Repository: hbase-site Updated Branches: refs/heads/asf-site 41d850387 -> 7e7efd3eb
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/7e7efd3e/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestMultiTableHFileOutputFormat.html ---------------------------------------------------------------------- diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestMultiTableHFileOutputFormat.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestMultiTableHFileOutputFormat.html new file mode 100644 index 0000000..be80ab3 --- /dev/null +++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestMultiTableHFileOutputFormat.html @@ -0,0 +1,454 @@ +<!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 static org.junit.Assert.assertTrue;<a name="line.20"></a> +<span class="sourceLineNo">021</span><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.util.ArrayList;<a name="line.23"></a> +<span class="sourceLineNo">024</span>import java.util.Arrays;<a name="line.24"></a> +<span class="sourceLineNo">025</span>import java.util.Collections;<a name="line.25"></a> +<span class="sourceLineNo">026</span>import java.util.List;<a name="line.26"></a> +<span class="sourceLineNo">027</span>import java.util.Random;<a name="line.27"></a> +<span class="sourceLineNo">028</span>import java.util.TreeSet;<a name="line.28"></a> +<span class="sourceLineNo">029</span><a name="line.29"></a> +<span class="sourceLineNo">030</span>import org.apache.commons.logging.Log;<a name="line.30"></a> +<span class="sourceLineNo">031</span>import org.apache.commons.logging.LogFactory;<a name="line.31"></a> +<span class="sourceLineNo">032</span>import org.apache.hadoop.conf.Configuration;<a name="line.32"></a> +<span class="sourceLineNo">033</span>import org.apache.hadoop.fs.FileStatus;<a name="line.33"></a> +<span class="sourceLineNo">034</span>import org.apache.hadoop.fs.FileSystem;<a name="line.34"></a> +<span class="sourceLineNo">035</span>import org.apache.hadoop.fs.Path;<a name="line.35"></a> +<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.Cell;<a name="line.36"></a> +<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.37"></a> +<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.HBaseTestingUtility;<a name="line.38"></a> +<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.HColumnDescriptor;<a name="line.39"></a> +<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.HConstants;<a name="line.40"></a> +<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.HTableDescriptor;<a name="line.41"></a> +<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.42"></a> +<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.PerformanceEvaluation;<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.io.ImmutableBytesWritable;<a name="line.45"></a> +<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.io.compress.Compression;<a name="line.46"></a> +<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.io.encoding.DataBlockEncoding;<a name="line.47"></a> +<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.io.hfile.CacheConfig;<a name="line.48"></a> +<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.49"></a> +<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.testclassification.MediumTests;<a name="line.50"></a> +<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.51"></a> +<span class="sourceLineNo">052</span>import org.apache.hadoop.io.NullWritable;<a name="line.52"></a> +<span class="sourceLineNo">053</span>import org.apache.hadoop.mapred.FileOutputCommitter;<a name="line.53"></a> +<span class="sourceLineNo">054</span>import org.apache.hadoop.mapreduce.Job;<a name="line.54"></a> +<span class="sourceLineNo">055</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.55"></a> +<span class="sourceLineNo">056</span>import org.apache.hadoop.mapreduce.Reducer;<a name="line.56"></a> +<span class="sourceLineNo">057</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.57"></a> +<span class="sourceLineNo">058</span>import org.junit.Test;<a name="line.58"></a> +<span class="sourceLineNo">059</span>import org.junit.experimental.categories.Category;<a name="line.59"></a> +<span class="sourceLineNo">060</span><a name="line.60"></a> +<span class="sourceLineNo">061</span>/**<a name="line.61"></a> +<span class="sourceLineNo">062</span> * Test for{@link MultiTableHFileOutputFormat}. Sets up and runs a mapreduce job that output directories and<a name="line.62"></a> +<span class="sourceLineNo">063</span> * writes hfiles.<a name="line.63"></a> +<span class="sourceLineNo">064</span> */<a name="line.64"></a> +<span class="sourceLineNo">065</span>@Category(MediumTests.class)<a name="line.65"></a> +<span class="sourceLineNo">066</span>public class TestMultiTableHFileOutputFormat {<a name="line.66"></a> +<span class="sourceLineNo">067</span> private static final Log LOG = LogFactory.getLog(TestMultiTableHFileOutputFormat.class);<a name="line.67"></a> +<span class="sourceLineNo">068</span><a name="line.68"></a> +<span class="sourceLineNo">069</span> private HBaseTestingUtility util = new HBaseTestingUtility();<a name="line.69"></a> +<span class="sourceLineNo">070</span><a name="line.70"></a> +<span class="sourceLineNo">071</span> private static int ROWSPERSPLIT = 10;<a name="line.71"></a> +<span class="sourceLineNo">072</span><a name="line.72"></a> +<span class="sourceLineNo">073</span> private static final int KEYLEN_DEFAULT = 10;<a name="line.73"></a> +<span class="sourceLineNo">074</span> private static final String KEYLEN_CONF = "randomkv.key.length";<a name="line.74"></a> +<span class="sourceLineNo">075</span><a name="line.75"></a> +<span class="sourceLineNo">076</span> private static final int VALLEN_DEFAULT = 10;<a name="line.76"></a> +<span class="sourceLineNo">077</span> private static final String VALLEN_CONF = "randomkv.val.length";<a name="line.77"></a> +<span class="sourceLineNo">078</span><a name="line.78"></a> +<span class="sourceLineNo">079</span> private static final byte[][] TABLES =<a name="line.79"></a> +<span class="sourceLineNo">080</span> { Bytes.add(Bytes.toBytes(PerformanceEvaluation.TABLE_NAME), Bytes.toBytes("-1")),<a name="line.80"></a> +<span class="sourceLineNo">081</span> Bytes.add(Bytes.toBytes(PerformanceEvaluation.TABLE_NAME), Bytes.toBytes("-2")) };<a name="line.81"></a> +<span class="sourceLineNo">082</span><a name="line.82"></a> +<span class="sourceLineNo">083</span> private static final byte[][] FAMILIES =<a name="line.83"></a> +<span class="sourceLineNo">084</span> { Bytes.add(PerformanceEvaluation.FAMILY_NAME, Bytes.toBytes("-A")),<a name="line.84"></a> +<span class="sourceLineNo">085</span> Bytes.add(PerformanceEvaluation.FAMILY_NAME, Bytes.toBytes("-B")) };<a name="line.85"></a> +<span class="sourceLineNo">086</span><a name="line.86"></a> +<span class="sourceLineNo">087</span> private static final byte[] QUALIFIER = Bytes.toBytes("data");<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> * Run small MR job. this MR job will write HFile into<a name="line.90"></a> +<span class="sourceLineNo">091</span> * testWritingDataIntoHFiles/tableNames/columnFamilies/<a name="line.91"></a> +<span class="sourceLineNo">092</span> */<a name="line.92"></a> +<span class="sourceLineNo">093</span> @Test<a name="line.93"></a> +<span class="sourceLineNo">094</span> public void testWritingDataIntoHFiles() throws Exception {<a name="line.94"></a> +<span class="sourceLineNo">095</span> Configuration conf = util.getConfiguration();<a name="line.95"></a> +<span class="sourceLineNo">096</span> util.startMiniCluster();<a name="line.96"></a> +<span class="sourceLineNo">097</span> Path testDir = util.getDataTestDirOnTestFS("testWritingDataIntoHFiles");<a name="line.97"></a> +<span class="sourceLineNo">098</span> FileSystem fs = testDir.getFileSystem(conf);<a name="line.98"></a> +<span class="sourceLineNo">099</span> LOG.info("testWritingDataIntoHFiles dir writing to dir: " + testDir);<a name="line.99"></a> +<span class="sourceLineNo">100</span><a name="line.100"></a> +<span class="sourceLineNo">101</span> // Set down this value or we OOME in eclipse.<a name="line.101"></a> +<span class="sourceLineNo">102</span> conf.setInt("mapreduce.task.io.sort.mb", 20);<a name="line.102"></a> +<span class="sourceLineNo">103</span> // Write a few files by setting max file size.<a name="line.103"></a> +<span class="sourceLineNo">104</span> conf.setLong(HConstants.HREGION_MAX_FILESIZE, 64 * 1024);<a name="line.104"></a> +<span class="sourceLineNo">105</span><a name="line.105"></a> +<span class="sourceLineNo">106</span> try {<a name="line.106"></a> +<span class="sourceLineNo">107</span> Job job = Job.getInstance(conf, "testWritingDataIntoHFiles");<a name="line.107"></a> +<span class="sourceLineNo">108</span><a name="line.108"></a> +<span class="sourceLineNo">109</span> FileOutputFormat.setOutputPath(job, testDir);<a name="line.109"></a> +<span class="sourceLineNo">110</span><a name="line.110"></a> +<span class="sourceLineNo">111</span> job.setInputFormatClass(NMapInputFormat.class);<a name="line.111"></a> +<span class="sourceLineNo">112</span> job.setMapperClass(Random_TableKV_GeneratingMapper.class);<a name="line.112"></a> +<span class="sourceLineNo">113</span> job.setMapOutputKeyClass(ImmutableBytesWritable.class);<a name="line.113"></a> +<span class="sourceLineNo">114</span> job.setMapOutputValueClass(KeyValue.class);<a name="line.114"></a> +<span class="sourceLineNo">115</span> job.setReducerClass(Table_KeyValueSortReducer.class);<a name="line.115"></a> +<span class="sourceLineNo">116</span> job.setOutputFormatClass(MultiTableHFileOutputFormat.class);<a name="line.116"></a> +<span class="sourceLineNo">117</span> job.getConfiguration().setStrings("io.serializations", conf.get("io.serializations"),<a name="line.117"></a> +<span class="sourceLineNo">118</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.118"></a> +<span class="sourceLineNo">119</span> KeyValueSerialization.class.getName());<a name="line.119"></a> +<span class="sourceLineNo">120</span><a name="line.120"></a> +<span class="sourceLineNo">121</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.121"></a> +<span class="sourceLineNo">122</span> TableMapReduceUtil.initCredentials(job);<a name="line.122"></a> +<span class="sourceLineNo">123</span> LOG.info("\nStarting test testWritingDataIntoHFiles\n");<a name="line.123"></a> +<span class="sourceLineNo">124</span> assertTrue(job.waitForCompletion(true));<a name="line.124"></a> +<span class="sourceLineNo">125</span> LOG.info("\nWaiting on checking MapReduce output\n");<a name="line.125"></a> +<span class="sourceLineNo">126</span> assertTrue(checkMROutput(fs, testDir, 0));<a name="line.126"></a> +<span class="sourceLineNo">127</span> } finally {<a name="line.127"></a> +<span class="sourceLineNo">128</span> testDir.getFileSystem(conf).delete(testDir, true);<a name="line.128"></a> +<span class="sourceLineNo">129</span> util.shutdownMiniCluster();<a name="line.129"></a> +<span class="sourceLineNo">130</span> }<a name="line.130"></a> +<span class="sourceLineNo">131</span> }<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> * check whether create directory and hfiles as format designed in MultiHFilePartitioner<a name="line.134"></a> +<span class="sourceLineNo">135</span> * and also check whether the output file has same related configuration as created table<a name="line.135"></a> +<span class="sourceLineNo">136</span> */<a name="line.136"></a> +<span class="sourceLineNo">137</span> @Test<a name="line.137"></a> +<span class="sourceLineNo">138</span> public void testMultiHFilePartitioner() throws Exception {<a name="line.138"></a> +<span class="sourceLineNo">139</span> Configuration conf = util.getConfiguration();<a name="line.139"></a> +<span class="sourceLineNo">140</span> util.startMiniCluster();<a name="line.140"></a> +<span class="sourceLineNo">141</span> Path testDir = util.getDataTestDirOnTestFS("testMultiHFilePartitioner");<a name="line.141"></a> +<span class="sourceLineNo">142</span> FileSystem fs = testDir.getFileSystem(conf);<a name="line.142"></a> +<span class="sourceLineNo">143</span> LOG.info("testMultiHFilePartitioner dir writing to : " + testDir);<a name="line.143"></a> +<span class="sourceLineNo">144</span><a name="line.144"></a> +<span class="sourceLineNo">145</span> // Set down this value or we OOME in eclipse.<a name="line.145"></a> +<span class="sourceLineNo">146</span> conf.setInt("mapreduce.task.io.sort.mb", 20);<a name="line.146"></a> +<span class="sourceLineNo">147</span> // Write a few files by setting max file size.<a name="line.147"></a> +<span class="sourceLineNo">148</span> conf.setLong(HConstants.HREGION_MAX_FILESIZE, 64 * 1024);<a name="line.148"></a> +<span class="sourceLineNo">149</span><a name="line.149"></a> +<span class="sourceLineNo">150</span> // Create several tables for testing<a name="line.150"></a> +<span class="sourceLineNo">151</span> List<TableName> tables = new ArrayList<TableName>();<a name="line.151"></a> +<span class="sourceLineNo">152</span><a name="line.152"></a> +<span class="sourceLineNo">153</span> // to store splitKeys for TABLE[0] for testing;<a name="line.153"></a> +<span class="sourceLineNo">154</span> byte[][] testKeys = new byte[0][0];<a name="line.154"></a> +<span class="sourceLineNo">155</span> for (int i = 0; i < TABLES.length; i++) {<a name="line.155"></a> +<span class="sourceLineNo">156</span> TableName tableName = TableName.valueOf(TABLES[i]);<a name="line.156"></a> +<span class="sourceLineNo">157</span> byte[][] splitKeys = generateRandomSplitKeys(3);<a name="line.157"></a> +<span class="sourceLineNo">158</span> if (i == 0) {<a name="line.158"></a> +<span class="sourceLineNo">159</span> testKeys = splitKeys;<a name="line.159"></a> +<span class="sourceLineNo">160</span> }<a name="line.160"></a> +<span class="sourceLineNo">161</span> HTableDescriptor tableDescriptor = new HTableDescriptor(tableName);<a name="line.161"></a> +<span class="sourceLineNo">162</span> for (int j = 0; j < FAMILIES.length; j++) {<a name="line.162"></a> +<span class="sourceLineNo">163</span> HColumnDescriptor familyDescriptor = new HColumnDescriptor(FAMILIES[j]);<a name="line.163"></a> +<span class="sourceLineNo">164</span> //only set Tables[0] configuration, and specify compression type and DataBlockEncode<a name="line.164"></a> +<span class="sourceLineNo">165</span> if (i == 0) {<a name="line.165"></a> +<span class="sourceLineNo">166</span> familyDescriptor.setCompressionType(Compression.Algorithm.GZ);<a name="line.166"></a> +<span class="sourceLineNo">167</span> familyDescriptor.setDataBlockEncoding(DataBlockEncoding.FAST_DIFF);<a name="line.167"></a> +<span class="sourceLineNo">168</span> }<a name="line.168"></a> +<span class="sourceLineNo">169</span> tableDescriptor.addFamily(familyDescriptor);<a name="line.169"></a> +<span class="sourceLineNo">170</span> }<a name="line.170"></a> +<span class="sourceLineNo">171</span> util.createTable(tableDescriptor, splitKeys, conf);<a name="line.171"></a> +<span class="sourceLineNo">172</span> tables.add(tableName);<a name="line.172"></a> +<span class="sourceLineNo">173</span> }<a name="line.173"></a> +<span class="sourceLineNo">174</span> // set up for MapReduce job<a name="line.174"></a> +<span class="sourceLineNo">175</span> try {<a name="line.175"></a> +<span class="sourceLineNo">176</span> Job job = Job.getInstance(conf, "testMultiHFilePartitioner");<a name="line.176"></a> +<span class="sourceLineNo">177</span> FileOutputFormat.setOutputPath(job, testDir);<a name="line.177"></a> +<span class="sourceLineNo">178</span><a name="line.178"></a> +<span class="sourceLineNo">179</span> job.setInputFormatClass(NMapInputFormat.class);<a name="line.179"></a> +<span class="sourceLineNo">180</span> job.setMapperClass(Random_TableKV_GeneratingMapper.class);<a name="line.180"></a> +<span class="sourceLineNo">181</span> job.setMapOutputKeyClass(ImmutableBytesWritable.class);<a name="line.181"></a> +<span class="sourceLineNo">182</span> job.setMapOutputValueClass(KeyValue.class);<a name="line.182"></a> +<span class="sourceLineNo">183</span><a name="line.183"></a> +<span class="sourceLineNo">184</span> MultiTableHFileOutputFormat.configureIncrementalLoad(job, tables);<a name="line.184"></a> +<span class="sourceLineNo">185</span><a name="line.185"></a> +<span class="sourceLineNo">186</span> LOG.info("Starting test testWritingDataIntoHFiles");<a name="line.186"></a> +<span class="sourceLineNo">187</span> assertTrue(job.waitForCompletion(true));<a name="line.187"></a> +<span class="sourceLineNo">188</span> LOG.info("Waiting on checking MapReduce output");<a name="line.188"></a> +<span class="sourceLineNo">189</span> assertTrue(checkMROutput(fs, testDir, 0));<a name="line.189"></a> +<span class="sourceLineNo">190</span> assertTrue(checkFileConfAndSplitKeys(conf, fs, testDir, testKeys));<a name="line.190"></a> +<span class="sourceLineNo">191</span> } finally {<a name="line.191"></a> +<span class="sourceLineNo">192</span> for (int i = 0; i < TABLES.length; i++) {<a name="line.192"></a> +<span class="sourceLineNo">193</span> TableName tName = TableName.valueOf(TABLES[i]);<a name="line.193"></a> +<span class="sourceLineNo">194</span> util.deleteTable(tName);<a name="line.194"></a> +<span class="sourceLineNo">195</span> }<a name="line.195"></a> +<span class="sourceLineNo">196</span> fs.delete(testDir, true);<a name="line.196"></a> +<span class="sourceLineNo">197</span> fs.close();<a name="line.197"></a> +<span class="sourceLineNo">198</span> util.shutdownMiniCluster();<a name="line.198"></a> +<span class="sourceLineNo">199</span> }<a name="line.199"></a> +<span class="sourceLineNo">200</span> }<a name="line.200"></a> +<span class="sourceLineNo">201</span><a name="line.201"></a> +<span class="sourceLineNo">202</span> /**<a name="line.202"></a> +<span class="sourceLineNo">203</span> * check the output hfile has same configuration as created test table<a name="line.203"></a> +<span class="sourceLineNo">204</span> * and also check whether hfiles get split correctly<a name="line.204"></a> +<span class="sourceLineNo">205</span> * only check TABLES[0]<a name="line.205"></a> +<span class="sourceLineNo">206</span> */<a name="line.206"></a> +<span class="sourceLineNo">207</span> private boolean checkFileConfAndSplitKeys(Configuration conf, FileSystem fs, Path testDir, byte[][] splitKeys) throws IOException {<a name="line.207"></a> +<span class="sourceLineNo">208</span> FileStatus[] fStats = fs.listStatus(testDir);<a name="line.208"></a> +<span class="sourceLineNo">209</span> for (FileStatus stats : fStats) {<a name="line.209"></a> +<span class="sourceLineNo">210</span> if (stats.getPath().getName().equals(new String(TABLES[0]))) {<a name="line.210"></a> +<span class="sourceLineNo">211</span> FileStatus[] cfStats = fs.listStatus(stats.getPath());<a name="line.211"></a> +<span class="sourceLineNo">212</span> for (FileStatus cfstat : cfStats) {<a name="line.212"></a> +<span class="sourceLineNo">213</span> FileStatus[] hfStats = fs.listStatus(cfstat.getPath());<a name="line.213"></a> +<span class="sourceLineNo">214</span><a name="line.214"></a> +<span class="sourceLineNo">215</span> List<byte[]> firsttKeys = new ArrayList<byte[]>();<a name="line.215"></a> +<span class="sourceLineNo">216</span> List<byte[]> lastKeys = new ArrayList<byte[]>();<a name="line.216"></a> +<span class="sourceLineNo">217</span> for (FileStatus hfstat : hfStats) {<a name="line.217"></a> +<span class="sourceLineNo">218</span> if (HFile.isHFileFormat(fs, hfstat)) {<a name="line.218"></a> +<span class="sourceLineNo">219</span> HFile.Reader hfr =<a name="line.219"></a> +<span class="sourceLineNo">220</span> HFile.createReader(fs, hfstat.getPath(), new CacheConfig(conf), true, conf);<a name="line.220"></a> +<span class="sourceLineNo">221</span> if (!hfr.getDataBlockEncoding().equals(DataBlockEncoding.FAST_DIFF) || !hfr<a name="line.221"></a> +<span class="sourceLineNo">222</span> .getCompressionAlgorithm().equals(Compression.Algorithm.GZ)) return false;<a name="line.222"></a> +<span class="sourceLineNo">223</span> firsttKeys.add(hfr.getFirstRowKey());<a name="line.223"></a> +<span class="sourceLineNo">224</span> lastKeys.add(hfr.getLastRowKey());<a name="line.224"></a> +<span class="sourceLineNo">225</span> }<a name="line.225"></a> +<span class="sourceLineNo">226</span> }<a name="line.226"></a> +<span class="sourceLineNo">227</span> if (checkFileSplit(splitKeys, firsttKeys, lastKeys) == false) {<a name="line.227"></a> +<span class="sourceLineNo">228</span> return false;<a name="line.228"></a> +<span class="sourceLineNo">229</span> }<a name="line.229"></a> +<span class="sourceLineNo">230</span> }<a name="line.230"></a> +<span class="sourceLineNo">231</span> }<a name="line.231"></a> +<span class="sourceLineNo">232</span> }<a name="line.232"></a> +<span class="sourceLineNo">233</span> return true;<a name="line.233"></a> +<span class="sourceLineNo">234</span> }<a name="line.234"></a> +<span class="sourceLineNo">235</span><a name="line.235"></a> +<span class="sourceLineNo">236</span> /**<a name="line.236"></a> +<span class="sourceLineNo">237</span> * Check whether the Hfile has been split by region boundaries<a name="line.237"></a> +<span class="sourceLineNo">238</span> * @param splitKeys split keys for that table<a name="line.238"></a> +<span class="sourceLineNo">239</span> * @param firstKeys first rowKey for hfiles<a name="line.239"></a> +<span class="sourceLineNo">240</span> * @param lastKeys last rowKey for hfiles<a name="line.240"></a> +<span class="sourceLineNo">241</span> */<a name="line.241"></a> +<span class="sourceLineNo">242</span> private boolean checkFileSplit(byte[][] splitKeys, List<byte[]> firstKeys, List<byte[]> lastKeys) {<a name="line.242"></a> +<span class="sourceLineNo">243</span> Collections.sort(firstKeys, Bytes.BYTES_RAWCOMPARATOR);<a name="line.243"></a> +<span class="sourceLineNo">244</span> Collections.sort(lastKeys, Bytes.BYTES_RAWCOMPARATOR);<a name="line.244"></a> +<span class="sourceLineNo">245</span> Arrays.sort(splitKeys, Bytes.BYTES_RAWCOMPARATOR);<a name="line.245"></a> +<span class="sourceLineNo">246</span><a name="line.246"></a> +<span class="sourceLineNo">247</span> int is = 0, il = 0;<a name="line.247"></a> +<span class="sourceLineNo">248</span> for (byte[] key : lastKeys) {<a name="line.248"></a> +<span class="sourceLineNo">249</span> while (is < splitKeys.length && Bytes.compareTo(key, splitKeys[is]) >= 0) is++;<a name="line.249"></a> +<span class="sourceLineNo">250</span> if (is == splitKeys.length) {<a name="line.250"></a> +<span class="sourceLineNo">251</span> break;<a name="line.251"></a> +<span class="sourceLineNo">252</span> }<a name="line.252"></a> +<span class="sourceLineNo">253</span> if (is > 0) {<a name="line.253"></a> +<span class="sourceLineNo">254</span> if (Bytes.compareTo(firstKeys.get(il), splitKeys[is - 1]) < 0) return false;<a name="line.254"></a> +<span class="sourceLineNo">255</span> }<a name="line.255"></a> +<span class="sourceLineNo">256</span> il++;<a name="line.256"></a> +<span class="sourceLineNo">257</span> }<a name="line.257"></a> +<span class="sourceLineNo">258</span><a name="line.258"></a> +<span class="sourceLineNo">259</span> if (is == splitKeys.length) {<a name="line.259"></a> +<span class="sourceLineNo">260</span> return il == lastKeys.size() - 1;<a name="line.260"></a> +<span class="sourceLineNo">261</span> }<a name="line.261"></a> +<span class="sourceLineNo">262</span> return true;<a name="line.262"></a> +<span class="sourceLineNo">263</span> }<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> * MR will output a 3 level directory, tableName->ColumnFamilyName->HFile this method to check the<a name="line.267"></a> +<span class="sourceLineNo">268</span> * created directory is correct or not A recursion method, the testDir had better be small size<a name="line.268"></a> +<span class="sourceLineNo">269</span> */<a name="line.269"></a> +<span class="sourceLineNo">270</span> private boolean checkMROutput(FileSystem fs, Path testDir, int level) throws IOException {<a name="line.270"></a> +<span class="sourceLineNo">271</span> if (level >= 3) {<a name="line.271"></a> +<span class="sourceLineNo">272</span> return HFile.isHFileFormat(fs, testDir);<a name="line.272"></a> +<span class="sourceLineNo">273</span> }<a name="line.273"></a> +<span class="sourceLineNo">274</span> FileStatus[] fStats = fs.listStatus(testDir);<a name="line.274"></a> +<span class="sourceLineNo">275</span> if (fStats == null || fStats.length <= 0) {<a name="line.275"></a> +<span class="sourceLineNo">276</span> LOG.info("Created directory format is not correct");<a name="line.276"></a> +<span class="sourceLineNo">277</span> return false;<a name="line.277"></a> +<span class="sourceLineNo">278</span> }<a name="line.278"></a> +<span class="sourceLineNo">279</span><a name="line.279"></a> +<span class="sourceLineNo">280</span> for (FileStatus stats : fStats) {<a name="line.280"></a> +<span class="sourceLineNo">281</span> // skip the _SUCCESS file created by MapReduce<a name="line.281"></a> +<span class="sourceLineNo">282</span> if (level == 0 && stats.getPath().getName().endsWith(FileOutputCommitter.SUCCEEDED_FILE_NAME))<a name="line.282"></a> +<span class="sourceLineNo">283</span> continue;<a name="line.283"></a> +<span class="sourceLineNo">284</span> if (level < 2 && !stats.isDirectory()) {<a name="line.284"></a> +<span class="sourceLineNo">285</span> LOG.info("Created directory format is not correct");<a name="line.285"></a> +<span class="sourceLineNo">286</span> return false;<a name="line.286"></a> +<span class="sourceLineNo">287</span> }<a name="line.287"></a> +<span class="sourceLineNo">288</span> boolean flag = checkMROutput(fs, stats.getPath(), level + 1);<a name="line.288"></a> +<span class="sourceLineNo">289</span> if (flag == false) return false;<a name="line.289"></a> +<span class="sourceLineNo">290</span> }<a name="line.290"></a> +<span class="sourceLineNo">291</span> return true;<a name="line.291"></a> +<span class="sourceLineNo">292</span> }<a name="line.292"></a> +<span class="sourceLineNo">293</span><a name="line.293"></a> +<span class="sourceLineNo">294</span><a name="line.294"></a> +<span class="sourceLineNo">295</span> private byte[][] generateRandomSplitKeys(int numKeys) {<a name="line.295"></a> +<span class="sourceLineNo">296</span> Random random = new Random();<a name="line.296"></a> +<span class="sourceLineNo">297</span> byte[][] ret = new byte[numKeys][];<a name="line.297"></a> +<span class="sourceLineNo">298</span> for (int i = 0; i < numKeys; i++) {<a name="line.298"></a> +<span class="sourceLineNo">299</span> ret[i] = PerformanceEvaluation.generateData(random, KEYLEN_DEFAULT);<a name="line.299"></a> +<span class="sourceLineNo">300</span> }<a name="line.300"></a> +<span class="sourceLineNo">301</span> return ret;<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> /**<a name="line.305"></a> +<span class="sourceLineNo">306</span> * Simple mapper that makes <TableName, KeyValue> output. With no input data<a name="line.306"></a> +<span class="sourceLineNo">307</span> */<a name="line.307"></a> +<span class="sourceLineNo">308</span> static class Random_TableKV_GeneratingMapper<a name="line.308"></a> +<span class="sourceLineNo">309</span> extends Mapper<NullWritable, NullWritable, ImmutableBytesWritable, Cell> {<a name="line.309"></a> +<span class="sourceLineNo">310</span><a name="line.310"></a> +<span class="sourceLineNo">311</span> private int keyLength;<a name="line.311"></a> +<span class="sourceLineNo">312</span> private int valLength;<a name="line.312"></a> +<span class="sourceLineNo">313</span><a name="line.313"></a> +<span class="sourceLineNo">314</span> @Override<a name="line.314"></a> +<span class="sourceLineNo">315</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.315"></a> +<span class="sourceLineNo">316</span> super.setup(context);<a name="line.316"></a> +<span class="sourceLineNo">317</span><a name="line.317"></a> +<span class="sourceLineNo">318</span> Configuration conf = context.getConfiguration();<a name="line.318"></a> +<span class="sourceLineNo">319</span> keyLength = conf.getInt(KEYLEN_CONF, KEYLEN_DEFAULT);<a name="line.319"></a> +<span class="sourceLineNo">320</span> valLength = conf.getInt(VALLEN_CONF, VALLEN_DEFAULT);<a name="line.320"></a> +<span class="sourceLineNo">321</span> }<a name="line.321"></a> +<span class="sourceLineNo">322</span><a name="line.322"></a> +<span class="sourceLineNo">323</span> @Override<a name="line.323"></a> +<span class="sourceLineNo">324</span> protected void map(NullWritable n1, NullWritable n2,<a name="line.324"></a> +<span class="sourceLineNo">325</span> Mapper<NullWritable, NullWritable, ImmutableBytesWritable, Cell>.Context context)<a name="line.325"></a> +<span class="sourceLineNo">326</span> throws java.io.IOException, InterruptedException {<a name="line.326"></a> +<span class="sourceLineNo">327</span><a name="line.327"></a> +<span class="sourceLineNo">328</span> byte keyBytes[] = new byte[keyLength];<a name="line.328"></a> +<span class="sourceLineNo">329</span> byte valBytes[] = new byte[valLength];<a name="line.329"></a> +<span class="sourceLineNo">330</span><a name="line.330"></a> +<span class="sourceLineNo">331</span> ArrayList<ImmutableBytesWritable> tables = new ArrayList<>();<a name="line.331"></a> +<span class="sourceLineNo">332</span> for (int i = 0; i < TABLES.length; i++) {<a name="line.332"></a> +<span class="sourceLineNo">333</span> tables.add(new ImmutableBytesWritable(TABLES[i]));<a name="line.333"></a> +<span class="sourceLineNo">334</span> }<a name="line.334"></a> +<span class="sourceLineNo">335</span><a name="line.335"></a> +<span class="sourceLineNo">336</span> int taskId = context.getTaskAttemptID().getTaskID().getId();<a name="line.336"></a> +<span class="sourceLineNo">337</span> assert taskId < Byte.MAX_VALUE : "Unit tests dont support > 127 tasks!";<a name="line.337"></a> +<span class="sourceLineNo">338</span> Random random = new Random();<a name="line.338"></a> +<span class="sourceLineNo">339</span><a name="line.339"></a> +<span class="sourceLineNo">340</span> for (int i = 0; i < ROWSPERSPLIT; i++) {<a name="line.340"></a> +<span class="sourceLineNo">341</span> random.nextBytes(keyBytes);<a name="line.341"></a> +<span class="sourceLineNo">342</span> // Ensure that unique tasks generate unique keys<a name="line.342"></a> +<span class="sourceLineNo">343</span> keyBytes[keyLength - 1] = (byte) (taskId & 0xFF);<a name="line.343"></a> +<span class="sourceLineNo">344</span> random.nextBytes(valBytes);<a name="line.344"></a> +<span class="sourceLineNo">345</span><a name="line.345"></a> +<span class="sourceLineNo">346</span> for (ImmutableBytesWritable table : tables) {<a name="line.346"></a> +<span class="sourceLineNo">347</span> for (byte[] family : FAMILIES) {<a name="line.347"></a> +<span class="sourceLineNo">348</span> Cell kv = new KeyValue(keyBytes, family, QUALIFIER, valBytes);<a name="line.348"></a> +<span class="sourceLineNo">349</span> context.write(table, kv);<a name="line.349"></a> +<span class="sourceLineNo">350</span> }<a name="line.350"></a> +<span class="sourceLineNo">351</span> }<a name="line.351"></a> +<span class="sourceLineNo">352</span> }<a name="line.352"></a> +<span class="sourceLineNo">353</span> }<a name="line.353"></a> +<span class="sourceLineNo">354</span> }<a name="line.354"></a> +<span class="sourceLineNo">355</span><a name="line.355"></a> +<span class="sourceLineNo">356</span> /**<a name="line.356"></a> +<span class="sourceLineNo">357</span> * Simple Reducer that have input <TableName, KeyValue>, with KeyValues have no order. and output<a name="line.357"></a> +<span class="sourceLineNo">358</span> * <TableName, KeyValue>, with KeyValues are ordered<a name="line.358"></a> +<span class="sourceLineNo">359</span> */<a name="line.359"></a> +<span class="sourceLineNo">360</span><a name="line.360"></a> +<span class="sourceLineNo">361</span> static class Table_KeyValueSortReducer<a name="line.361"></a> +<span class="sourceLineNo">362</span> extends Reducer<ImmutableBytesWritable, KeyValue, ImmutableBytesWritable, KeyValue> {<a name="line.362"></a> +<span class="sourceLineNo">363</span> protected void reduce(ImmutableBytesWritable table, java.lang.Iterable<KeyValue> kvs,<a name="line.363"></a> +<span class="sourceLineNo">364</span> org.apache.hadoop.mapreduce.Reducer<ImmutableBytesWritable, KeyValue, ImmutableBytesWritable, KeyValue>.Context context)<a name="line.364"></a> +<span class="sourceLineNo">365</span> throws java.io.IOException, InterruptedException {<a name="line.365"></a> +<span class="sourceLineNo">366</span> TreeSet<KeyValue> map = new TreeSet<>(CellComparator.COMPARATOR);<a name="line.366"></a> +<span class="sourceLineNo">367</span> for (KeyValue kv : kvs) {<a name="line.367"></a> +<span class="sourceLineNo">368</span> try {<a name="line.368"></a> +<span class="sourceLineNo">369</span> map.add(kv.clone());<a name="line.369"></a> +<span class="sourceLineNo">370</span> } catch (CloneNotSupportedException e) {<a name="line.370"></a> +<span class="sourceLineNo">371</span> throw new java.io.IOException(e);<a name="line.371"></a> +<span class="sourceLineNo">372</span> }<a name="line.372"></a> +<span class="sourceLineNo">373</span> }<a name="line.373"></a> +<span class="sourceLineNo">374</span> context.setStatus("Read " + map.getClass());<a name="line.374"></a> +<span class="sourceLineNo">375</span> int index = 0;<a name="line.375"></a> +<span class="sourceLineNo">376</span> for (KeyValue kv : map) {<a name="line.376"></a> +<span class="sourceLineNo">377</span> context.write(table, kv);<a name="line.377"></a> +<span class="sourceLineNo">378</span> if (++index % 100 == 0) context.setStatus("Wrote " + index);<a name="line.378"></a> +<span class="sourceLineNo">379</span> }<a name="line.379"></a> +<span class="sourceLineNo">380</span> }<a name="line.380"></a> +<span class="sourceLineNo">381</span> }<a name="line.381"></a> +<span class="sourceLineNo">382</span>}<a name="line.382"></a> + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +</pre> +</div> +</body> +</html>
