blob: 0eb19650381045a1112aac1c766adcd8bafd2fcf [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 static org.junit.Assert.assertEquals;<a name="line.20"></a>
<span class="sourceLineNo">021</span><a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.HashMap;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.Map;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.fs.FileStatus;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.fs.FileSystem;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.fs.Path;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.HBaseClassTestRule;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HBaseTestingUtil;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.TableName;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.Put;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.client.Table;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.testclassification.LargeTests;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.io.MapFile;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.junit.AfterClass;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.junit.Assert;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.junit.BeforeClass;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.junit.ClassRule;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.junit.Rule;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.junit.Test;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.junit.experimental.categories.Category;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.junit.rules.TestName;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.slf4j.Logger;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.slf4j.LoggerFactory;<a name="line.45"></a>
<span class="sourceLineNo">046</span><a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hbase.thirdparty.com.google.common.collect.ImmutableMap;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hbase.thirdparty.com.google.common.collect.Maps;<a name="line.48"></a>
<span class="sourceLineNo">049</span><a name="line.49"></a>
<span class="sourceLineNo">050</span>/**<a name="line.50"></a>
<span class="sourceLineNo">051</span> * Basic test for the HashTable M/R tool<a name="line.51"></a>
<span class="sourceLineNo">052</span> */<a name="line.52"></a>
<span class="sourceLineNo">053</span>@Category(LargeTests.class)<a name="line.53"></a>
<span class="sourceLineNo">054</span>public class TestHashTable {<a name="line.54"></a>
<span class="sourceLineNo">055</span> @ClassRule<a name="line.55"></a>
<span class="sourceLineNo">056</span> public static final HBaseClassTestRule CLASS_RULE =<a name="line.56"></a>
<span class="sourceLineNo">057</span> HBaseClassTestRule.forClass(TestHashTable.class);<a name="line.57"></a>
<span class="sourceLineNo">058</span><a name="line.58"></a>
<span class="sourceLineNo">059</span> private static final Logger LOG = LoggerFactory.getLogger(TestHashTable.class);<a name="line.59"></a>
<span class="sourceLineNo">060</span><a name="line.60"></a>
<span class="sourceLineNo">061</span> private static final HBaseTestingUtil TEST_UTIL = new HBaseTestingUtil();<a name="line.61"></a>
<span class="sourceLineNo">062</span><a name="line.62"></a>
<span class="sourceLineNo">063</span> @Rule<a name="line.63"></a>
<span class="sourceLineNo">064</span> public TestName name = new TestName();<a name="line.64"></a>
<span class="sourceLineNo">065</span><a name="line.65"></a>
<span class="sourceLineNo">066</span> @BeforeClass<a name="line.66"></a>
<span class="sourceLineNo">067</span> public static void beforeClass() throws Exception {<a name="line.67"></a>
<span class="sourceLineNo">068</span> TEST_UTIL.startMiniCluster(3);<a name="line.68"></a>
<span class="sourceLineNo">069</span> }<a name="line.69"></a>
<span class="sourceLineNo">070</span><a name="line.70"></a>
<span class="sourceLineNo">071</span> @AfterClass<a name="line.71"></a>
<span class="sourceLineNo">072</span> public static void afterClass() throws Exception {<a name="line.72"></a>
<span class="sourceLineNo">073</span> TEST_UTIL.shutdownMiniCluster();<a name="line.73"></a>
<span class="sourceLineNo">074</span> }<a name="line.74"></a>
<span class="sourceLineNo">075</span><a name="line.75"></a>
<span class="sourceLineNo">076</span> @Test<a name="line.76"></a>
<span class="sourceLineNo">077</span> public void testHashTable() throws Exception {<a name="line.77"></a>
<span class="sourceLineNo">078</span> final TableName tableName = TableName.valueOf(name.getMethodName());<a name="line.78"></a>
<span class="sourceLineNo">079</span> final byte[] family = Bytes.toBytes("family");<a name="line.79"></a>
<span class="sourceLineNo">080</span> final byte[] column1 = Bytes.toBytes("c1");<a name="line.80"></a>
<span class="sourceLineNo">081</span> final byte[] column2 = Bytes.toBytes("c2");<a name="line.81"></a>
<span class="sourceLineNo">082</span> final byte[] column3 = Bytes.toBytes("c3");<a name="line.82"></a>
<span class="sourceLineNo">083</span><a name="line.83"></a>
<span class="sourceLineNo">084</span> int numRows = 100;<a name="line.84"></a>
<span class="sourceLineNo">085</span> int numRegions = 10;<a name="line.85"></a>
<span class="sourceLineNo">086</span> int numHashFiles = 3;<a name="line.86"></a>
<span class="sourceLineNo">087</span><a name="line.87"></a>
<span class="sourceLineNo">088</span> byte[][] splitRows = new byte[numRegions - 1][];<a name="line.88"></a>
<span class="sourceLineNo">089</span> for (int i = 1; i &lt; numRegions; i++) {<a name="line.89"></a>
<span class="sourceLineNo">090</span> splitRows[i - 1] = Bytes.toBytes(numRows * i / numRegions);<a name="line.90"></a>
<span class="sourceLineNo">091</span> }<a name="line.91"></a>
<span class="sourceLineNo">092</span><a name="line.92"></a>
<span class="sourceLineNo">093</span> long timestamp = 1430764183454L;<a name="line.93"></a>
<span class="sourceLineNo">094</span> // put rows into the first table<a name="line.94"></a>
<span class="sourceLineNo">095</span> Table t1 = TEST_UTIL.createTable(tableName, family, splitRows);<a name="line.95"></a>
<span class="sourceLineNo">096</span> for (int i = 0; i &lt; numRows; i++) {<a name="line.96"></a>
<span class="sourceLineNo">097</span> Put p = new Put(Bytes.toBytes(i), timestamp);<a name="line.97"></a>
<span class="sourceLineNo">098</span> p.addColumn(family, column1, column1);<a name="line.98"></a>
<span class="sourceLineNo">099</span> p.addColumn(family, column2, column2);<a name="line.99"></a>
<span class="sourceLineNo">100</span> p.addColumn(family, column3, column3);<a name="line.100"></a>
<span class="sourceLineNo">101</span> t1.put(p);<a name="line.101"></a>
<span class="sourceLineNo">102</span> }<a name="line.102"></a>
<span class="sourceLineNo">103</span> t1.close();<a name="line.103"></a>
<span class="sourceLineNo">104</span><a name="line.104"></a>
<span class="sourceLineNo">105</span> HashTable hashTable = new HashTable(TEST_UTIL.getConfiguration());<a name="line.105"></a>
<span class="sourceLineNo">106</span><a name="line.106"></a>
<span class="sourceLineNo">107</span> Path testDir = TEST_UTIL.getDataTestDirOnTestFS(tableName.getNameAsString());<a name="line.107"></a>
<span class="sourceLineNo">108</span><a name="line.108"></a>
<span class="sourceLineNo">109</span> long batchSize = 300;<a name="line.109"></a>
<span class="sourceLineNo">110</span> int code =<a name="line.110"></a>
<span class="sourceLineNo">111</span> hashTable.run(new String[] { "--batchsize=" + batchSize, "--numhashfiles=" + numHashFiles,<a name="line.111"></a>
<span class="sourceLineNo">112</span> "--scanbatch=2", tableName.getNameAsString(), testDir.toString() });<a name="line.112"></a>
<span class="sourceLineNo">113</span> assertEquals("test job failed", 0, code);<a name="line.113"></a>
<span class="sourceLineNo">114</span><a name="line.114"></a>
<span class="sourceLineNo">115</span> FileSystem fs = TEST_UTIL.getTestFileSystem();<a name="line.115"></a>
<span class="sourceLineNo">116</span><a name="line.116"></a>
<span class="sourceLineNo">117</span> HashTable.TableHash tableHash = HashTable.TableHash.read(fs.getConf(), testDir);<a name="line.117"></a>
<span class="sourceLineNo">118</span> assertEquals(tableName.getNameAsString(), tableHash.tableName);<a name="line.118"></a>
<span class="sourceLineNo">119</span> assertEquals(batchSize, tableHash.batchSize);<a name="line.119"></a>
<span class="sourceLineNo">120</span> assertEquals(numHashFiles, tableHash.numHashFiles);<a name="line.120"></a>
<span class="sourceLineNo">121</span> assertEquals(numHashFiles - 1, tableHash.partitions.size());<a name="line.121"></a>
<span class="sourceLineNo">122</span> for (ImmutableBytesWritable bytes : tableHash.partitions) {<a name="line.122"></a>
<span class="sourceLineNo">123</span> LOG.debug("partition: " + Bytes.toInt(bytes.get()));<a name="line.123"></a>
<span class="sourceLineNo">124</span> }<a name="line.124"></a>
<span class="sourceLineNo">125</span><a name="line.125"></a>
<span class="sourceLineNo">126</span> ImmutableMap&lt;Integer, ImmutableBytesWritable&gt; expectedHashes =<a name="line.126"></a>
<span class="sourceLineNo">127</span> ImmutableMap.&lt;Integer, ImmutableBytesWritable&gt; builder()<a name="line.127"></a>
<span class="sourceLineNo">128</span> .put(-1, new ImmutableBytesWritable(Bytes.fromHex("714cb10a9e3b5569852980edd8c6ca2f")))<a name="line.128"></a>
<span class="sourceLineNo">129</span> .put(5, new ImmutableBytesWritable(Bytes.fromHex("28d961d9252ce8f8d44a07b38d3e1d96")))<a name="line.129"></a>
<span class="sourceLineNo">130</span> .put(10, new ImmutableBytesWritable(Bytes.fromHex("f6bbc4a224d8fd929b783a92599eaffa")))<a name="line.130"></a>
<span class="sourceLineNo">131</span> .put(15, new ImmutableBytesWritable(Bytes.fromHex("522deb5d97f73a414ecc11457be46881")))<a name="line.131"></a>
<span class="sourceLineNo">132</span> .put(20, new ImmutableBytesWritable(Bytes.fromHex("b026f2611aaa46f7110116d807545352")))<a name="line.132"></a>
<span class="sourceLineNo">133</span> .put(25, new ImmutableBytesWritable(Bytes.fromHex("39ffc1a3094aa12a2e90ffd9cef2ce93")))<a name="line.133"></a>
<span class="sourceLineNo">134</span> .put(30, new ImmutableBytesWritable(Bytes.fromHex("f6b4d75727ce9a30ac29e4f08f601666")))<a name="line.134"></a>
<span class="sourceLineNo">135</span> .put(35, new ImmutableBytesWritable(Bytes.fromHex("422e2d2f1eb79a8f02171a705a42c090")))<a name="line.135"></a>
<span class="sourceLineNo">136</span> .put(40, new ImmutableBytesWritable(Bytes.fromHex("559ad61c900fffefea0a15abf8a97bc3")))<a name="line.136"></a>
<span class="sourceLineNo">137</span> .put(45, new ImmutableBytesWritable(Bytes.fromHex("23019084513eca41cee436b2a29611cb")))<a name="line.137"></a>
<span class="sourceLineNo">138</span> .put(50, new ImmutableBytesWritable(Bytes.fromHex("b40467d222ddb4949b142fe145ee9edc")))<a name="line.138"></a>
<span class="sourceLineNo">139</span> .put(55, new ImmutableBytesWritable(Bytes.fromHex("372bf89fcd8ca4b7ab3c1add9d07f7e4")))<a name="line.139"></a>
<span class="sourceLineNo">140</span> .put(60, new ImmutableBytesWritable(Bytes.fromHex("69ae0585e6255de27dce974e332b8f8b")))<a name="line.140"></a>
<span class="sourceLineNo">141</span> .put(65, new ImmutableBytesWritable(Bytes.fromHex("8029610044297aad0abdbecd485d8e59")))<a name="line.141"></a>
<span class="sourceLineNo">142</span> .put(70, new ImmutableBytesWritable(Bytes.fromHex("de5f784f7f78987b6e57ecfd81c8646f")))<a name="line.142"></a>
<span class="sourceLineNo">143</span> .put(75, new ImmutableBytesWritable(Bytes.fromHex("1cd757cc4e1715c8c3b1c24447a1ec56")))<a name="line.143"></a>
<span class="sourceLineNo">144</span> .put(80, new ImmutableBytesWritable(Bytes.fromHex("f9a53aacfeb6142b08066615e7038095")))<a name="line.144"></a>
<span class="sourceLineNo">145</span> .put(85, new ImmutableBytesWritable(Bytes.fromHex("89b872b7e639df32d3276b33928c0c91")))<a name="line.145"></a>
<span class="sourceLineNo">146</span> .put(90, new ImmutableBytesWritable(Bytes.fromHex("45eeac0646d46a474ea0484175faed38")))<a name="line.146"></a>
<span class="sourceLineNo">147</span> .put(95, new ImmutableBytesWritable(Bytes.fromHex("f57c447e32a08f4bf1abb2892839ac56")))<a name="line.147"></a>
<span class="sourceLineNo">148</span> .build();<a name="line.148"></a>
<span class="sourceLineNo">149</span><a name="line.149"></a>
<span class="sourceLineNo">150</span> Map&lt;Integer, ImmutableBytesWritable&gt; actualHashes = new HashMap&lt;&gt;();<a name="line.150"></a>
<span class="sourceLineNo">151</span> Path dataDir = new Path(testDir, HashTable.HASH_DATA_DIR);<a name="line.151"></a>
<span class="sourceLineNo">152</span> for (int i = 0; i &lt; numHashFiles; i++) {<a name="line.152"></a>
<span class="sourceLineNo">153</span> Path hashPath = new Path(dataDir, HashTable.TableHash.getDataFileName(i));<a name="line.153"></a>
<span class="sourceLineNo">154</span><a name="line.154"></a>
<span class="sourceLineNo">155</span> MapFile.Reader reader = new MapFile.Reader(hashPath, fs.getConf());<a name="line.155"></a>
<span class="sourceLineNo">156</span> ImmutableBytesWritable key = new ImmutableBytesWritable();<a name="line.156"></a>
<span class="sourceLineNo">157</span> ImmutableBytesWritable hash = new ImmutableBytesWritable();<a name="line.157"></a>
<span class="sourceLineNo">158</span> while (reader.next(key, hash)) {<a name="line.158"></a>
<span class="sourceLineNo">159</span> String keyString = Bytes.toHex(key.get(), key.getOffset(), key.getLength());<a name="line.159"></a>
<span class="sourceLineNo">160</span> LOG.debug("Key: " + (keyString.isEmpty() ? "-1" : Integer.parseInt(keyString, 16))<a name="line.160"></a>
<span class="sourceLineNo">161</span> + " Hash: " + Bytes.toHex(hash.get(), hash.getOffset(), hash.getLength()));<a name="line.161"></a>
<span class="sourceLineNo">162</span><a name="line.162"></a>
<span class="sourceLineNo">163</span> int intKey = -1;<a name="line.163"></a>
<span class="sourceLineNo">164</span> if (key.getLength() &gt; 0) {<a name="line.164"></a>
<span class="sourceLineNo">165</span> intKey = Bytes.toInt(key.get(), key.getOffset(), key.getLength());<a name="line.165"></a>
<span class="sourceLineNo">166</span> }<a name="line.166"></a>
<span class="sourceLineNo">167</span> if (actualHashes.containsKey(intKey)) {<a name="line.167"></a>
<span class="sourceLineNo">168</span> Assert.fail("duplicate key in data files: " + intKey);<a name="line.168"></a>
<span class="sourceLineNo">169</span> }<a name="line.169"></a>
<span class="sourceLineNo">170</span> actualHashes.put(intKey, new ImmutableBytesWritable(hash.copyBytes()));<a name="line.170"></a>
<span class="sourceLineNo">171</span> }<a name="line.171"></a>
<span class="sourceLineNo">172</span> reader.close();<a name="line.172"></a>
<span class="sourceLineNo">173</span> }<a name="line.173"></a>
<span class="sourceLineNo">174</span><a name="line.174"></a>
<span class="sourceLineNo">175</span> FileStatus[] files = fs.listStatus(testDir);<a name="line.175"></a>
<span class="sourceLineNo">176</span> for (FileStatus file : files) {<a name="line.176"></a>
<span class="sourceLineNo">177</span> LOG.debug("Output file: " + file.getPath());<a name="line.177"></a>
<span class="sourceLineNo">178</span> }<a name="line.178"></a>
<span class="sourceLineNo">179</span><a name="line.179"></a>
<span class="sourceLineNo">180</span> files = fs.listStatus(dataDir);<a name="line.180"></a>
<span class="sourceLineNo">181</span> for (FileStatus file : files) {<a name="line.181"></a>
<span class="sourceLineNo">182</span> LOG.debug("Data file: " + file.getPath());<a name="line.182"></a>
<span class="sourceLineNo">183</span> }<a name="line.183"></a>
<span class="sourceLineNo">184</span><a name="line.184"></a>
<span class="sourceLineNo">185</span> if (!expectedHashes.equals(actualHashes)) {<a name="line.185"></a>
<span class="sourceLineNo">186</span> LOG.error("Diff: " + Maps.difference(expectedHashes, actualHashes));<a name="line.186"></a>
<span class="sourceLineNo">187</span> }<a name="line.187"></a>
<span class="sourceLineNo">188</span> Assert.assertEquals(expectedHashes, actualHashes);<a name="line.188"></a>
<span class="sourceLineNo">189</span><a name="line.189"></a>
<span class="sourceLineNo">190</span> TEST_UTIL.deleteTable(tableName);<a name="line.190"></a>
<span class="sourceLineNo">191</span> TEST_UTIL.cleanupDataTestDirOnTestFS();<a name="line.191"></a>
<span class="sourceLineNo">192</span> }<a name="line.192"></a>
<span class="sourceLineNo">193</span>}<a name="line.193"></a>
</pre>
</div>
</body>
</html>