blob: 6cbe9f1c4f6fc3c39b27f843072211dcfac5092f [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.replication.regionserver;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.Closeable;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.FileNotFoundException;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.IOException;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.InterruptedIOException;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.math.BigInteger;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.security.SecureRandom;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.ArrayList;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.util.Deque;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import java.util.HashMap;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import java.util.LinkedList;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import java.util.List;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import java.util.Map;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import java.util.Map.Entry;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import java.util.concurrent.Callable;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import java.util.concurrent.ExecutionException;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import java.util.concurrent.Future;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import java.util.concurrent.ThreadPoolExecutor;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import java.util.concurrent.TimeUnit;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.conf.Configuration;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.fs.FileSystem;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.fs.FileUtil;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.fs.Path;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.fs.permission.FsPermission;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.HConstants;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.TableName;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.client.AsyncClusterConnection;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.security.User;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.security.UserProvider;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.security.token.FsDelegationToken;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.tool.BulkLoadHFiles;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.tool.BulkLoadHFiles.LoadQueueItem;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.tool.BulkLoadHFilesTool;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.apache.hadoop.hbase.util.CommonFSUtils;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.54"></a>
<span class="sourceLineNo">055</span>import org.apache.hadoop.hbase.util.Threads;<a name="line.55"></a>
<span class="sourceLineNo">056</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.56"></a>
<span class="sourceLineNo">057</span>import org.slf4j.Logger;<a name="line.57"></a>
<span class="sourceLineNo">058</span>import org.slf4j.LoggerFactory;<a name="line.58"></a>
<span class="sourceLineNo">059</span><a name="line.59"></a>
<span class="sourceLineNo">060</span>import org.apache.hbase.thirdparty.com.google.common.util.concurrent.ThreadFactoryBuilder;<a name="line.60"></a>
<span class="sourceLineNo">061</span><a name="line.61"></a>
<span class="sourceLineNo">062</span>/**<a name="line.62"></a>
<span class="sourceLineNo">063</span> * It is used for replicating HFile entries. It will first copy parallely all the hfiles to a local<a name="line.63"></a>
<span class="sourceLineNo">064</span> * staging directory and then it will use ({@link BulkLoadHFiles} to prepare a collection of<a name="line.64"></a>
<span class="sourceLineNo">065</span> * {@link LoadQueueItem} which will finally be loaded(replicated) into the table of this cluster.<a name="line.65"></a>
<span class="sourceLineNo">066</span> * Call {@link #close()} when done.<a name="line.66"></a>
<span class="sourceLineNo">067</span> */<a name="line.67"></a>
<span class="sourceLineNo">068</span>@InterfaceAudience.Private<a name="line.68"></a>
<span class="sourceLineNo">069</span>public class HFileReplicator implements Closeable {<a name="line.69"></a>
<span class="sourceLineNo">070</span> /** Maximum number of threads to allow in pool to copy hfiles during replication */<a name="line.70"></a>
<span class="sourceLineNo">071</span> public static final String REPLICATION_BULKLOAD_COPY_MAXTHREADS_KEY =<a name="line.71"></a>
<span class="sourceLineNo">072</span> "hbase.replication.bulkload.copy.maxthreads";<a name="line.72"></a>
<span class="sourceLineNo">073</span> public static final int REPLICATION_BULKLOAD_COPY_MAXTHREADS_DEFAULT = 10;<a name="line.73"></a>
<span class="sourceLineNo">074</span> /** Number of hfiles to copy per thread during replication */<a name="line.74"></a>
<span class="sourceLineNo">075</span> public static final String REPLICATION_BULKLOAD_COPY_HFILES_PERTHREAD_KEY =<a name="line.75"></a>
<span class="sourceLineNo">076</span> "hbase.replication.bulkload.copy.hfiles.perthread";<a name="line.76"></a>
<span class="sourceLineNo">077</span> public static final int REPLICATION_BULKLOAD_COPY_HFILES_PERTHREAD_DEFAULT = 10;<a name="line.77"></a>
<span class="sourceLineNo">078</span><a name="line.78"></a>
<span class="sourceLineNo">079</span> private static final Logger LOG = LoggerFactory.getLogger(HFileReplicator.class);<a name="line.79"></a>
<span class="sourceLineNo">080</span> private static final String UNDERSCORE = "_";<a name="line.80"></a>
<span class="sourceLineNo">081</span> private final static FsPermission PERM_ALL_ACCESS = FsPermission.valueOf("-rwxrwxrwx");<a name="line.81"></a>
<span class="sourceLineNo">082</span><a name="line.82"></a>
<span class="sourceLineNo">083</span> private Configuration sourceClusterConf;<a name="line.83"></a>
<span class="sourceLineNo">084</span> private String sourceBaseNamespaceDirPath;<a name="line.84"></a>
<span class="sourceLineNo">085</span> private String sourceHFileArchiveDirPath;<a name="line.85"></a>
<span class="sourceLineNo">086</span> private Map&lt;String, List&lt;Pair&lt;byte[], List&lt;String&gt;&gt;&gt;&gt; bulkLoadHFileMap;<a name="line.86"></a>
<span class="sourceLineNo">087</span> private FileSystem sinkFs;<a name="line.87"></a>
<span class="sourceLineNo">088</span> private FsDelegationToken fsDelegationToken;<a name="line.88"></a>
<span class="sourceLineNo">089</span> private UserProvider userProvider;<a name="line.89"></a>
<span class="sourceLineNo">090</span> private Configuration conf;<a name="line.90"></a>
<span class="sourceLineNo">091</span> private AsyncClusterConnection connection;<a name="line.91"></a>
<span class="sourceLineNo">092</span> private Path hbaseStagingDir;<a name="line.92"></a>
<span class="sourceLineNo">093</span> private ThreadPoolExecutor exec;<a name="line.93"></a>
<span class="sourceLineNo">094</span> private int maxCopyThreads;<a name="line.94"></a>
<span class="sourceLineNo">095</span> private int copiesPerThread;<a name="line.95"></a>
<span class="sourceLineNo">096</span> private List&lt;String&gt; sourceClusterIds;<a name="line.96"></a>
<span class="sourceLineNo">097</span><a name="line.97"></a>
<span class="sourceLineNo">098</span> public HFileReplicator(Configuration sourceClusterConf,<a name="line.98"></a>
<span class="sourceLineNo">099</span> String sourceBaseNamespaceDirPath, String sourceHFileArchiveDirPath,<a name="line.99"></a>
<span class="sourceLineNo">100</span> Map&lt;String, List&lt;Pair&lt;byte[], List&lt;String&gt;&gt;&gt;&gt; tableQueueMap, Configuration conf,<a name="line.100"></a>
<span class="sourceLineNo">101</span> AsyncClusterConnection connection, List&lt;String&gt; sourceClusterIds) throws IOException {<a name="line.101"></a>
<span class="sourceLineNo">102</span> this.sourceClusterConf = sourceClusterConf;<a name="line.102"></a>
<span class="sourceLineNo">103</span> this.sourceBaseNamespaceDirPath = sourceBaseNamespaceDirPath;<a name="line.103"></a>
<span class="sourceLineNo">104</span> this.sourceHFileArchiveDirPath = sourceHFileArchiveDirPath;<a name="line.104"></a>
<span class="sourceLineNo">105</span> this.bulkLoadHFileMap = tableQueueMap;<a name="line.105"></a>
<span class="sourceLineNo">106</span> this.conf = conf;<a name="line.106"></a>
<span class="sourceLineNo">107</span> this.connection = connection;<a name="line.107"></a>
<span class="sourceLineNo">108</span> this.sourceClusterIds = sourceClusterIds;<a name="line.108"></a>
<span class="sourceLineNo">109</span><a name="line.109"></a>
<span class="sourceLineNo">110</span> userProvider = UserProvider.instantiate(conf);<a name="line.110"></a>
<span class="sourceLineNo">111</span> fsDelegationToken = new FsDelegationToken(userProvider, "renewer");<a name="line.111"></a>
<span class="sourceLineNo">112</span> this.hbaseStagingDir =<a name="line.112"></a>
<span class="sourceLineNo">113</span> new Path(CommonFSUtils.getRootDir(conf), HConstants.BULKLOAD_STAGING_DIR_NAME);<a name="line.113"></a>
<span class="sourceLineNo">114</span> this.maxCopyThreads =<a name="line.114"></a>
<span class="sourceLineNo">115</span> this.conf.getInt(REPLICATION_BULKLOAD_COPY_MAXTHREADS_KEY,<a name="line.115"></a>
<span class="sourceLineNo">116</span> REPLICATION_BULKLOAD_COPY_MAXTHREADS_DEFAULT);<a name="line.116"></a>
<span class="sourceLineNo">117</span> this.exec = Threads.getBoundedCachedThreadPool(maxCopyThreads, 60, TimeUnit.SECONDS,<a name="line.117"></a>
<span class="sourceLineNo">118</span> new ThreadFactoryBuilder().setDaemon(true)<a name="line.118"></a>
<span class="sourceLineNo">119</span> .setNameFormat("HFileReplicationCopier-%1$d-" + this.sourceBaseNamespaceDirPath).<a name="line.119"></a>
<span class="sourceLineNo">120</span> build());<a name="line.120"></a>
<span class="sourceLineNo">121</span> this.copiesPerThread =<a name="line.121"></a>
<span class="sourceLineNo">122</span> conf.getInt(REPLICATION_BULKLOAD_COPY_HFILES_PERTHREAD_KEY,<a name="line.122"></a>
<span class="sourceLineNo">123</span> REPLICATION_BULKLOAD_COPY_HFILES_PERTHREAD_DEFAULT);<a name="line.123"></a>
<span class="sourceLineNo">124</span><a name="line.124"></a>
<span class="sourceLineNo">125</span> sinkFs = FileSystem.get(conf);<a name="line.125"></a>
<span class="sourceLineNo">126</span> }<a name="line.126"></a>
<span class="sourceLineNo">127</span><a name="line.127"></a>
<span class="sourceLineNo">128</span> @Override<a name="line.128"></a>
<span class="sourceLineNo">129</span> public void close() throws IOException {<a name="line.129"></a>
<span class="sourceLineNo">130</span> if (this.exec != null) {<a name="line.130"></a>
<span class="sourceLineNo">131</span> this.exec.shutdown();<a name="line.131"></a>
<span class="sourceLineNo">132</span> }<a name="line.132"></a>
<span class="sourceLineNo">133</span> }<a name="line.133"></a>
<span class="sourceLineNo">134</span><a name="line.134"></a>
<span class="sourceLineNo">135</span> public Void replicate() throws IOException {<a name="line.135"></a>
<span class="sourceLineNo">136</span> // Copy all the hfiles to the local file system<a name="line.136"></a>
<span class="sourceLineNo">137</span> Map&lt;String, Path&gt; tableStagingDirsMap = copyHFilesToStagingDir();<a name="line.137"></a>
<span class="sourceLineNo">138</span><a name="line.138"></a>
<span class="sourceLineNo">139</span> int maxRetries = conf.getInt(HConstants.BULKLOAD_MAX_RETRIES_NUMBER, 10);<a name="line.139"></a>
<span class="sourceLineNo">140</span><a name="line.140"></a>
<span class="sourceLineNo">141</span> for (Entry&lt;String, Path&gt; tableStagingDir : tableStagingDirsMap.entrySet()) {<a name="line.141"></a>
<span class="sourceLineNo">142</span> String tableNameString = tableStagingDir.getKey();<a name="line.142"></a>
<span class="sourceLineNo">143</span> Path stagingDir = tableStagingDir.getValue();<a name="line.143"></a>
<span class="sourceLineNo">144</span> TableName tableName = TableName.valueOf(tableNameString);<a name="line.144"></a>
<span class="sourceLineNo">145</span><a name="line.145"></a>
<span class="sourceLineNo">146</span> // Prepare collection of queue of hfiles to be loaded(replicated)<a name="line.146"></a>
<span class="sourceLineNo">147</span> Deque&lt;LoadQueueItem&gt; queue = new LinkedList&lt;&gt;();<a name="line.147"></a>
<span class="sourceLineNo">148</span> BulkLoadHFilesTool.prepareHFileQueue(conf, connection, tableName, stagingDir, queue, false,<a name="line.148"></a>
<span class="sourceLineNo">149</span> false);<a name="line.149"></a>
<span class="sourceLineNo">150</span><a name="line.150"></a>
<span class="sourceLineNo">151</span> if (queue.isEmpty()) {<a name="line.151"></a>
<span class="sourceLineNo">152</span> LOG.warn("Did not find any files to replicate in directory {}", stagingDir.toUri());<a name="line.152"></a>
<span class="sourceLineNo">153</span> return null;<a name="line.153"></a>
<span class="sourceLineNo">154</span> }<a name="line.154"></a>
<span class="sourceLineNo">155</span> fsDelegationToken.acquireDelegationToken(sinkFs);<a name="line.155"></a>
<span class="sourceLineNo">156</span> try {<a name="line.156"></a>
<span class="sourceLineNo">157</span> doBulkLoad(conf, tableName, stagingDir, queue, maxRetries);<a name="line.157"></a>
<span class="sourceLineNo">158</span> } finally {<a name="line.158"></a>
<span class="sourceLineNo">159</span> cleanup(stagingDir);<a name="line.159"></a>
<span class="sourceLineNo">160</span> }<a name="line.160"></a>
<span class="sourceLineNo">161</span> }<a name="line.161"></a>
<span class="sourceLineNo">162</span> return null;<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> private void doBulkLoad(Configuration conf, TableName tableName, Path stagingDir,<a name="line.165"></a>
<span class="sourceLineNo">166</span> Deque&lt;LoadQueueItem&gt; queue, int maxRetries) throws IOException {<a name="line.166"></a>
<span class="sourceLineNo">167</span> BulkLoadHFilesTool loader = new BulkLoadHFilesTool(conf);<a name="line.167"></a>
<span class="sourceLineNo">168</span> // Set the staging directory which will be used by BulkLoadHFilesTool for loading the data<a name="line.168"></a>
<span class="sourceLineNo">169</span> loader.setBulkToken(stagingDir.toString());<a name="line.169"></a>
<span class="sourceLineNo">170</span> //updating list of cluster ids where this bulkload event has already been processed<a name="line.170"></a>
<span class="sourceLineNo">171</span> loader.setClusterIds(sourceClusterIds);<a name="line.171"></a>
<span class="sourceLineNo">172</span> for (int count = 0; !queue.isEmpty(); count++) {<a name="line.172"></a>
<span class="sourceLineNo">173</span> if (count != 0) {<a name="line.173"></a>
<span class="sourceLineNo">174</span> LOG.warn("Error replicating HFiles; retry={} with {} remaining.", count, queue.size());<a name="line.174"></a>
<span class="sourceLineNo">175</span> }<a name="line.175"></a>
<span class="sourceLineNo">176</span><a name="line.176"></a>
<span class="sourceLineNo">177</span> if (maxRetries != 0 &amp;&amp; count &gt;= maxRetries) {<a name="line.177"></a>
<span class="sourceLineNo">178</span> throw new IOException("Retry attempted " + count + " times without completing, bailing.");<a name="line.178"></a>
<span class="sourceLineNo">179</span> }<a name="line.179"></a>
<span class="sourceLineNo">180</span><a name="line.180"></a>
<span class="sourceLineNo">181</span> // Try bulk load<a name="line.181"></a>
<span class="sourceLineNo">182</span> loader.loadHFileQueue(connection, tableName, queue, false);<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><a name="line.185"></a>
<span class="sourceLineNo">186</span> private void cleanup(Path stagingDir) {<a name="line.186"></a>
<span class="sourceLineNo">187</span> // Release the file system delegation token<a name="line.187"></a>
<span class="sourceLineNo">188</span> fsDelegationToken.releaseDelegationToken();<a name="line.188"></a>
<span class="sourceLineNo">189</span> // Delete the staging directory<a name="line.189"></a>
<span class="sourceLineNo">190</span> if (stagingDir != null) {<a name="line.190"></a>
<span class="sourceLineNo">191</span> try {<a name="line.191"></a>
<span class="sourceLineNo">192</span> sinkFs.delete(stagingDir, true);<a name="line.192"></a>
<span class="sourceLineNo">193</span> } catch (IOException e) {<a name="line.193"></a>
<span class="sourceLineNo">194</span> LOG.warn("Failed to delete the staging directory " + stagingDir, e);<a name="line.194"></a>
<span class="sourceLineNo">195</span> }<a name="line.195"></a>
<span class="sourceLineNo">196</span> }<a name="line.196"></a>
<span class="sourceLineNo">197</span> // Do not close the file system<a name="line.197"></a>
<span class="sourceLineNo">198</span> }<a name="line.198"></a>
<span class="sourceLineNo">199</span><a name="line.199"></a>
<span class="sourceLineNo">200</span> private Map&lt;String, Path&gt; copyHFilesToStagingDir() throws IOException {<a name="line.200"></a>
<span class="sourceLineNo">201</span> Map&lt;String, Path&gt; mapOfCopiedHFiles = new HashMap&lt;&gt;();<a name="line.201"></a>
<span class="sourceLineNo">202</span> Pair&lt;byte[], List&lt;String&gt;&gt; familyHFilePathsPair;<a name="line.202"></a>
<span class="sourceLineNo">203</span> List&lt;String&gt; hfilePaths;<a name="line.203"></a>
<span class="sourceLineNo">204</span> byte[] family;<a name="line.204"></a>
<span class="sourceLineNo">205</span> Path familyStagingDir;<a name="line.205"></a>
<span class="sourceLineNo">206</span> int familyHFilePathsPairsListSize;<a name="line.206"></a>
<span class="sourceLineNo">207</span> int totalNoOfHFiles;<a name="line.207"></a>
<span class="sourceLineNo">208</span> List&lt;Pair&lt;byte[], List&lt;String&gt;&gt;&gt; familyHFilePathsPairsList;<a name="line.208"></a>
<span class="sourceLineNo">209</span> FileSystem sourceFs = null;<a name="line.209"></a>
<span class="sourceLineNo">210</span><a name="line.210"></a>
<span class="sourceLineNo">211</span> try {<a name="line.211"></a>
<span class="sourceLineNo">212</span> Path sourceClusterPath = new Path(sourceBaseNamespaceDirPath);<a name="line.212"></a>
<span class="sourceLineNo">213</span> /*<a name="line.213"></a>
<span class="sourceLineNo">214</span> * Path#getFileSystem will by default get the FS from cache. If both source and sink cluster<a name="line.214"></a>
<span class="sourceLineNo">215</span> * has same FS name service then it will return peer cluster FS. To avoid this we explicitly<a name="line.215"></a>
<span class="sourceLineNo">216</span> * disable the loading of FS from cache, so that a new FS is created with source cluster<a name="line.216"></a>
<span class="sourceLineNo">217</span> * configuration.<a name="line.217"></a>
<span class="sourceLineNo">218</span> */<a name="line.218"></a>
<span class="sourceLineNo">219</span> String sourceScheme = sourceClusterPath.toUri().getScheme();<a name="line.219"></a>
<span class="sourceLineNo">220</span> String disableCacheName =<a name="line.220"></a>
<span class="sourceLineNo">221</span> String.format("fs.%s.impl.disable.cache", new Object[] { sourceScheme });<a name="line.221"></a>
<span class="sourceLineNo">222</span> sourceClusterConf.setBoolean(disableCacheName, true);<a name="line.222"></a>
<span class="sourceLineNo">223</span><a name="line.223"></a>
<span class="sourceLineNo">224</span> sourceFs = sourceClusterPath.getFileSystem(sourceClusterConf);<a name="line.224"></a>
<span class="sourceLineNo">225</span><a name="line.225"></a>
<span class="sourceLineNo">226</span> User user = userProvider.getCurrent();<a name="line.226"></a>
<span class="sourceLineNo">227</span> // For each table name in the map<a name="line.227"></a>
<span class="sourceLineNo">228</span> for (Entry&lt;String, List&lt;Pair&lt;byte[], List&lt;String&gt;&gt;&gt;&gt; tableEntry : bulkLoadHFileMap<a name="line.228"></a>
<span class="sourceLineNo">229</span> .entrySet()) {<a name="line.229"></a>
<span class="sourceLineNo">230</span> String tableName = tableEntry.getKey();<a name="line.230"></a>
<span class="sourceLineNo">231</span><a name="line.231"></a>
<span class="sourceLineNo">232</span> // Create staging directory for each table<a name="line.232"></a>
<span class="sourceLineNo">233</span> Path stagingDir =<a name="line.233"></a>
<span class="sourceLineNo">234</span> createStagingDir(hbaseStagingDir, user, TableName.valueOf(tableName));<a name="line.234"></a>
<span class="sourceLineNo">235</span><a name="line.235"></a>
<span class="sourceLineNo">236</span> familyHFilePathsPairsList = tableEntry.getValue();<a name="line.236"></a>
<span class="sourceLineNo">237</span> familyHFilePathsPairsListSize = familyHFilePathsPairsList.size();<a name="line.237"></a>
<span class="sourceLineNo">238</span><a name="line.238"></a>
<span class="sourceLineNo">239</span> // For each list of family hfile paths pair in the table<a name="line.239"></a>
<span class="sourceLineNo">240</span> for (int i = 0; i &lt; familyHFilePathsPairsListSize; i++) {<a name="line.240"></a>
<span class="sourceLineNo">241</span> familyHFilePathsPair = familyHFilePathsPairsList.get(i);<a name="line.241"></a>
<span class="sourceLineNo">242</span><a name="line.242"></a>
<span class="sourceLineNo">243</span> family = familyHFilePathsPair.getFirst();<a name="line.243"></a>
<span class="sourceLineNo">244</span> hfilePaths = familyHFilePathsPair.getSecond();<a name="line.244"></a>
<span class="sourceLineNo">245</span><a name="line.245"></a>
<span class="sourceLineNo">246</span> familyStagingDir = new Path(stagingDir, Bytes.toString(family));<a name="line.246"></a>
<span class="sourceLineNo">247</span> totalNoOfHFiles = hfilePaths.size();<a name="line.247"></a>
<span class="sourceLineNo">248</span><a name="line.248"></a>
<span class="sourceLineNo">249</span> // For each list of hfile paths for the family<a name="line.249"></a>
<span class="sourceLineNo">250</span> List&lt;Future&lt;Void&gt;&gt; futures = new ArrayList&lt;&gt;();<a name="line.250"></a>
<span class="sourceLineNo">251</span> Callable&lt;Void&gt; c;<a name="line.251"></a>
<span class="sourceLineNo">252</span> Future&lt;Void&gt; future;<a name="line.252"></a>
<span class="sourceLineNo">253</span> int currentCopied = 0;<a name="line.253"></a>
<span class="sourceLineNo">254</span> // Copy the hfiles parallely<a name="line.254"></a>
<span class="sourceLineNo">255</span> while (totalNoOfHFiles &gt; currentCopied + this.copiesPerThread) {<a name="line.255"></a>
<span class="sourceLineNo">256</span> c =<a name="line.256"></a>
<span class="sourceLineNo">257</span> new Copier(sourceFs, familyStagingDir, hfilePaths.subList(currentCopied,<a name="line.257"></a>
<span class="sourceLineNo">258</span> currentCopied + this.copiesPerThread));<a name="line.258"></a>
<span class="sourceLineNo">259</span> future = exec.submit(c);<a name="line.259"></a>
<span class="sourceLineNo">260</span> futures.add(future);<a name="line.260"></a>
<span class="sourceLineNo">261</span> currentCopied += this.copiesPerThread;<a name="line.261"></a>
<span class="sourceLineNo">262</span> }<a name="line.262"></a>
<span class="sourceLineNo">263</span><a name="line.263"></a>
<span class="sourceLineNo">264</span> int remaining = totalNoOfHFiles - currentCopied;<a name="line.264"></a>
<span class="sourceLineNo">265</span> if (remaining &gt; 0) {<a name="line.265"></a>
<span class="sourceLineNo">266</span> c =<a name="line.266"></a>
<span class="sourceLineNo">267</span> new Copier(sourceFs, familyStagingDir, hfilePaths.subList(currentCopied,<a name="line.267"></a>
<span class="sourceLineNo">268</span> currentCopied + remaining));<a name="line.268"></a>
<span class="sourceLineNo">269</span> future = exec.submit(c);<a name="line.269"></a>
<span class="sourceLineNo">270</span> futures.add(future);<a name="line.270"></a>
<span class="sourceLineNo">271</span> }<a name="line.271"></a>
<span class="sourceLineNo">272</span><a name="line.272"></a>
<span class="sourceLineNo">273</span> for (Future&lt;Void&gt; f : futures) {<a name="line.273"></a>
<span class="sourceLineNo">274</span> try {<a name="line.274"></a>
<span class="sourceLineNo">275</span> f.get();<a name="line.275"></a>
<span class="sourceLineNo">276</span> } catch (InterruptedException e) {<a name="line.276"></a>
<span class="sourceLineNo">277</span> InterruptedIOException iioe =<a name="line.277"></a>
<span class="sourceLineNo">278</span> new InterruptedIOException(<a name="line.278"></a>
<span class="sourceLineNo">279</span> "Failed to copy HFiles to local file system. This will be retried again "<a name="line.279"></a>
<span class="sourceLineNo">280</span> + "by the source cluster.");<a name="line.280"></a>
<span class="sourceLineNo">281</span> iioe.initCause(e);<a name="line.281"></a>
<span class="sourceLineNo">282</span> throw iioe;<a name="line.282"></a>
<span class="sourceLineNo">283</span> } catch (ExecutionException e) {<a name="line.283"></a>
<span class="sourceLineNo">284</span> throw new IOException("Failed to copy HFiles to local file system. This will "<a name="line.284"></a>
<span class="sourceLineNo">285</span> + "be retried again by the source cluster.", e);<a name="line.285"></a>
<span class="sourceLineNo">286</span> }<a name="line.286"></a>
<span class="sourceLineNo">287</span> }<a name="line.287"></a>
<span class="sourceLineNo">288</span> }<a name="line.288"></a>
<span class="sourceLineNo">289</span> // Add the staging directory to this table. Staging directory contains all the hfiles<a name="line.289"></a>
<span class="sourceLineNo">290</span> // belonging to this table<a name="line.290"></a>
<span class="sourceLineNo">291</span> mapOfCopiedHFiles.put(tableName, stagingDir);<a name="line.291"></a>
<span class="sourceLineNo">292</span> }<a name="line.292"></a>
<span class="sourceLineNo">293</span> return mapOfCopiedHFiles;<a name="line.293"></a>
<span class="sourceLineNo">294</span> } finally {<a name="line.294"></a>
<span class="sourceLineNo">295</span> if (sourceFs != null) {<a name="line.295"></a>
<span class="sourceLineNo">296</span> sourceFs.close();<a name="line.296"></a>
<span class="sourceLineNo">297</span> }<a name="line.297"></a>
<span class="sourceLineNo">298</span> if(exec != null) {<a name="line.298"></a>
<span class="sourceLineNo">299</span> exec.shutdown();<a name="line.299"></a>
<span class="sourceLineNo">300</span> }<a name="line.300"></a>
<span class="sourceLineNo">301</span> }<a name="line.301"></a>
<span class="sourceLineNo">302</span> }<a name="line.302"></a>
<span class="sourceLineNo">303</span><a name="line.303"></a>
<span class="sourceLineNo">304</span> private Path createStagingDir(Path baseDir, User user, TableName tableName) throws IOException {<a name="line.304"></a>
<span class="sourceLineNo">305</span> String tblName = tableName.getNameAsString().replace(":", UNDERSCORE);<a name="line.305"></a>
<span class="sourceLineNo">306</span> int RANDOM_WIDTH = 320;<a name="line.306"></a>
<span class="sourceLineNo">307</span> int RANDOM_RADIX = 32;<a name="line.307"></a>
<span class="sourceLineNo">308</span> String doubleUnderScore = UNDERSCORE + UNDERSCORE;<a name="line.308"></a>
<span class="sourceLineNo">309</span> String randomDir = user.getShortName() + doubleUnderScore + tblName + doubleUnderScore<a name="line.309"></a>
<span class="sourceLineNo">310</span> + (new BigInteger(RANDOM_WIDTH, new SecureRandom()).toString(RANDOM_RADIX));<a name="line.310"></a>
<span class="sourceLineNo">311</span> return createStagingDir(baseDir, user, randomDir);<a name="line.311"></a>
<span class="sourceLineNo">312</span> }<a name="line.312"></a>
<span class="sourceLineNo">313</span><a name="line.313"></a>
<span class="sourceLineNo">314</span> private Path createStagingDir(Path baseDir, User user, String randomDir) throws IOException {<a name="line.314"></a>
<span class="sourceLineNo">315</span> Path p = new Path(baseDir, randomDir);<a name="line.315"></a>
<span class="sourceLineNo">316</span> sinkFs.mkdirs(p, PERM_ALL_ACCESS);<a name="line.316"></a>
<span class="sourceLineNo">317</span> sinkFs.setPermission(p, PERM_ALL_ACCESS);<a name="line.317"></a>
<span class="sourceLineNo">318</span> return p;<a name="line.318"></a>
<span class="sourceLineNo">319</span> }<a name="line.319"></a>
<span class="sourceLineNo">320</span><a name="line.320"></a>
<span class="sourceLineNo">321</span> /**<a name="line.321"></a>
<span class="sourceLineNo">322</span> * This class will copy the given hfiles from the given source file system to the given local file<a name="line.322"></a>
<span class="sourceLineNo">323</span> * system staging directory.<a name="line.323"></a>
<span class="sourceLineNo">324</span> */<a name="line.324"></a>
<span class="sourceLineNo">325</span> private class Copier implements Callable&lt;Void&gt; {<a name="line.325"></a>
<span class="sourceLineNo">326</span> private FileSystem sourceFs;<a name="line.326"></a>
<span class="sourceLineNo">327</span> private Path stagingDir;<a name="line.327"></a>
<span class="sourceLineNo">328</span> private List&lt;String&gt; hfiles;<a name="line.328"></a>
<span class="sourceLineNo">329</span><a name="line.329"></a>
<span class="sourceLineNo">330</span> public Copier(FileSystem sourceFs, final Path stagingDir, final List&lt;String&gt; hfiles)<a name="line.330"></a>
<span class="sourceLineNo">331</span> throws IOException {<a name="line.331"></a>
<span class="sourceLineNo">332</span> this.sourceFs = sourceFs;<a name="line.332"></a>
<span class="sourceLineNo">333</span> this.stagingDir = stagingDir;<a name="line.333"></a>
<span class="sourceLineNo">334</span> this.hfiles = hfiles;<a name="line.334"></a>
<span class="sourceLineNo">335</span> }<a name="line.335"></a>
<span class="sourceLineNo">336</span><a name="line.336"></a>
<span class="sourceLineNo">337</span> @Override<a name="line.337"></a>
<span class="sourceLineNo">338</span> public Void call() throws IOException {<a name="line.338"></a>
<span class="sourceLineNo">339</span> Path sourceHFilePath;<a name="line.339"></a>
<span class="sourceLineNo">340</span> Path localHFilePath;<a name="line.340"></a>
<span class="sourceLineNo">341</span> int totalHFiles = hfiles.size();<a name="line.341"></a>
<span class="sourceLineNo">342</span> for (int i = 0; i &lt; totalHFiles; i++) {<a name="line.342"></a>
<span class="sourceLineNo">343</span> sourceHFilePath = new Path(sourceBaseNamespaceDirPath, hfiles.get(i));<a name="line.343"></a>
<span class="sourceLineNo">344</span> localHFilePath = new Path(stagingDir, sourceHFilePath.getName());<a name="line.344"></a>
<span class="sourceLineNo">345</span> try {<a name="line.345"></a>
<span class="sourceLineNo">346</span> FileUtil.copy(sourceFs, sourceHFilePath, sinkFs, localHFilePath, false, conf);<a name="line.346"></a>
<span class="sourceLineNo">347</span> // If any other exception other than FNFE then we will fail the replication requests and<a name="line.347"></a>
<span class="sourceLineNo">348</span> // source will retry to replicate these data.<a name="line.348"></a>
<span class="sourceLineNo">349</span> } catch (FileNotFoundException e) {<a name="line.349"></a>
<span class="sourceLineNo">350</span> LOG.info("Failed to copy hfile from " + sourceHFilePath + " to " + localHFilePath<a name="line.350"></a>
<span class="sourceLineNo">351</span> + ". Trying to copy from hfile archive directory.",<a name="line.351"></a>
<span class="sourceLineNo">352</span> e);<a name="line.352"></a>
<span class="sourceLineNo">353</span> sourceHFilePath = new Path(sourceHFileArchiveDirPath, hfiles.get(i));<a name="line.353"></a>
<span class="sourceLineNo">354</span><a name="line.354"></a>
<span class="sourceLineNo">355</span> try {<a name="line.355"></a>
<span class="sourceLineNo">356</span> FileUtil.copy(sourceFs, sourceHFilePath, sinkFs, localHFilePath, false, conf);<a name="line.356"></a>
<span class="sourceLineNo">357</span> } catch (FileNotFoundException e1) {<a name="line.357"></a>
<span class="sourceLineNo">358</span> // This will mean that the hfile does not exists any where in source cluster FS. So we<a name="line.358"></a>
<span class="sourceLineNo">359</span> // cannot do anything here just log and continue.<a name="line.359"></a>
<span class="sourceLineNo">360</span> LOG.debug("Failed to copy hfile from " + sourceHFilePath + " to " + localHFilePath<a name="line.360"></a>
<span class="sourceLineNo">361</span> + ". Hence ignoring this hfile from replication..",<a name="line.361"></a>
<span class="sourceLineNo">362</span> e1);<a name="line.362"></a>
<span class="sourceLineNo">363</span> continue;<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> sinkFs.setPermission(localHFilePath, PERM_ALL_ACCESS);<a name="line.366"></a>
<span class="sourceLineNo">367</span> }<a name="line.367"></a>
<span class="sourceLineNo">368</span> return null;<a name="line.368"></a>
<span class="sourceLineNo">369</span> }<a name="line.369"></a>
<span class="sourceLineNo">370</span> }<a name="line.370"></a>
<span class="sourceLineNo">371</span>}<a name="line.371"></a>
</pre>
</div>
</body>
</html>