blob: e037eeb78660c9156c09d1f6f92b841dc2e080c9 [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.DataInput;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.DataOutput;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.EOFException;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.FileNotFoundException;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.io.IOException;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.time.Instant;<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.Collections;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import java.util.List;<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.LocatedFileStatus;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.Path;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.RemoteIterator;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.util.LeaseNotRecoveredException;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.wal.AbstractFSWALProvider;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.wal.WAL;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.wal.WAL.Entry;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.wal.WALEdit;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.wal.WALFactory;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.wal.WALKey;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.wal.WALStreamReader;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.io.Writable;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.mapreduce.InputFormat;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.mapreduce.InputSplit;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.mapreduce.RecordReader;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.hadoop.mapreduce.security.TokenCache;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.hadoop.util.StringUtils;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.slf4j.Logger;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.slf4j.LoggerFactory;<a name="line.54"></a>
<span class="sourceLineNo">055</span><a name="line.55"></a>
<span class="sourceLineNo">056</span>/**<a name="line.56"></a>
<span class="sourceLineNo">057</span> * Simple {@link InputFormat} for {@link org.apache.hadoop.hbase.wal.WAL} files.<a name="line.57"></a>
<span class="sourceLineNo">058</span> */<a name="line.58"></a>
<span class="sourceLineNo">059</span>@InterfaceAudience.Public<a name="line.59"></a>
<span class="sourceLineNo">060</span>public class WALInputFormat extends InputFormat&lt;WALKey, WALEdit&gt; {<a name="line.60"></a>
<span class="sourceLineNo">061</span> private static final Logger LOG = LoggerFactory.getLogger(WALInputFormat.class);<a name="line.61"></a>
<span class="sourceLineNo">062</span><a name="line.62"></a>
<span class="sourceLineNo">063</span> public static final String START_TIME_KEY = "wal.start.time";<a name="line.63"></a>
<span class="sourceLineNo">064</span> public static final String END_TIME_KEY = "wal.end.time";<a name="line.64"></a>
<span class="sourceLineNo">065</span><a name="line.65"></a>
<span class="sourceLineNo">066</span> /**<a name="line.66"></a>
<span class="sourceLineNo">067</span> * {@link InputSplit} for {@link WAL} files. Each split represent exactly one log file.<a name="line.67"></a>
<span class="sourceLineNo">068</span> */<a name="line.68"></a>
<span class="sourceLineNo">069</span> static class WALSplit extends InputSplit implements Writable {<a name="line.69"></a>
<span class="sourceLineNo">070</span> private String logFileName;<a name="line.70"></a>
<span class="sourceLineNo">071</span> private long fileSize;<a name="line.71"></a>
<span class="sourceLineNo">072</span> private long startTime;<a name="line.72"></a>
<span class="sourceLineNo">073</span> private long endTime;<a name="line.73"></a>
<span class="sourceLineNo">074</span><a name="line.74"></a>
<span class="sourceLineNo">075</span> /** for serialization */<a name="line.75"></a>
<span class="sourceLineNo">076</span> public WALSplit() {<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> /**<a name="line.79"></a>
<span class="sourceLineNo">080</span> * Represent an WALSplit, i.e. a single WAL file. Start- and EndTime are managed by the split,<a name="line.80"></a>
<span class="sourceLineNo">081</span> * so that WAL files can be filtered before WALEdits are passed to the mapper(s).<a name="line.81"></a>
<span class="sourceLineNo">082</span> */<a name="line.82"></a>
<span class="sourceLineNo">083</span> public WALSplit(String logFileName, long fileSize, long startTime, long endTime) {<a name="line.83"></a>
<span class="sourceLineNo">084</span> this.logFileName = logFileName;<a name="line.84"></a>
<span class="sourceLineNo">085</span> this.fileSize = fileSize;<a name="line.85"></a>
<span class="sourceLineNo">086</span> this.startTime = startTime;<a name="line.86"></a>
<span class="sourceLineNo">087</span> this.endTime = endTime;<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> @Override<a name="line.90"></a>
<span class="sourceLineNo">091</span> public long getLength() throws IOException, InterruptedException {<a name="line.91"></a>
<span class="sourceLineNo">092</span> return fileSize;<a name="line.92"></a>
<span class="sourceLineNo">093</span> }<a name="line.93"></a>
<span class="sourceLineNo">094</span><a name="line.94"></a>
<span class="sourceLineNo">095</span> @Override<a name="line.95"></a>
<span class="sourceLineNo">096</span> public String[] getLocations() throws IOException, InterruptedException {<a name="line.96"></a>
<span class="sourceLineNo">097</span> // TODO: Find the data node with the most blocks for this WAL?<a name="line.97"></a>
<span class="sourceLineNo">098</span> return new String[] {};<a name="line.98"></a>
<span class="sourceLineNo">099</span> }<a name="line.99"></a>
<span class="sourceLineNo">100</span><a name="line.100"></a>
<span class="sourceLineNo">101</span> public String getLogFileName() {<a name="line.101"></a>
<span class="sourceLineNo">102</span> return logFileName;<a name="line.102"></a>
<span class="sourceLineNo">103</span> }<a name="line.103"></a>
<span class="sourceLineNo">104</span><a name="line.104"></a>
<span class="sourceLineNo">105</span> public long getStartTime() {<a name="line.105"></a>
<span class="sourceLineNo">106</span> return startTime;<a name="line.106"></a>
<span class="sourceLineNo">107</span> }<a name="line.107"></a>
<span class="sourceLineNo">108</span><a name="line.108"></a>
<span class="sourceLineNo">109</span> public long getEndTime() {<a name="line.109"></a>
<span class="sourceLineNo">110</span> return endTime;<a name="line.110"></a>
<span class="sourceLineNo">111</span> }<a name="line.111"></a>
<span class="sourceLineNo">112</span><a name="line.112"></a>
<span class="sourceLineNo">113</span> @Override<a name="line.113"></a>
<span class="sourceLineNo">114</span> public void readFields(DataInput in) throws IOException {<a name="line.114"></a>
<span class="sourceLineNo">115</span> logFileName = in.readUTF();<a name="line.115"></a>
<span class="sourceLineNo">116</span> fileSize = in.readLong();<a name="line.116"></a>
<span class="sourceLineNo">117</span> startTime = in.readLong();<a name="line.117"></a>
<span class="sourceLineNo">118</span> endTime = in.readLong();<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> @Override<a name="line.121"></a>
<span class="sourceLineNo">122</span> public void write(DataOutput out) throws IOException {<a name="line.122"></a>
<span class="sourceLineNo">123</span> out.writeUTF(logFileName);<a name="line.123"></a>
<span class="sourceLineNo">124</span> out.writeLong(fileSize);<a name="line.124"></a>
<span class="sourceLineNo">125</span> out.writeLong(startTime);<a name="line.125"></a>
<span class="sourceLineNo">126</span> out.writeLong(endTime);<a name="line.126"></a>
<span class="sourceLineNo">127</span> }<a name="line.127"></a>
<span class="sourceLineNo">128</span><a name="line.128"></a>
<span class="sourceLineNo">129</span> @Override<a name="line.129"></a>
<span class="sourceLineNo">130</span> public String toString() {<a name="line.130"></a>
<span class="sourceLineNo">131</span> return logFileName + " (" + startTime + ":" + endTime + ") length:" + fileSize;<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> /**<a name="line.135"></a>
<span class="sourceLineNo">136</span> * {@link RecordReader} for an {@link WAL} file. Implementation shared with deprecated<a name="line.136"></a>
<span class="sourceLineNo">137</span> * HLogInputFormat.<a name="line.137"></a>
<span class="sourceLineNo">138</span> */<a name="line.138"></a>
<span class="sourceLineNo">139</span> static abstract class WALRecordReader&lt;K extends WALKey&gt; extends RecordReader&lt;K, WALEdit&gt; {<a name="line.139"></a>
<span class="sourceLineNo">140</span> private WALStreamReader reader = null;<a name="line.140"></a>
<span class="sourceLineNo">141</span> // visible until we can remove the deprecated HLogInputFormat<a name="line.141"></a>
<span class="sourceLineNo">142</span> Entry currentEntry = new Entry();<a name="line.142"></a>
<span class="sourceLineNo">143</span> private long startTime;<a name="line.143"></a>
<span class="sourceLineNo">144</span> private long endTime;<a name="line.144"></a>
<span class="sourceLineNo">145</span> private Configuration conf;<a name="line.145"></a>
<span class="sourceLineNo">146</span> private Path logFile;<a name="line.146"></a>
<span class="sourceLineNo">147</span> private long currentPos;<a name="line.147"></a>
<span class="sourceLineNo">148</span><a name="line.148"></a>
<span class="sourceLineNo">149</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "DCN_NULLPOINTER_EXCEPTION",<a name="line.149"></a>
<span class="sourceLineNo">150</span> justification = "HDFS-4380")<a name="line.150"></a>
<span class="sourceLineNo">151</span> private WALStreamReader openReader(Path path, long startPosition) throws IOException {<a name="line.151"></a>
<span class="sourceLineNo">152</span> long retryInterval = 2000; // 2 sec<a name="line.152"></a>
<span class="sourceLineNo">153</span> int maxAttempts = 30;<a name="line.153"></a>
<span class="sourceLineNo">154</span> int attempt = 0;<a name="line.154"></a>
<span class="sourceLineNo">155</span> Exception ee = null;<a name="line.155"></a>
<span class="sourceLineNo">156</span> WALStreamReader reader = null;<a name="line.156"></a>
<span class="sourceLineNo">157</span> while (reader == null &amp;&amp; attempt++ &lt; maxAttempts) {<a name="line.157"></a>
<span class="sourceLineNo">158</span> try {<a name="line.158"></a>
<span class="sourceLineNo">159</span> // Detect if this is a new file, if so get a new reader else<a name="line.159"></a>
<span class="sourceLineNo">160</span> // reset the current reader so that we see the new data<a name="line.160"></a>
<span class="sourceLineNo">161</span> reader =<a name="line.161"></a>
<span class="sourceLineNo">162</span> WALFactory.createStreamReader(path.getFileSystem(conf), path, conf, startPosition);<a name="line.162"></a>
<span class="sourceLineNo">163</span> return reader;<a name="line.163"></a>
<span class="sourceLineNo">164</span> } catch (LeaseNotRecoveredException lnre) {<a name="line.164"></a>
<span class="sourceLineNo">165</span> // HBASE-15019 the WAL was not closed due to some hiccup.<a name="line.165"></a>
<span class="sourceLineNo">166</span> LOG.warn("Try to recover the WAL lease " + path, lnre);<a name="line.166"></a>
<span class="sourceLineNo">167</span> AbstractFSWALProvider.recoverLease(conf, path);<a name="line.167"></a>
<span class="sourceLineNo">168</span> reader = null;<a name="line.168"></a>
<span class="sourceLineNo">169</span> ee = lnre;<a name="line.169"></a>
<span class="sourceLineNo">170</span> } catch (NullPointerException npe) {<a name="line.170"></a>
<span class="sourceLineNo">171</span> // Workaround for race condition in HDFS-4380<a name="line.171"></a>
<span class="sourceLineNo">172</span> // which throws a NPE if we open a file before any data node has the most recent block<a name="line.172"></a>
<span class="sourceLineNo">173</span> // Just sleep and retry. Will require re-reading compressed WALs for compressionContext.<a name="line.173"></a>
<span class="sourceLineNo">174</span> LOG.warn("Got NPE opening reader, will retry.");<a name="line.174"></a>
<span class="sourceLineNo">175</span> reader = null;<a name="line.175"></a>
<span class="sourceLineNo">176</span> ee = npe;<a name="line.176"></a>
<span class="sourceLineNo">177</span> }<a name="line.177"></a>
<span class="sourceLineNo">178</span> if (reader == null) {<a name="line.178"></a>
<span class="sourceLineNo">179</span> // sleep before next attempt<a name="line.179"></a>
<span class="sourceLineNo">180</span> try {<a name="line.180"></a>
<span class="sourceLineNo">181</span> Thread.sleep(retryInterval);<a name="line.181"></a>
<span class="sourceLineNo">182</span> } catch (InterruptedException e) {<a name="line.182"></a>
<span class="sourceLineNo">183</span> Thread.currentThread().interrupt();<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> }<a name="line.186"></a>
<span class="sourceLineNo">187</span> throw new IOException("Could not open reader", ee);<a name="line.187"></a>
<span class="sourceLineNo">188</span> }<a name="line.188"></a>
<span class="sourceLineNo">189</span><a name="line.189"></a>
<span class="sourceLineNo">190</span> @Override<a name="line.190"></a>
<span class="sourceLineNo">191</span> public void initialize(InputSplit split, TaskAttemptContext context)<a name="line.191"></a>
<span class="sourceLineNo">192</span> throws IOException, InterruptedException {<a name="line.192"></a>
<span class="sourceLineNo">193</span> WALSplit hsplit = (WALSplit) split;<a name="line.193"></a>
<span class="sourceLineNo">194</span> logFile = new Path(hsplit.getLogFileName());<a name="line.194"></a>
<span class="sourceLineNo">195</span> conf = context.getConfiguration();<a name="line.195"></a>
<span class="sourceLineNo">196</span> LOG.info("Opening {} for {}", logFile, split);<a name="line.196"></a>
<span class="sourceLineNo">197</span> openReader(logFile);<a name="line.197"></a>
<span class="sourceLineNo">198</span> this.startTime = hsplit.getStartTime();<a name="line.198"></a>
<span class="sourceLineNo">199</span> this.endTime = hsplit.getEndTime();<a name="line.199"></a>
<span class="sourceLineNo">200</span> }<a name="line.200"></a>
<span class="sourceLineNo">201</span><a name="line.201"></a>
<span class="sourceLineNo">202</span> private void openReader(Path path) throws IOException {<a name="line.202"></a>
<span class="sourceLineNo">203</span> closeReader();<a name="line.203"></a>
<span class="sourceLineNo">204</span> reader = openReader(path, currentPos &gt; 0 ? currentPos : -1);<a name="line.204"></a>
<span class="sourceLineNo">205</span> setCurrentPath(path);<a name="line.205"></a>
<span class="sourceLineNo">206</span> }<a name="line.206"></a>
<span class="sourceLineNo">207</span><a name="line.207"></a>
<span class="sourceLineNo">208</span> private void setCurrentPath(Path path) {<a name="line.208"></a>
<span class="sourceLineNo">209</span> this.logFile = path;<a name="line.209"></a>
<span class="sourceLineNo">210</span> }<a name="line.210"></a>
<span class="sourceLineNo">211</span><a name="line.211"></a>
<span class="sourceLineNo">212</span> private void closeReader() throws IOException {<a name="line.212"></a>
<span class="sourceLineNo">213</span> if (reader != null) {<a name="line.213"></a>
<span class="sourceLineNo">214</span> reader.close();<a name="line.214"></a>
<span class="sourceLineNo">215</span> reader = null;<a name="line.215"></a>
<span class="sourceLineNo">216</span> }<a name="line.216"></a>
<span class="sourceLineNo">217</span> }<a name="line.217"></a>
<span class="sourceLineNo">218</span><a name="line.218"></a>
<span class="sourceLineNo">219</span> @Override<a name="line.219"></a>
<span class="sourceLineNo">220</span> public boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.220"></a>
<span class="sourceLineNo">221</span> if (reader == null) {<a name="line.221"></a>
<span class="sourceLineNo">222</span> return false;<a name="line.222"></a>
<span class="sourceLineNo">223</span> }<a name="line.223"></a>
<span class="sourceLineNo">224</span> this.currentPos = reader.getPosition();<a name="line.224"></a>
<span class="sourceLineNo">225</span> Entry temp;<a name="line.225"></a>
<span class="sourceLineNo">226</span> long i = -1;<a name="line.226"></a>
<span class="sourceLineNo">227</span> try {<a name="line.227"></a>
<span class="sourceLineNo">228</span> do {<a name="line.228"></a>
<span class="sourceLineNo">229</span> // skip older entries<a name="line.229"></a>
<span class="sourceLineNo">230</span> try {<a name="line.230"></a>
<span class="sourceLineNo">231</span> temp = reader.next(currentEntry);<a name="line.231"></a>
<span class="sourceLineNo">232</span> i++;<a name="line.232"></a>
<span class="sourceLineNo">233</span> } catch (EOFException x) {<a name="line.233"></a>
<span class="sourceLineNo">234</span> LOG.warn("Corrupted entry detected. Ignoring the rest of the file."<a name="line.234"></a>
<span class="sourceLineNo">235</span> + " (This is normal when a RegionServer crashed.)");<a name="line.235"></a>
<span class="sourceLineNo">236</span> return false;<a name="line.236"></a>
<span class="sourceLineNo">237</span> }<a name="line.237"></a>
<span class="sourceLineNo">238</span> } while (temp != null &amp;&amp; temp.getKey().getWriteTime() &lt; startTime);<a name="line.238"></a>
<span class="sourceLineNo">239</span><a name="line.239"></a>
<span class="sourceLineNo">240</span> if (temp == null) {<a name="line.240"></a>
<span class="sourceLineNo">241</span> if (i &gt; 0) {<a name="line.241"></a>
<span class="sourceLineNo">242</span> LOG.info("Skipped " + i + " entries.");<a name="line.242"></a>
<span class="sourceLineNo">243</span> }<a name="line.243"></a>
<span class="sourceLineNo">244</span> LOG.info("Reached end of file.");<a name="line.244"></a>
<span class="sourceLineNo">245</span> return false;<a name="line.245"></a>
<span class="sourceLineNo">246</span> } else if (i &gt; 0) {<a name="line.246"></a>
<span class="sourceLineNo">247</span> LOG.info("Skipped " + i + " entries, until ts: " + temp.getKey().getWriteTime() + ".");<a name="line.247"></a>
<span class="sourceLineNo">248</span> }<a name="line.248"></a>
<span class="sourceLineNo">249</span> boolean res = temp.getKey().getWriteTime() &lt;= endTime;<a name="line.249"></a>
<span class="sourceLineNo">250</span> if (!res) {<a name="line.250"></a>
<span class="sourceLineNo">251</span> LOG.info(<a name="line.251"></a>
<span class="sourceLineNo">252</span> "Reached ts: " + temp.getKey().getWriteTime() + " ignoring the rest of the file.");<a name="line.252"></a>
<span class="sourceLineNo">253</span> }<a name="line.253"></a>
<span class="sourceLineNo">254</span> return res;<a name="line.254"></a>
<span class="sourceLineNo">255</span> } catch (IOException e) {<a name="line.255"></a>
<span class="sourceLineNo">256</span> Path archivedLog = AbstractFSWALProvider.findArchivedLog(logFile, conf);<a name="line.256"></a>
<span class="sourceLineNo">257</span> // archivedLog can be null if unable to locate in archiveDir.<a name="line.257"></a>
<span class="sourceLineNo">258</span> if (archivedLog != null) {<a name="line.258"></a>
<span class="sourceLineNo">259</span> openReader(archivedLog);<a name="line.259"></a>
<span class="sourceLineNo">260</span> // Try call again in recursion<a name="line.260"></a>
<span class="sourceLineNo">261</span> return nextKeyValue();<a name="line.261"></a>
<span class="sourceLineNo">262</span> } else {<a name="line.262"></a>
<span class="sourceLineNo">263</span> throw e;<a name="line.263"></a>
<span class="sourceLineNo">264</span> }<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><a name="line.267"></a>
<span class="sourceLineNo">268</span> @Override<a name="line.268"></a>
<span class="sourceLineNo">269</span> public WALEdit getCurrentValue() throws IOException, InterruptedException {<a name="line.269"></a>
<span class="sourceLineNo">270</span> return currentEntry.getEdit();<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> @Override<a name="line.273"></a>
<span class="sourceLineNo">274</span> public float getProgress() throws IOException, InterruptedException {<a name="line.274"></a>
<span class="sourceLineNo">275</span> // N/A depends on total number of entries, which is unknown<a name="line.275"></a>
<span class="sourceLineNo">276</span> return 0;<a name="line.276"></a>
<span class="sourceLineNo">277</span> }<a name="line.277"></a>
<span class="sourceLineNo">278</span><a name="line.278"></a>
<span class="sourceLineNo">279</span> @Override<a name="line.279"></a>
<span class="sourceLineNo">280</span> public void close() throws IOException {<a name="line.280"></a>
<span class="sourceLineNo">281</span> LOG.info("Closing reader");<a name="line.281"></a>
<span class="sourceLineNo">282</span> if (reader != null) {<a name="line.282"></a>
<span class="sourceLineNo">283</span> this.reader.close();<a name="line.283"></a>
<span class="sourceLineNo">284</span> }<a name="line.284"></a>
<span class="sourceLineNo">285</span> }<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> * handler for non-deprecated WALKey version. fold into WALRecordReader once we no longer need to<a name="line.289"></a>
<span class="sourceLineNo">290</span> * support HLogInputFormat.<a name="line.290"></a>
<span class="sourceLineNo">291</span> */<a name="line.291"></a>
<span class="sourceLineNo">292</span> static class WALKeyRecordReader extends WALRecordReader&lt;WALKey&gt; {<a name="line.292"></a>
<span class="sourceLineNo">293</span> @Override<a name="line.293"></a>
<span class="sourceLineNo">294</span> public WALKey getCurrentKey() throws IOException, InterruptedException {<a name="line.294"></a>
<span class="sourceLineNo">295</span> return currentEntry.getKey();<a name="line.295"></a>
<span class="sourceLineNo">296</span> }<a name="line.296"></a>
<span class="sourceLineNo">297</span> }<a name="line.297"></a>
<span class="sourceLineNo">298</span><a name="line.298"></a>
<span class="sourceLineNo">299</span> @Override<a name="line.299"></a>
<span class="sourceLineNo">300</span> public List&lt;InputSplit&gt; getSplits(JobContext context) throws IOException, InterruptedException {<a name="line.300"></a>
<span class="sourceLineNo">301</span> return getSplits(context, START_TIME_KEY, END_TIME_KEY);<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> /**<a name="line.304"></a>
<span class="sourceLineNo">305</span> * implementation shared with deprecated HLogInputFormat<a name="line.305"></a>
<span class="sourceLineNo">306</span> */<a name="line.306"></a>
<span class="sourceLineNo">307</span> List&lt;InputSplit&gt; getSplits(final JobContext context, final String startKey, final String endKey)<a name="line.307"></a>
<span class="sourceLineNo">308</span> throws IOException, InterruptedException {<a name="line.308"></a>
<span class="sourceLineNo">309</span> Configuration conf = context.getConfiguration();<a name="line.309"></a>
<span class="sourceLineNo">310</span> boolean ignoreMissing = conf.getBoolean(WALPlayer.IGNORE_MISSING_FILES, false);<a name="line.310"></a>
<span class="sourceLineNo">311</span> Path[] inputPaths = getInputPaths(conf);<a name="line.311"></a>
<span class="sourceLineNo">312</span> // get delegation token for the filesystem<a name="line.312"></a>
<span class="sourceLineNo">313</span> TokenCache.obtainTokensForNamenodes(context.getCredentials(), inputPaths, conf);<a name="line.313"></a>
<span class="sourceLineNo">314</span> long startTime = conf.getLong(startKey, Long.MIN_VALUE);<a name="line.314"></a>
<span class="sourceLineNo">315</span> long endTime = conf.getLong(endKey, Long.MAX_VALUE);<a name="line.315"></a>
<span class="sourceLineNo">316</span><a name="line.316"></a>
<span class="sourceLineNo">317</span> List&lt;FileStatus&gt; allFiles = new ArrayList&lt;FileStatus&gt;();<a name="line.317"></a>
<span class="sourceLineNo">318</span> for (Path inputPath : inputPaths) {<a name="line.318"></a>
<span class="sourceLineNo">319</span> FileSystem fs = inputPath.getFileSystem(conf);<a name="line.319"></a>
<span class="sourceLineNo">320</span> try {<a name="line.320"></a>
<span class="sourceLineNo">321</span> List&lt;FileStatus&gt; files = getFiles(fs, inputPath, startTime, endTime);<a name="line.321"></a>
<span class="sourceLineNo">322</span> allFiles.addAll(files);<a name="line.322"></a>
<span class="sourceLineNo">323</span> } catch (FileNotFoundException e) {<a name="line.323"></a>
<span class="sourceLineNo">324</span> if (ignoreMissing) {<a name="line.324"></a>
<span class="sourceLineNo">325</span> LOG.warn("File " + inputPath + " is missing. Skipping it.");<a name="line.325"></a>
<span class="sourceLineNo">326</span> continue;<a name="line.326"></a>
<span class="sourceLineNo">327</span> }<a name="line.327"></a>
<span class="sourceLineNo">328</span> throw e;<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> List&lt;InputSplit&gt; splits = new ArrayList&lt;InputSplit&gt;(allFiles.size());<a name="line.331"></a>
<span class="sourceLineNo">332</span> for (FileStatus file : allFiles) {<a name="line.332"></a>
<span class="sourceLineNo">333</span> splits.add(new WALSplit(file.getPath().toString(), file.getLen(), startTime, endTime));<a name="line.333"></a>
<span class="sourceLineNo">334</span> }<a name="line.334"></a>
<span class="sourceLineNo">335</span> return splits;<a name="line.335"></a>
<span class="sourceLineNo">336</span> }<a name="line.336"></a>
<span class="sourceLineNo">337</span><a name="line.337"></a>
<span class="sourceLineNo">338</span> private Path[] getInputPaths(Configuration conf) {<a name="line.338"></a>
<span class="sourceLineNo">339</span> String inpDirs = conf.get(FileInputFormat.INPUT_DIR);<a name="line.339"></a>
<span class="sourceLineNo">340</span> return StringUtils<a name="line.340"></a>
<span class="sourceLineNo">341</span> .stringToPath(inpDirs.split(conf.get(WALPlayer.INPUT_FILES_SEPARATOR_KEY, ",")));<a name="line.341"></a>
<span class="sourceLineNo">342</span> }<a name="line.342"></a>
<span class="sourceLineNo">343</span><a name="line.343"></a>
<span class="sourceLineNo">344</span> /**<a name="line.344"></a>
<span class="sourceLineNo">345</span> * @param startTime If file looks like it has a timestamp in its name, we'll check if newer or<a name="line.345"></a>
<span class="sourceLineNo">346</span> * equal to this value else we will filter out the file. If name does not seem to<a name="line.346"></a>
<span class="sourceLineNo">347</span> * have a timestamp, we will just return it w/o filtering.<a name="line.347"></a>
<span class="sourceLineNo">348</span> * @param endTime If file looks like it has a timestamp in its name, we'll check if older or<a name="line.348"></a>
<span class="sourceLineNo">349</span> * equal to this value else we will filter out the file. If name does not seem to<a name="line.349"></a>
<span class="sourceLineNo">350</span> * have a timestamp, we will just return it w/o filtering.<a name="line.350"></a>
<span class="sourceLineNo">351</span> */<a name="line.351"></a>
<span class="sourceLineNo">352</span> private List&lt;FileStatus&gt; getFiles(FileSystem fs, Path dir, long startTime, long endTime)<a name="line.352"></a>
<span class="sourceLineNo">353</span> throws IOException {<a name="line.353"></a>
<span class="sourceLineNo">354</span> List&lt;FileStatus&gt; result = new ArrayList&lt;&gt;();<a name="line.354"></a>
<span class="sourceLineNo">355</span> LOG.debug("Scanning " + dir.toString() + " for WAL files");<a name="line.355"></a>
<span class="sourceLineNo">356</span> RemoteIterator&lt;LocatedFileStatus&gt; iter = fs.listLocatedStatus(dir);<a name="line.356"></a>
<span class="sourceLineNo">357</span> if (!iter.hasNext()) {<a name="line.357"></a>
<span class="sourceLineNo">358</span> return Collections.emptyList();<a name="line.358"></a>
<span class="sourceLineNo">359</span> }<a name="line.359"></a>
<span class="sourceLineNo">360</span> while (iter.hasNext()) {<a name="line.360"></a>
<span class="sourceLineNo">361</span> LocatedFileStatus file = iter.next();<a name="line.361"></a>
<span class="sourceLineNo">362</span> if (file.isDirectory()) {<a name="line.362"></a>
<span class="sourceLineNo">363</span> // Recurse into sub directories<a name="line.363"></a>
<span class="sourceLineNo">364</span> result.addAll(getFiles(fs, file.getPath(), startTime, endTime));<a name="line.364"></a>
<span class="sourceLineNo">365</span> } else {<a name="line.365"></a>
<span class="sourceLineNo">366</span> addFile(result, file, startTime, endTime);<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> // TODO: These results should be sorted? Results could be content of recovered.edits directory<a name="line.369"></a>
<span class="sourceLineNo">370</span> // -- null padded increasing numeric -- or a WAL file w/ timestamp suffix or timestamp and<a name="line.370"></a>
<span class="sourceLineNo">371</span> // then meta suffix. See AbstractFSWALProvider#WALStartTimeComparator<a name="line.371"></a>
<span class="sourceLineNo">372</span> return result;<a name="line.372"></a>
<span class="sourceLineNo">373</span> }<a name="line.373"></a>
<span class="sourceLineNo">374</span><a name="line.374"></a>
<span class="sourceLineNo">375</span> static void addFile(List&lt;FileStatus&gt; result, LocatedFileStatus lfs, long startTime,<a name="line.375"></a>
<span class="sourceLineNo">376</span> long endTime) {<a name="line.376"></a>
<span class="sourceLineNo">377</span> long timestamp = AbstractFSWALProvider.getTimestamp(lfs.getPath().getName());<a name="line.377"></a>
<span class="sourceLineNo">378</span> if (timestamp &gt; 0) {<a name="line.378"></a>
<span class="sourceLineNo">379</span> // Looks like a valid timestamp.<a name="line.379"></a>
<span class="sourceLineNo">380</span> if (timestamp &lt;= endTime &amp;&amp; timestamp &gt;= startTime) {<a name="line.380"></a>
<span class="sourceLineNo">381</span> LOG.info("Found {}", lfs.getPath());<a name="line.381"></a>
<span class="sourceLineNo">382</span> result.add(lfs);<a name="line.382"></a>
<span class="sourceLineNo">383</span> } else {<a name="line.383"></a>
<span class="sourceLineNo">384</span> LOG.info("Skipped {}, outside range [{}/{} - {}/{}]", lfs.getPath(), startTime,<a name="line.384"></a>
<span class="sourceLineNo">385</span> Instant.ofEpochMilli(startTime), endTime, Instant.ofEpochMilli(endTime));<a name="line.385"></a>
<span class="sourceLineNo">386</span> }<a name="line.386"></a>
<span class="sourceLineNo">387</span> } else {<a name="line.387"></a>
<span class="sourceLineNo">388</span> // If no timestamp, add it regardless.<a name="line.388"></a>
<span class="sourceLineNo">389</span> LOG.info("Found (no-timestamp!) {}", lfs);<a name="line.389"></a>
<span class="sourceLineNo">390</span> result.add(lfs);<a name="line.390"></a>
<span class="sourceLineNo">391</span> }<a name="line.391"></a>
<span class="sourceLineNo">392</span> }<a name="line.392"></a>
<span class="sourceLineNo">393</span><a name="line.393"></a>
<span class="sourceLineNo">394</span> @Override<a name="line.394"></a>
<span class="sourceLineNo">395</span> public RecordReader&lt;WALKey, WALEdit&gt; createRecordReader(InputSplit split,<a name="line.395"></a>
<span class="sourceLineNo">396</span> TaskAttemptContext context) throws IOException, InterruptedException {<a name="line.396"></a>
<span class="sourceLineNo">397</span> return new WALKeyRecordReader();<a name="line.397"></a>
<span class="sourceLineNo">398</span> }<a name="line.398"></a>
<span class="sourceLineNo">399</span>}<a name="line.399"></a>
</pre>
</div>
</body>
</html>