323</span> } catch (NoSuchFieldException | SecurityException | IllegalArgumentException<a name="line.323"></a>
<span class="sourceLineNo">324</span> | IllegalAccessException | NoSuchMethodException | ClassNotFoundException<a name="line.324"></a>
<span class="sourceLineNo">325</span> | InvocationTargetException e) {<a name="line.325"></a>
<span class="sourceLineNo">326</span> throw new IOException(e);<a name="line.326"></a>
<span class="sourceLineNo">327</span> }<a name="line.327"></a>
<span class="sourceLineNo">328</span> return fileListingPath;<a name="line.328"></a>
<span class="sourceLineNo">329</span> }<a name="line.329"></a>
<span class="sourceLineNo">330</span><a name="line.330"></a>
<span class="sourceLineNo">331</span> private Text getKey(Path path) {<a name="line.331"></a>
<span class="sourceLineNo">332</span> int level = conf.getInt(NUMBER_OF_LEVELS_TO_PRESERVE_KEY, 1);<a name="line.332"></a>
<span class="sourceLineNo">333</span> int count = 0;<a name="line.333"></a>
<span class="sourceLineNo">334</span> String relPath = "";<a name="line.334"></a>
<span class="sourceLineNo">335</span> while (count++ &lt; level) {<a name="line.335"></a>
<span class="sourceLineNo">336</span> relPath = Path.SEPARATOR + path.getName() + relPath;<a name="line.336"></a>
<span class="sourceLineNo">337</span> path = path.getParent();<a name="line.337"></a>
<span class="sourceLineNo">338</span> }<a name="line.338"></a>
<span class="sourceLineNo">339</span> return new Text(relPath);<a name="line.339"></a>
<span class="sourceLineNo">340</span> }<a name="line.340"></a>
<span class="sourceLineNo">341</span><a name="line.341"></a>
<span class="sourceLineNo">342</span> private List&lt;Path&gt; getSourceFiles() throws NoSuchFieldException, SecurityException,<a name="line.342"></a>
<span class="sourceLineNo">343</span> IllegalArgumentException, IllegalAccessException, NoSuchMethodException,<a name="line.343"></a>
<span class="sourceLineNo">344</span> ClassNotFoundException, InvocationTargetException, IOException {<a name="line.344"></a>
<span class="sourceLineNo">345</span> Field options = null;<a name="line.345"></a>
<span class="sourceLineNo">346</span> try {<a name="line.346"></a>
<span class="sourceLineNo">347</span> options = DistCp.class.getDeclaredField("inputOptions");<a name="line.347"></a>
<span class="sourceLineNo">348</span> } catch (NoSuchFieldException | SecurityException e) {<a name="line.348"></a>
<span class="sourceLineNo">349</span> options = DistCp.class.getDeclaredField("context");<a name="line.349"></a>
<span class="sourceLineNo">350</span> }<a name="line.350"></a>
<span class="sourceLineNo">351</span> options.setAccessible(true);<a name="line.351"></a>
<span class="sourceLineNo">352</span> return getSourcePaths(options);<a name="line.352"></a>
<span class="sourceLineNo">353</span> }<a name="line.353"></a>
<span class="sourceLineNo">354</span><a name="line.354"></a>
<span class="sourceLineNo">355</span><a name="line.355"></a>
<span class="sourceLineNo">356</span><a name="line.356"></a>
<span class="sourceLineNo">357</span> private SequenceFile.Writer getWriter(Path pathToListFile) throws IOException {<a name="line.357"></a>
<span class="sourceLineNo">358</span> FileSystem fs = pathToListFile.getFileSystem(conf);<a name="line.358"></a>
<span class="sourceLineNo">359</span> fs.delete(pathToListFile, false);<a name="line.359"></a>
<span class="sourceLineNo">360</span> return SequenceFile.createWriter(conf, SequenceFile.Writer.file(pathToListFile),<a name="line.360"></a>
<span class="sourceLineNo">361</span> SequenceFile.Writer.keyClass(Text.class),<a name="line.361"></a>
<span class="sourceLineNo">362</span> SequenceFile.Writer.valueClass(CopyListingFileStatus.class),<a name="line.362"></a>
<span class="sourceLineNo">363</span> SequenceFile.Writer.compression(SequenceFile.CompressionType.NONE));<a name="line.363"></a>
<span class="sourceLineNo">364</span> }<a name="line.364"></a>
<span class="sourceLineNo">365</span><a name="line.365"></a>
<span class="sourceLineNo">366</span> }<a name="line.366"></a>
<span class="sourceLineNo">367</span><a name="line.367"></a>
<span class="sourceLineNo">368</span> /**<a name="line.368"></a>
<span class="sourceLineNo">369</span> * Do backup copy based on different types.<a name="line.369"></a>
<span class="sourceLineNo">370</span> * @param context The backup info<a name="line.370"></a>
<span class="sourceLineNo">371</span> * @param conf The hadoop configuration<a name="line.371"></a>
<span class="sourceLineNo">372</span> * @param copyType The backup copy type<a name="line.372"></a>
<span class="sourceLineNo">373</span> * @param options Options for customized ExportSnapshot or DistCp<a name="line.373"></a>
<span class="sourceLineNo">374</span> * @throws Exception exception<a name="line.374"></a>
<span class="sourceLineNo">375</span> */<a name="line.375"></a>
<span class="sourceLineNo">376</span> @Override<a name="line.376"></a>
<span class="sourceLineNo">377</span> public int copy(BackupInfo context, BackupManager backupManager, Configuration conf,<a name="line.377"></a>
<span class="sourceLineNo">378</span> BackupType copyType, String[] options) throws IOException {<a name="line.378"></a>
<span class="sourceLineNo">379</span> int res = 0;<a name="line.379"></a>
<span class="sourceLineNo">380</span><a name="line.380"></a>
<span class="sourceLineNo">381</span> try {<a name="line.381"></a>
<span class="sourceLineNo">382</span> if (copyType == BackupType.FULL) {<a name="line.382"></a>
<span class="sourceLineNo">383</span> SnapshotCopy snapshotCp = new SnapshotCopy(context, context.getTableBySnapshot(options[1]));<a name="line.383"></a>
<span class="sourceLineNo">384</span> LOG.debug("Doing SNAPSHOT_COPY");<a name="line.384"></a>
<span class="sourceLineNo">385</span> // Make a new instance of conf to be used by the snapshot copy class.<a name="line.385"></a>
<span class="sourceLineNo">386</span> snapshotCp.setConf(new Configuration(conf));<a name="line.386"></a>
<span class="sourceLineNo">387</span> res = snapshotCp.run(options);<a name="line.387"></a>
<span class="sourceLineNo">388</span><a name="line.388"></a>
<span class="sourceLineNo">389</span> } else if (copyType == BackupType.INCREMENTAL) {<a name="line.389"></a>
<span class="sourceLineNo">390</span> LOG.debug("Doing COPY_TYPE_DISTCP");<a name="line.390"></a>
<span class="sourceLineNo">391</span> setSubTaskPercntgInWholeTask(1f);<a name="line.391"></a>
<span class="sourceLineNo">392</span><a name="line.392"></a>
<span class="sourceLineNo">393</span> BackupDistCp distcp =<a name="line.393"></a>
<span class="sourceLineNo">394</span> new BackupDistCp(new Configuration(conf), null, context, backupManager);<a name="line.394"></a>
<span class="sourceLineNo">395</span> // Handle a special case where the source file is a single file.<a name="line.395"></a>
<span class="sourceLineNo">396</span> // In this case, distcp will not create the target dir. It just take the<a name="line.396"></a>
<span class="sourceLineNo">397</span> // target as a file name and copy source file to the target (as a file name).<a name="line.397"></a>
<span class="sourceLineNo">398</span> // We need to create the target dir before run distcp.<a name="line.398"></a>
<span class="sourceLineNo">399</span> LOG.debug("DistCp options: " + Arrays.toString(options));<a name="line.399"></a>
<span class="sourceLineNo">400</span> Path dest = new Path(options[options.length - 1]);<a name="line.400"></a>
<span class="sourceLineNo">401</span> String[] newOptions = new String[options.length + 1];<a name="line.401"></a>
<span class="sourceLineNo">402</span> System.arraycopy(options, 0, newOptions, 1, options.length);<a name="line.402"></a>
<span class="sourceLineNo">403</span> newOptions[0] = "-async"; // run DisCp in async mode<a name="line.403"></a>
<span class="sourceLineNo">404</span> FileSystem destfs = dest.getFileSystem(conf);<a name="line.404"></a>
<span class="sourceLineNo">405</span> if (!destfs.exists(dest)) {<a name="line.405"></a>
<span class="sourceLineNo">406</span> destfs.mkdirs(dest);<a name="line.406"></a>
<span class="sourceLineNo">407</span> }<a name="line.407"></a>
<span class="sourceLineNo">408</span> res = distcp.run(newOptions);<a name="line.408"></a>
<span class="sourceLineNo">409</span> }<a name="line.409"></a>
<span class="sourceLineNo">410</span> return res;<a name="line.410"></a>
<span class="sourceLineNo">411</span><a name="line.411"></a>
<span class="sourceLineNo">412</span> } catch (Exception e) {<a name="line.412"></a>
<span class="sourceLineNo">413</span> throw new IOException(e);<a name="line.413"></a>
<span class="sourceLineNo">414</span> }<a name="line.414"></a>
<span class="sourceLineNo">415</span> }<a name="line.415"></a>
<span class="sourceLineNo">416</span><a name="line.416"></a>
<span class="sourceLineNo">417</span> @Override<a name="line.417"></a>
<span class="sourceLineNo">418</span> public void cancel(String jobId) throws IOException {<a name="line.418"></a>
<span class="sourceLineNo">419</span> JobID id = JobID.forName(jobId);<a name="line.419"></a>
<span class="sourceLineNo">420</span> Cluster cluster = new Cluster(this.getConf());<a name="line.420"></a>
<span class="sourceLineNo">421</span> try {<a name="line.421"></a>
<span class="sourceLineNo">422</span> Job job = cluster.getJob(id);<a name="line.422"></a>
<span class="sourceLineNo">423</span> if (job == null) {<a name="line.423"></a>
<span class="sourceLineNo">424</span> LOG.error("No job found for " + id);<a name="line.424"></a>
<span class="sourceLineNo">425</span> // should we throw exception<a name="line.425"></a>
<span class="sourceLineNo">426</span> return;<a name="line.426"></a>
<span class="sourceLineNo">427</span> }<a name="line.427"></a>
<span class="sourceLineNo">428</span> if (job.isComplete() || job.isRetired()) {<a name="line.428"></a>
<span class="sourceLineNo">429</span> return;<a name="line.429"></a>
<span class="sourceLineNo">430</span> }<a name="line.430"></a>
<span class="sourceLineNo">431</span><a name="line.431"></a>
<span class="sourceLineNo">432</span> job.killJob();<a name="line.432"></a>
<span class="sourceLineNo">433</span> LOG.debug("Killed copy job " + id);<a name="line.433"></a>
<span class="sourceLineNo">434</span> } catch (InterruptedException e) {<a name="line.434"></a>
<span class="sourceLineNo">435</span> throw new IOException(e);<a name="line.435"></a>
<span class="sourceLineNo">436</span> }<a name="line.436"></a>
<span class="sourceLineNo">437</span> }<a name="line.437"></a>
<span class="sourceLineNo">438</span><a name="line.438"></a>
<span class="sourceLineNo">439</span>}<a name="line.439"></a>
</pre>
</div>
</body>
</html>