| <!DOCTYPE HTML> |
| <html lang="en"> |
| <head> |
| <!-- Generated by javadoc (17) --> |
| <title>Source code</title> |
| <meta name="viewport" content="width=device-width, initial-scale=1"> |
| <meta name="description" content="source: package: org.apache.hadoop.hbase.regionserver, class: SplitLogWorker"> |
| <meta name="generator" content="javadoc/SourceToHTMLConverter"> |
| <link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style"> |
| </head> |
| <body class="source-page"> |
| <main role="main"> |
| <div class="source-container"> |
| <pre><span class="source-line-no">001</span><span id="line-1">/*</span> |
| <span class="source-line-no">002</span><span id="line-2"> * Licensed to the Apache Software Foundation (ASF) under one</span> |
| <span class="source-line-no">003</span><span id="line-3"> * or more contributor license agreements. See the NOTICE file</span> |
| <span class="source-line-no">004</span><span id="line-4"> * distributed with this work for additional information</span> |
| <span class="source-line-no">005</span><span id="line-5"> * regarding copyright ownership. The ASF licenses this file</span> |
| <span class="source-line-no">006</span><span id="line-6"> * to you under the Apache License, Version 2.0 (the</span> |
| <span class="source-line-no">007</span><span id="line-7"> * "License"); you may not use this file except in compliance</span> |
| <span class="source-line-no">008</span><span id="line-8"> * with the License. You may obtain a copy of the License at</span> |
| <span class="source-line-no">009</span><span id="line-9"> *</span> |
| <span class="source-line-no">010</span><span id="line-10"> * http://www.apache.org/licenses/LICENSE-2.0</span> |
| <span class="source-line-no">011</span><span id="line-11"> *</span> |
| <span class="source-line-no">012</span><span id="line-12"> * Unless required by applicable law or agreed to in writing, software</span> |
| <span class="source-line-no">013</span><span id="line-13"> * distributed under the License is distributed on an "AS IS" BASIS,</span> |
| <span class="source-line-no">014</span><span id="line-14"> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span> |
| <span class="source-line-no">015</span><span id="line-15"> * See the License for the specific language governing permissions and</span> |
| <span class="source-line-no">016</span><span id="line-16"> * limitations under the License.</span> |
| <span class="source-line-no">017</span><span id="line-17"> */</span> |
| <span class="source-line-no">018</span><span id="line-18">package org.apache.hadoop.hbase.regionserver;</span> |
| <span class="source-line-no">019</span><span id="line-19"></span> |
| <span class="source-line-no">020</span><span id="line-20">import java.io.FileNotFoundException;</span> |
| <span class="source-line-no">021</span><span id="line-21">import java.io.IOException;</span> |
| <span class="source-line-no">022</span><span id="line-22">import java.io.InterruptedIOException;</span> |
| <span class="source-line-no">023</span><span id="line-23">import java.net.ConnectException;</span> |
| <span class="source-line-no">024</span><span id="line-24">import java.net.SocketTimeoutException;</span> |
| <span class="source-line-no">025</span><span id="line-25">import java.util.Optional;</span> |
| <span class="source-line-no">026</span><span id="line-26">import org.apache.commons.io.IOUtils;</span> |
| <span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.conf.Configuration;</span> |
| <span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.fs.FSDataInputStream;</span> |
| <span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.fs.FSDataOutputStream;</span> |
| <span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.fs.FileSystem;</span> |
| <span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.fs.Path;</span> |
| <span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.NotServingRegionException;</span> |
| <span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.Server;</span> |
| <span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.client.RetriesExhaustedException;</span> |
| <span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.coordination.SplitLogWorkerCoordination;</span> |
| <span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.regionserver.SplitLogWorker.TaskExecutor.Status;</span> |
| <span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.replication.ReplicationPeerImpl;</span> |
| <span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.replication.ReplicationUtils;</span> |
| <span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.hbase.replication.SyncReplicationState;</span> |
| <span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.hbase.util.CancelableProgressable;</span> |
| <span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.hbase.util.CommonFSUtils;</span> |
| <span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.hbase.util.ExceptionUtil;</span> |
| <span class="source-line-no">043</span><span id="line-43">import org.apache.hadoop.hbase.util.FSUtils;</span> |
| <span class="source-line-no">044</span><span id="line-44">import org.apache.hadoop.hbase.util.Pair;</span> |
| <span class="source-line-no">045</span><span id="line-45">import org.apache.hadoop.hbase.wal.AbstractWALProvider;</span> |
| <span class="source-line-no">046</span><span id="line-46">import org.apache.hadoop.hbase.wal.WALFactory;</span> |
| <span class="source-line-no">047</span><span id="line-47">import org.apache.hadoop.hbase.wal.WALSplitter;</span> |
| <span class="source-line-no">048</span><span id="line-48">import org.apache.yetus.audience.InterfaceAudience;</span> |
| <span class="source-line-no">049</span><span id="line-49">import org.slf4j.Logger;</span> |
| <span class="source-line-no">050</span><span id="line-50">import org.slf4j.LoggerFactory;</span> |
| <span class="source-line-no">051</span><span id="line-51"></span> |
| <span class="source-line-no">052</span><span id="line-52">/**</span> |
| <span class="source-line-no">053</span><span id="line-53"> * This worker is spawned in every regionserver, including master. The Worker waits for log</span> |
| <span class="source-line-no">054</span><span id="line-54"> * splitting tasks to be put up by the {@link org.apache.hadoop.hbase.master.SplitLogManager}</span> |
| <span class="source-line-no">055</span><span id="line-55"> * running in the master and races with other workers in other serves to acquire those tasks. The</span> |
| <span class="source-line-no">056</span><span id="line-56"> * coordination is done via coordination engine.</span> |
| <span class="source-line-no">057</span><span id="line-57"> * <p></span> |
| <span class="source-line-no">058</span><span id="line-58"> * If a worker has successfully moved the task from state UNASSIGNED to OWNED then it owns the task.</span> |
| <span class="source-line-no">059</span><span id="line-59"> * It keeps heart beating the manager by periodically moving the task from UNASSIGNED to OWNED</span> |
| <span class="source-line-no">060</span><span id="line-60"> * state. On success it moves the task to TASK_DONE. On unrecoverable error it moves task state to</span> |
| <span class="source-line-no">061</span><span id="line-61"> * ERR. If it cannot continue but wants the master to retry the task then it moves the task state to</span> |
| <span class="source-line-no">062</span><span id="line-62"> * RESIGNED.</span> |
| <span class="source-line-no">063</span><span id="line-63"> * <p></span> |
| <span class="source-line-no">064</span><span id="line-64"> * The manager can take a task away from a worker by moving the task from OWNED to UNASSIGNED. In</span> |
| <span class="source-line-no">065</span><span id="line-65"> * the absence of a global lock there is a unavoidable race here - a worker might have just finished</span> |
| <span class="source-line-no">066</span><span id="line-66"> * its task when it is stripped of its ownership. Here we rely on the idempotency of the log</span> |
| <span class="source-line-no">067</span><span id="line-67"> * splitting task for correctness</span> |
| <span class="source-line-no">068</span><span id="line-68"> * @deprecated since 2.4.0 and in 3.0.0, to be removed in 4.0.0, replaced by procedure-based</span> |
| <span class="source-line-no">069</span><span id="line-69"> * distributed WAL splitter, see SplitWALRemoteProcedure</span> |
| <span class="source-line-no">070</span><span id="line-70"> */</span> |
| <span class="source-line-no">071</span><span id="line-71">@Deprecated</span> |
| <span class="source-line-no">072</span><span id="line-72">@InterfaceAudience.Private</span> |
| <span class="source-line-no">073</span><span id="line-73">public class SplitLogWorker implements Runnable {</span> |
| <span class="source-line-no">074</span><span id="line-74"></span> |
| <span class="source-line-no">075</span><span id="line-75"> private static final Logger LOG = LoggerFactory.getLogger(SplitLogWorker.class);</span> |
| <span class="source-line-no">076</span><span id="line-76"></span> |
| <span class="source-line-no">077</span><span id="line-77"> Thread worker;</span> |
| <span class="source-line-no">078</span><span id="line-78"> // thread pool which executes recovery work</span> |
| <span class="source-line-no">079</span><span id="line-79"> private final SplitLogWorkerCoordination coordination;</span> |
| <span class="source-line-no">080</span><span id="line-80"> private final RegionServerServices server;</span> |
| <span class="source-line-no">081</span><span id="line-81"></span> |
| <span class="source-line-no">082</span><span id="line-82"> public SplitLogWorker(Server hserver, Configuration conf, RegionServerServices server,</span> |
| <span class="source-line-no">083</span><span id="line-83"> TaskExecutor splitTaskExecutor) {</span> |
| <span class="source-line-no">084</span><span id="line-84"> this.server = server;</span> |
| <span class="source-line-no">085</span><span id="line-85"> this.coordination = hserver.getCoordinatedStateManager().getSplitLogWorkerCoordination();</span> |
| <span class="source-line-no">086</span><span id="line-86"> coordination.init(server, conf, splitTaskExecutor, this);</span> |
| <span class="source-line-no">087</span><span id="line-87"> }</span> |
| <span class="source-line-no">088</span><span id="line-88"></span> |
| <span class="source-line-no">089</span><span id="line-89"> public SplitLogWorker(Configuration conf, RegionServerServices server,</span> |
| <span class="source-line-no">090</span><span id="line-90"> LastSequenceId sequenceIdChecker, WALFactory factory) {</span> |
| <span class="source-line-no">091</span><span id="line-91"> this(server, conf, server, (f, p) -> splitLog(f, p, conf, server, sequenceIdChecker, factory));</span> |
| <span class="source-line-no">092</span><span id="line-92"> }</span> |
| <span class="source-line-no">093</span><span id="line-93"></span> |
| <span class="source-line-no">094</span><span id="line-94"> // returns whether we need to continue the split work</span> |
| <span class="source-line-no">095</span><span id="line-95"> private static boolean processSyncReplicationWAL(String name, Configuration conf,</span> |
| <span class="source-line-no">096</span><span id="line-96"> RegionServerServices server, FileSystem fs, Path walDir) throws IOException {</span> |
| <span class="source-line-no">097</span><span id="line-97"> Path walFile = new Path(walDir, name);</span> |
| <span class="source-line-no">098</span><span id="line-98"> String filename = walFile.getName();</span> |
| <span class="source-line-no">099</span><span id="line-99"> Optional<String> optSyncPeerId =</span> |
| <span class="source-line-no">100</span><span id="line-100"> AbstractWALProvider.getSyncReplicationPeerIdFromWALName(filename);</span> |
| <span class="source-line-no">101</span><span id="line-101"> if (!optSyncPeerId.isPresent()) {</span> |
| <span class="source-line-no">102</span><span id="line-102"> return true;</span> |
| <span class="source-line-no">103</span><span id="line-103"> }</span> |
| <span class="source-line-no">104</span><span id="line-104"> String peerId = optSyncPeerId.get();</span> |
| <span class="source-line-no">105</span><span id="line-105"> ReplicationPeerImpl peer =</span> |
| <span class="source-line-no">106</span><span id="line-106"> server.getReplicationSourceService().getReplicationPeers().getPeer(peerId);</span> |
| <span class="source-line-no">107</span><span id="line-107"> if (peer == null || !peer.getPeerConfig().isSyncReplication()) {</span> |
| <span class="source-line-no">108</span><span id="line-108"> return true;</span> |
| <span class="source-line-no">109</span><span id="line-109"> }</span> |
| <span class="source-line-no">110</span><span id="line-110"> Pair<SyncReplicationState, SyncReplicationState> stateAndNewState =</span> |
| <span class="source-line-no">111</span><span id="line-111"> peer.getSyncReplicationStateAndNewState();</span> |
| <span class="source-line-no">112</span><span id="line-112"> if (</span> |
| <span class="source-line-no">113</span><span id="line-113"> stateAndNewState.getFirst().equals(SyncReplicationState.ACTIVE)</span> |
| <span class="source-line-no">114</span><span id="line-114"> && stateAndNewState.getSecond().equals(SyncReplicationState.NONE)</span> |
| <span class="source-line-no">115</span><span id="line-115"> ) {</span> |
| <span class="source-line-no">116</span><span id="line-116"> // copy the file to remote and overwrite the previous one</span> |
| <span class="source-line-no">117</span><span id="line-117"> String remoteWALDir = peer.getPeerConfig().getRemoteWALDir();</span> |
| <span class="source-line-no">118</span><span id="line-118"> Path remoteWALDirForPeer = ReplicationUtils.getPeerRemoteWALDir(remoteWALDir, peerId);</span> |
| <span class="source-line-no">119</span><span id="line-119"> Path tmpRemoteWAL = new Path(remoteWALDirForPeer, filename + ".tmp");</span> |
| <span class="source-line-no">120</span><span id="line-120"> FileSystem remoteFs = ReplicationUtils.getRemoteWALFileSystem(conf, remoteWALDir);</span> |
| <span class="source-line-no">121</span><span id="line-121"> try (FSDataInputStream in = fs.open(walFile);</span> |
| <span class="source-line-no">122</span><span id="line-122"> FSDataOutputStream out = remoteFs.createNonRecursive(tmpRemoteWAL, true,</span> |
| <span class="source-line-no">123</span><span id="line-123"> CommonFSUtils.getDefaultBufferSize(remoteFs),</span> |
| <span class="source-line-no">124</span><span id="line-124"> remoteFs.getDefaultReplication(tmpRemoteWAL), remoteFs.getDefaultBlockSize(tmpRemoteWAL),</span> |
| <span class="source-line-no">125</span><span id="line-125"> null)) {</span> |
| <span class="source-line-no">126</span><span id="line-126"> IOUtils.copy(in, out);</span> |
| <span class="source-line-no">127</span><span id="line-127"> }</span> |
| <span class="source-line-no">128</span><span id="line-128"> Path toCommitRemoteWAL =</span> |
| <span class="source-line-no">129</span><span id="line-129"> new Path(remoteWALDirForPeer, filename + ReplicationUtils.RENAME_WAL_SUFFIX);</span> |
| <span class="source-line-no">130</span><span id="line-130"> // Some FileSystem implementations may not support atomic rename so we need to do it in two</span> |
| <span class="source-line-no">131</span><span id="line-131"> // phases</span> |
| <span class="source-line-no">132</span><span id="line-132"> FSUtils.renameFile(remoteFs, tmpRemoteWAL, toCommitRemoteWAL);</span> |
| <span class="source-line-no">133</span><span id="line-133"> FSUtils.renameFile(remoteFs, toCommitRemoteWAL, new Path(remoteWALDirForPeer, filename));</span> |
| <span class="source-line-no">134</span><span id="line-134"> } else if (</span> |
| <span class="source-line-no">135</span><span id="line-135"> (stateAndNewState.getFirst().equals(SyncReplicationState.ACTIVE)</span> |
| <span class="source-line-no">136</span><span id="line-136"> && stateAndNewState.getSecond().equals(SyncReplicationState.STANDBY))</span> |
| <span class="source-line-no">137</span><span id="line-137"> || stateAndNewState.getFirst().equals(SyncReplicationState.STANDBY)</span> |
| <span class="source-line-no">138</span><span id="line-138"> ) {</span> |
| <span class="source-line-no">139</span><span id="line-139"> // check whether we still need to process this file</span> |
| <span class="source-line-no">140</span><span id="line-140"> // actually we only write wal file which name is ended with .syncrep in A state, and after</span> |
| <span class="source-line-no">141</span><span id="line-141"> // transiting to a state other than A, we will reopen all the regions so the data in the wal</span> |
| <span class="source-line-no">142</span><span id="line-142"> // will be flushed so the wal file will be archived soon. But it is still possible that there</span> |
| <span class="source-line-no">143</span><span id="line-143"> // is a server crash when we are transiting from A to S, to simplify the logic of the transit</span> |
| <span class="source-line-no">144</span><span id="line-144"> // procedure, here we will also check the remote snapshot directory in state S, so that we do</span> |
| <span class="source-line-no">145</span><span id="line-145"> // not need wait until all the wal files with .syncrep suffix to be archived before finishing</span> |
| <span class="source-line-no">146</span><span id="line-146"> // the procedure.</span> |
| <span class="source-line-no">147</span><span id="line-147"> String remoteWALDir = peer.getPeerConfig().getRemoteWALDir();</span> |
| <span class="source-line-no">148</span><span id="line-148"> Path remoteSnapshotDirForPeer = ReplicationUtils.getPeerSnapshotWALDir(remoteWALDir, peerId);</span> |
| <span class="source-line-no">149</span><span id="line-149"> FileSystem remoteFs = ReplicationUtils.getRemoteWALFileSystem(conf, remoteWALDir);</span> |
| <span class="source-line-no">150</span><span id="line-150"> if (remoteFs.exists(new Path(remoteSnapshotDirForPeer, filename))) {</span> |
| <span class="source-line-no">151</span><span id="line-151"> // the file has been replayed when the remote cluster was transited from S to DA, the</span> |
| <span class="source-line-no">152</span><span id="line-152"> // content will be replicated back to us so give up split it.</span> |
| <span class="source-line-no">153</span><span id="line-153"> LOG.warn("Giveup splitting {} since it has been replayed in the remote cluster and "</span> |
| <span class="source-line-no">154</span><span id="line-154"> + "the content will be replicated back", filename);</span> |
| <span class="source-line-no">155</span><span id="line-155"> return false;</span> |
| <span class="source-line-no">156</span><span id="line-156"> }</span> |
| <span class="source-line-no">157</span><span id="line-157"> }</span> |
| <span class="source-line-no">158</span><span id="line-158"> return true;</span> |
| <span class="source-line-no">159</span><span id="line-159"> }</span> |
| <span class="source-line-no">160</span><span id="line-160"></span> |
| <span class="source-line-no">161</span><span id="line-161"> /** Returns Result either DONE, RESIGNED, or ERR. */</span> |
| <span class="source-line-no">162</span><span id="line-162"> static Status splitLog(String filename, CancelableProgressable p, Configuration conf,</span> |
| <span class="source-line-no">163</span><span id="line-163"> RegionServerServices server, LastSequenceId sequenceIdChecker, WALFactory factory) {</span> |
| <span class="source-line-no">164</span><span id="line-164"> Path walDir;</span> |
| <span class="source-line-no">165</span><span id="line-165"> FileSystem fs;</span> |
| <span class="source-line-no">166</span><span id="line-166"> try {</span> |
| <span class="source-line-no">167</span><span id="line-167"> walDir = CommonFSUtils.getWALRootDir(conf);</span> |
| <span class="source-line-no">168</span><span id="line-168"> fs = walDir.getFileSystem(conf);</span> |
| <span class="source-line-no">169</span><span id="line-169"> } catch (IOException e) {</span> |
| <span class="source-line-no">170</span><span id="line-170"> LOG.warn("Resigning, could not find root dir or fs", e);</span> |
| <span class="source-line-no">171</span><span id="line-171"> return Status.RESIGNED;</span> |
| <span class="source-line-no">172</span><span id="line-172"> }</span> |
| <span class="source-line-no">173</span><span id="line-173"> try {</span> |
| <span class="source-line-no">174</span><span id="line-174"> if (!processSyncReplicationWAL(filename, conf, server, fs, walDir)) {</span> |
| <span class="source-line-no">175</span><span id="line-175"> return Status.DONE;</span> |
| <span class="source-line-no">176</span><span id="line-176"> }</span> |
| <span class="source-line-no">177</span><span id="line-177"> } catch (IOException e) {</span> |
| <span class="source-line-no">178</span><span id="line-178"> LOG.warn("failed to process sync replication wal {}", filename, e);</span> |
| <span class="source-line-no">179</span><span id="line-179"> return Status.RESIGNED;</span> |
| <span class="source-line-no">180</span><span id="line-180"> }</span> |
| <span class="source-line-no">181</span><span id="line-181"> // TODO have to correctly figure out when log splitting has been</span> |
| <span class="source-line-no">182</span><span id="line-182"> // interrupted or has encountered a transient error and when it has</span> |
| <span class="source-line-no">183</span><span id="line-183"> // encountered a bad non-retry-able persistent error.</span> |
| <span class="source-line-no">184</span><span id="line-184"> try {</span> |
| <span class="source-line-no">185</span><span id="line-185"> SplitLogWorkerCoordination splitLogWorkerCoordination =</span> |
| <span class="source-line-no">186</span><span id="line-186"> server.getCoordinatedStateManager() == null</span> |
| <span class="source-line-no">187</span><span id="line-187"> ? null</span> |
| <span class="source-line-no">188</span><span id="line-188"> : server.getCoordinatedStateManager().getSplitLogWorkerCoordination();</span> |
| <span class="source-line-no">189</span><span id="line-189"> if (</span> |
| <span class="source-line-no">190</span><span id="line-190"> !WALSplitter.splitLogFile(walDir, fs.getFileStatus(new Path(walDir, filename)), fs, conf, p,</span> |
| <span class="source-line-no">191</span><span id="line-191"> sequenceIdChecker, splitLogWorkerCoordination, factory, server)</span> |
| <span class="source-line-no">192</span><span id="line-192"> ) {</span> |
| <span class="source-line-no">193</span><span id="line-193"> return Status.PREEMPTED;</span> |
| <span class="source-line-no">194</span><span id="line-194"> }</span> |
| <span class="source-line-no">195</span><span id="line-195"> } catch (InterruptedIOException iioe) {</span> |
| <span class="source-line-no">196</span><span id="line-196"> LOG.warn("Resigning, interrupted splitting WAL {}", filename, iioe);</span> |
| <span class="source-line-no">197</span><span id="line-197"> return Status.RESIGNED;</span> |
| <span class="source-line-no">198</span><span id="line-198"> } catch (IOException e) {</span> |
| <span class="source-line-no">199</span><span id="line-199"> if (e instanceof FileNotFoundException) {</span> |
| <span class="source-line-no">200</span><span id="line-200"> // A wal file may not exist anymore. Nothing can be recovered so move on</span> |
| <span class="source-line-no">201</span><span id="line-201"> LOG.warn("Done, WAL {} does not exist anymore", filename, e);</span> |
| <span class="source-line-no">202</span><span id="line-202"> return Status.DONE;</span> |
| <span class="source-line-no">203</span><span id="line-203"> }</span> |
| <span class="source-line-no">204</span><span id="line-204"> Throwable cause = e.getCause();</span> |
| <span class="source-line-no">205</span><span id="line-205"> if (</span> |
| <span class="source-line-no">206</span><span id="line-206"> e instanceof RetriesExhaustedException && (cause instanceof NotServingRegionException</span> |
| <span class="source-line-no">207</span><span id="line-207"> || cause instanceof ConnectException || cause instanceof SocketTimeoutException)</span> |
| <span class="source-line-no">208</span><span id="line-208"> ) {</span> |
| <span class="source-line-no">209</span><span id="line-209"> LOG.warn("Resigning, can't connect to target regionserver splitting WAL {}", filename, e);</span> |
| <span class="source-line-no">210</span><span id="line-210"> return Status.RESIGNED;</span> |
| <span class="source-line-no">211</span><span id="line-211"> } else if (cause instanceof InterruptedException) {</span> |
| <span class="source-line-no">212</span><span id="line-212"> LOG.warn("Resigning, interrupted splitting WAL {}", filename, e);</span> |
| <span class="source-line-no">213</span><span id="line-213"> return Status.RESIGNED;</span> |
| <span class="source-line-no">214</span><span id="line-214"> }</span> |
| <span class="source-line-no">215</span><span id="line-215"> LOG.warn("Error splitting WAL {}", filename, e);</span> |
| <span class="source-line-no">216</span><span id="line-216"> return Status.ERR;</span> |
| <span class="source-line-no">217</span><span id="line-217"> }</span> |
| <span class="source-line-no">218</span><span id="line-218"> LOG.debug("Done splitting WAL {}", filename);</span> |
| <span class="source-line-no">219</span><span id="line-219"> return Status.DONE;</span> |
| <span class="source-line-no">220</span><span id="line-220"> }</span> |
| <span class="source-line-no">221</span><span id="line-221"></span> |
| <span class="source-line-no">222</span><span id="line-222"> @Override</span> |
| <span class="source-line-no">223</span><span id="line-223"> public void run() {</span> |
| <span class="source-line-no">224</span><span id="line-224"> try {</span> |
| <span class="source-line-no">225</span><span id="line-225"> LOG.info("SplitLogWorker " + server.getServerName() + " starting");</span> |
| <span class="source-line-no">226</span><span id="line-226"> coordination.registerListener();</span> |
| <span class="source-line-no">227</span><span id="line-227"> // wait for Coordination Engine is ready</span> |
| <span class="source-line-no">228</span><span id="line-228"> boolean res = false;</span> |
| <span class="source-line-no">229</span><span id="line-229"> while (!res && !coordination.isStop()) {</span> |
| <span class="source-line-no">230</span><span id="line-230"> res = coordination.isReady();</span> |
| <span class="source-line-no">231</span><span id="line-231"> }</span> |
| <span class="source-line-no">232</span><span id="line-232"> if (!coordination.isStop()) {</span> |
| <span class="source-line-no">233</span><span id="line-233"> coordination.taskLoop();</span> |
| <span class="source-line-no">234</span><span id="line-234"> }</span> |
| <span class="source-line-no">235</span><span id="line-235"> } catch (Throwable t) {</span> |
| <span class="source-line-no">236</span><span id="line-236"> if (ExceptionUtil.isInterrupt(t)) {</span> |
| <span class="source-line-no">237</span><span id="line-237"> LOG.info("SplitLogWorker interrupted. Exiting. "</span> |
| <span class="source-line-no">238</span><span id="line-238"> + (coordination.isStop() ? "" : " (ERROR: exitWorker is not set, exiting anyway)"));</span> |
| <span class="source-line-no">239</span><span id="line-239"> } else {</span> |
| <span class="source-line-no">240</span><span id="line-240"> // only a logical error can cause here. Printing it out</span> |
| <span class="source-line-no">241</span><span id="line-241"> // to make debugging easier</span> |
| <span class="source-line-no">242</span><span id="line-242"> LOG.error("unexpected error ", t);</span> |
| <span class="source-line-no">243</span><span id="line-243"> }</span> |
| <span class="source-line-no">244</span><span id="line-244"> } finally {</span> |
| <span class="source-line-no">245</span><span id="line-245"> coordination.removeListener();</span> |
| <span class="source-line-no">246</span><span id="line-246"> LOG.info("SplitLogWorker " + server.getServerName() + " exiting");</span> |
| <span class="source-line-no">247</span><span id="line-247"> }</span> |
| <span class="source-line-no">248</span><span id="line-248"> }</span> |
| <span class="source-line-no">249</span><span id="line-249"></span> |
| <span class="source-line-no">250</span><span id="line-250"> /**</span> |
| <span class="source-line-no">251</span><span id="line-251"> * If the worker is doing a task i.e. splitting a log file then stop the task. It doesn't exit the</span> |
| <span class="source-line-no">252</span><span id="line-252"> * worker thread.</span> |
| <span class="source-line-no">253</span><span id="line-253"> */</span> |
| <span class="source-line-no">254</span><span id="line-254"> public void stopTask() {</span> |
| <span class="source-line-no">255</span><span id="line-255"> LOG.info("Sending interrupt to stop the worker thread");</span> |
| <span class="source-line-no">256</span><span id="line-256"> worker.interrupt(); // TODO interrupt often gets swallowed, do what else?</span> |
| <span class="source-line-no">257</span><span id="line-257"> }</span> |
| <span class="source-line-no">258</span><span id="line-258"></span> |
| <span class="source-line-no">259</span><span id="line-259"> /**</span> |
| <span class="source-line-no">260</span><span id="line-260"> * start the SplitLogWorker thread</span> |
| <span class="source-line-no">261</span><span id="line-261"> */</span> |
| <span class="source-line-no">262</span><span id="line-262"> public void start() {</span> |
| <span class="source-line-no">263</span><span id="line-263"> worker = new Thread(null, this, "SplitLogWorker-" + server.getServerName().toShortString());</span> |
| <span class="source-line-no">264</span><span id="line-264"> worker.start();</span> |
| <span class="source-line-no">265</span><span id="line-265"> }</span> |
| <span class="source-line-no">266</span><span id="line-266"></span> |
| <span class="source-line-no">267</span><span id="line-267"> /**</span> |
| <span class="source-line-no">268</span><span id="line-268"> * stop the SplitLogWorker thread</span> |
| <span class="source-line-no">269</span><span id="line-269"> */</span> |
| <span class="source-line-no">270</span><span id="line-270"> public void stop() {</span> |
| <span class="source-line-no">271</span><span id="line-271"> coordination.stopProcessingTasks();</span> |
| <span class="source-line-no">272</span><span id="line-272"> stopTask();</span> |
| <span class="source-line-no">273</span><span id="line-273"> }</span> |
| <span class="source-line-no">274</span><span id="line-274"></span> |
| <span class="source-line-no">275</span><span id="line-275"> /**</span> |
| <span class="source-line-no">276</span><span id="line-276"> * Objects implementing this interface actually do the task that has been acquired by a</span> |
| <span class="source-line-no">277</span><span id="line-277"> * {@link SplitLogWorker}. Since there isn't a water-tight guarantee that two workers will not be</span> |
| <span class="source-line-no">278</span><span id="line-278"> * executing the same task therefore it is better to have workers prepare the task and then have</span> |
| <span class="source-line-no">279</span><span id="line-279"> * the {@link org.apache.hadoop.hbase.master.SplitLogManager} commit the work in</span> |
| <span class="source-line-no">280</span><span id="line-280"> * SplitLogManager.TaskFinisher</span> |
| <span class="source-line-no">281</span><span id="line-281"> */</span> |
| <span class="source-line-no">282</span><span id="line-282"> @FunctionalInterface</span> |
| <span class="source-line-no">283</span><span id="line-283"> public interface TaskExecutor {</span> |
| <span class="source-line-no">284</span><span id="line-284"> enum Status {</span> |
| <span class="source-line-no">285</span><span id="line-285"> DONE(),</span> |
| <span class="source-line-no">286</span><span id="line-286"> ERR(),</span> |
| <span class="source-line-no">287</span><span id="line-287"> RESIGNED(),</span> |
| <span class="source-line-no">288</span><span id="line-288"> PREEMPTED()</span> |
| <span class="source-line-no">289</span><span id="line-289"> }</span> |
| <span class="source-line-no">290</span><span id="line-290"></span> |
| <span class="source-line-no">291</span><span id="line-291"> Status exec(String name, CancelableProgressable p);</span> |
| <span class="source-line-no">292</span><span id="line-292"> }</span> |
| <span class="source-line-no">293</span><span id="line-293"></span> |
| <span class="source-line-no">294</span><span id="line-294"> /**</span> |
| <span class="source-line-no">295</span><span id="line-295"> * Returns the number of tasks processed by coordination. This method is used by tests only</span> |
| <span class="source-line-no">296</span><span id="line-296"> */</span> |
| <span class="source-line-no">297</span><span id="line-297"> public int getTaskReadySeq() {</span> |
| <span class="source-line-no">298</span><span id="line-298"> return coordination.getTaskReadySeq();</span> |
| <span class="source-line-no">299</span><span id="line-299"> }</span> |
| <span class="source-line-no">300</span><span id="line-300">}</span> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </main> |
| </body> |
| </html> |