blob: c259629e5b9fdecfcdaa22ce933f82a96fefcd72 [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.assertTrue;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import static org.junit.Assert.fail;<a name="line.21"></a>
<span class="sourceLineNo">022</span><a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.IOException;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.util.Iterator;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.Map;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.NavigableMap;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.conf.Configuration;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.Cell;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.HBaseTestingUtil;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.HConstants;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.TableName;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.Put;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Result;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.ResultScanner;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.Table;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.junit.AfterClass;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.junit.BeforeClass;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.junit.Test;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.slf4j.Logger;<a name="line.43"></a>
<span class="sourceLineNo">044</span><a name="line.44"></a>
<span class="sourceLineNo">045</span>/**<a name="line.45"></a>
<span class="sourceLineNo">046</span> * A base class for a test Map/Reduce job over HBase tables. The map/reduce process we're testing on<a name="line.46"></a>
<span class="sourceLineNo">047</span> * our tables is simple - take every row in the table, reverse the value of a particular cell, and<a name="line.47"></a>
<span class="sourceLineNo">048</span> * write it back to the table. Implements common components between mapred and mapreduce<a name="line.48"></a>
<span class="sourceLineNo">049</span> * implementations.<a name="line.49"></a>
<span class="sourceLineNo">050</span> */<a name="line.50"></a>
<span class="sourceLineNo">051</span>public abstract class TestTableMapReduceBase {<a name="line.51"></a>
<span class="sourceLineNo">052</span> protected static final HBaseTestingUtil UTIL = new HBaseTestingUtil();<a name="line.52"></a>
<span class="sourceLineNo">053</span> protected static final TableName MULTI_REGION_TABLE_NAME = TableName.valueOf("mrtest");<a name="line.53"></a>
<span class="sourceLineNo">054</span> protected static final TableName TABLE_FOR_NEGATIVE_TESTS = TableName.valueOf("testfailuretable");<a name="line.54"></a>
<span class="sourceLineNo">055</span> protected static final byte[] INPUT_FAMILY = Bytes.toBytes("contents");<a name="line.55"></a>
<span class="sourceLineNo">056</span> protected static final byte[] OUTPUT_FAMILY = Bytes.toBytes("text");<a name="line.56"></a>
<span class="sourceLineNo">057</span><a name="line.57"></a>
<span class="sourceLineNo">058</span> protected static final byte[][] columns = new byte[][] { INPUT_FAMILY, OUTPUT_FAMILY };<a name="line.58"></a>
<span class="sourceLineNo">059</span><a name="line.59"></a>
<span class="sourceLineNo">060</span> /**<a name="line.60"></a>
<span class="sourceLineNo">061</span> * Retrieve my logger instance.<a name="line.61"></a>
<span class="sourceLineNo">062</span> */<a name="line.62"></a>
<span class="sourceLineNo">063</span> protected abstract Logger getLog();<a name="line.63"></a>
<span class="sourceLineNo">064</span><a name="line.64"></a>
<span class="sourceLineNo">065</span> /**<a name="line.65"></a>
<span class="sourceLineNo">066</span> * Handles API-specifics for setting up and executing the job.<a name="line.66"></a>
<span class="sourceLineNo">067</span> */<a name="line.67"></a>
<span class="sourceLineNo">068</span> protected abstract void runTestOnTable(Table table) throws IOException;<a name="line.68"></a>
<span class="sourceLineNo">069</span><a name="line.69"></a>
<span class="sourceLineNo">070</span> @BeforeClass<a name="line.70"></a>
<span class="sourceLineNo">071</span> public static void beforeClass() throws Exception {<a name="line.71"></a>
<span class="sourceLineNo">072</span> UTIL.startMiniCluster();<a name="line.72"></a>
<span class="sourceLineNo">073</span> Table table = UTIL.createMultiRegionTable(MULTI_REGION_TABLE_NAME,<a name="line.73"></a>
<span class="sourceLineNo">074</span> new byte[][] { INPUT_FAMILY, OUTPUT_FAMILY });<a name="line.74"></a>
<span class="sourceLineNo">075</span> UTIL.loadTable(table, INPUT_FAMILY, false);<a name="line.75"></a>
<span class="sourceLineNo">076</span> UTIL.createTable(TABLE_FOR_NEGATIVE_TESTS, new byte[][] { INPUT_FAMILY, OUTPUT_FAMILY });<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> @AfterClass<a name="line.79"></a>
<span class="sourceLineNo">080</span> public static void afterClass() throws Exception {<a name="line.80"></a>
<span class="sourceLineNo">081</span> UTIL.deleteTable(TABLE_FOR_NEGATIVE_TESTS);<a name="line.81"></a>
<span class="sourceLineNo">082</span> UTIL.shutdownMiniCluster();<a name="line.82"></a>
<span class="sourceLineNo">083</span> }<a name="line.83"></a>
<span class="sourceLineNo">084</span><a name="line.84"></a>
<span class="sourceLineNo">085</span> /**<a name="line.85"></a>
<span class="sourceLineNo">086</span> * Test a map/reduce against a multi-region table n<a name="line.86"></a>
<span class="sourceLineNo">087</span> */<a name="line.87"></a>
<span class="sourceLineNo">088</span> @Test<a name="line.88"></a>
<span class="sourceLineNo">089</span> public void testMultiRegionTable() throws IOException {<a name="line.89"></a>
<span class="sourceLineNo">090</span> runTestOnTable(UTIL.getConnection().getTable(MULTI_REGION_TABLE_NAME));<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> @Test<a name="line.93"></a>
<span class="sourceLineNo">094</span> public void testCombiner() throws IOException {<a name="line.94"></a>
<span class="sourceLineNo">095</span> Configuration conf = new Configuration(UTIL.getConfiguration());<a name="line.95"></a>
<span class="sourceLineNo">096</span> // force use of combiner for testing purposes<a name="line.96"></a>
<span class="sourceLineNo">097</span> conf.setInt("mapreduce.map.combine.minspills", 1);<a name="line.97"></a>
<span class="sourceLineNo">098</span> runTestOnTable(UTIL.getConnection().getTable(MULTI_REGION_TABLE_NAME));<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> /**<a name="line.101"></a>
<span class="sourceLineNo">102</span> * Implements mapper logic for use across APIs.<a name="line.102"></a>
<span class="sourceLineNo">103</span> */<a name="line.103"></a>
<span class="sourceLineNo">104</span> protected static Put map(ImmutableBytesWritable key, Result value) throws IOException {<a name="line.104"></a>
<span class="sourceLineNo">105</span> if (value.size() != 1) {<a name="line.105"></a>
<span class="sourceLineNo">106</span> throw new IOException("There should only be one input column");<a name="line.106"></a>
<span class="sourceLineNo">107</span> }<a name="line.107"></a>
<span class="sourceLineNo">108</span> Map&lt;byte[], NavigableMap&lt;byte[], NavigableMap&lt;Long, byte[]&gt;&gt;&gt; cf = value.getMap();<a name="line.108"></a>
<span class="sourceLineNo">109</span> if (!cf.containsKey(INPUT_FAMILY)) {<a name="line.109"></a>
<span class="sourceLineNo">110</span> throw new IOException(<a name="line.110"></a>
<span class="sourceLineNo">111</span> "Wrong input columns. Missing: '" + Bytes.toString(INPUT_FAMILY) + "'.");<a name="line.111"></a>
<span class="sourceLineNo">112</span> }<a name="line.112"></a>
<span class="sourceLineNo">113</span><a name="line.113"></a>
<span class="sourceLineNo">114</span> // Get the original value and reverse it<a name="line.114"></a>
<span class="sourceLineNo">115</span><a name="line.115"></a>
<span class="sourceLineNo">116</span> String originalValue = Bytes.toString(value.getValue(INPUT_FAMILY, INPUT_FAMILY));<a name="line.116"></a>
<span class="sourceLineNo">117</span> StringBuilder newValue = new StringBuilder(originalValue);<a name="line.117"></a>
<span class="sourceLineNo">118</span> newValue.reverse();<a name="line.118"></a>
<span class="sourceLineNo">119</span><a name="line.119"></a>
<span class="sourceLineNo">120</span> // Now set the value to be collected<a name="line.120"></a>
<span class="sourceLineNo">121</span><a name="line.121"></a>
<span class="sourceLineNo">122</span> Put outval = new Put(key.get());<a name="line.122"></a>
<span class="sourceLineNo">123</span> outval.addColumn(OUTPUT_FAMILY, null, Bytes.toBytes(newValue.toString()));<a name="line.123"></a>
<span class="sourceLineNo">124</span> return outval;<a name="line.124"></a>
<span class="sourceLineNo">125</span> }<a name="line.125"></a>
<span class="sourceLineNo">126</span><a name="line.126"></a>
<span class="sourceLineNo">127</span> protected void verify(TableName tableName) throws IOException {<a name="line.127"></a>
<span class="sourceLineNo">128</span> Table table = UTIL.getConnection().getTable(tableName);<a name="line.128"></a>
<span class="sourceLineNo">129</span> boolean verified = false;<a name="line.129"></a>
<span class="sourceLineNo">130</span> long pause = UTIL.getConfiguration().getLong("hbase.client.pause", 5 * 1000);<a name="line.130"></a>
<span class="sourceLineNo">131</span> int numRetries = UTIL.getConfiguration().getInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER, 5);<a name="line.131"></a>
<span class="sourceLineNo">132</span> for (int i = 0; i &lt; numRetries; i++) {<a name="line.132"></a>
<span class="sourceLineNo">133</span> try {<a name="line.133"></a>
<span class="sourceLineNo">134</span> getLog().info("Verification attempt #" + i);<a name="line.134"></a>
<span class="sourceLineNo">135</span> verifyAttempt(table);<a name="line.135"></a>
<span class="sourceLineNo">136</span> verified = true;<a name="line.136"></a>
<span class="sourceLineNo">137</span> break;<a name="line.137"></a>
<span class="sourceLineNo">138</span> } catch (NullPointerException e) {<a name="line.138"></a>
<span class="sourceLineNo">139</span> // If here, a cell was empty. Presume its because updates came in<a name="line.139"></a>
<span class="sourceLineNo">140</span> // after the scanner had been opened. Wait a while and retry.<a name="line.140"></a>
<span class="sourceLineNo">141</span> getLog().debug("Verification attempt failed: " + e.getMessage());<a name="line.141"></a>
<span class="sourceLineNo">142</span> }<a name="line.142"></a>
<span class="sourceLineNo">143</span> try {<a name="line.143"></a>
<span class="sourceLineNo">144</span> Thread.sleep(pause);<a name="line.144"></a>
<span class="sourceLineNo">145</span> } catch (InterruptedException e) {<a name="line.145"></a>
<span class="sourceLineNo">146</span> // continue<a name="line.146"></a>
<span class="sourceLineNo">147</span> }<a name="line.147"></a>
<span class="sourceLineNo">148</span> }<a name="line.148"></a>
<span class="sourceLineNo">149</span> assertTrue(verified);<a name="line.149"></a>
<span class="sourceLineNo">150</span> }<a name="line.150"></a>
<span class="sourceLineNo">151</span><a name="line.151"></a>
<span class="sourceLineNo">152</span> /**<a name="line.152"></a>
<span class="sourceLineNo">153</span> * Looks at every value of the mapreduce output and verifies that indeed the values have been<a name="line.153"></a>
<span class="sourceLineNo">154</span> * reversed.<a name="line.154"></a>
<span class="sourceLineNo">155</span> * @param table Table to scan. n * @throws NullPointerException if we failed to find a cell value<a name="line.155"></a>
<span class="sourceLineNo">156</span> */<a name="line.156"></a>
<span class="sourceLineNo">157</span> private void verifyAttempt(final Table table) throws IOException, NullPointerException {<a name="line.157"></a>
<span class="sourceLineNo">158</span> Scan scan = new Scan();<a name="line.158"></a>
<span class="sourceLineNo">159</span> TableInputFormat.addColumns(scan, columns);<a name="line.159"></a>
<span class="sourceLineNo">160</span> ResultScanner scanner = table.getScanner(scan);<a name="line.160"></a>
<span class="sourceLineNo">161</span> try {<a name="line.161"></a>
<span class="sourceLineNo">162</span> Iterator&lt;Result&gt; itr = scanner.iterator();<a name="line.162"></a>
<span class="sourceLineNo">163</span> assertTrue(itr.hasNext());<a name="line.163"></a>
<span class="sourceLineNo">164</span> while (itr.hasNext()) {<a name="line.164"></a>
<span class="sourceLineNo">165</span> Result r = itr.next();<a name="line.165"></a>
<span class="sourceLineNo">166</span> if (getLog().isDebugEnabled()) {<a name="line.166"></a>
<span class="sourceLineNo">167</span> if (r.size() &gt; 2) {<a name="line.167"></a>
<span class="sourceLineNo">168</span> throw new IOException("Too many results, expected 2 got " + r.size());<a name="line.168"></a>
<span class="sourceLineNo">169</span> }<a name="line.169"></a>
<span class="sourceLineNo">170</span> }<a name="line.170"></a>
<span class="sourceLineNo">171</span> byte[] firstValue = null;<a name="line.171"></a>
<span class="sourceLineNo">172</span> byte[] secondValue = null;<a name="line.172"></a>
<span class="sourceLineNo">173</span> int count = 0;<a name="line.173"></a>
<span class="sourceLineNo">174</span> for (Cell kv : r.listCells()) {<a name="line.174"></a>
<span class="sourceLineNo">175</span> if (count == 0) {<a name="line.175"></a>
<span class="sourceLineNo">176</span> firstValue = CellUtil.cloneValue(kv);<a name="line.176"></a>
<span class="sourceLineNo">177</span> }<a name="line.177"></a>
<span class="sourceLineNo">178</span> if (count == 1) {<a name="line.178"></a>
<span class="sourceLineNo">179</span> secondValue = CellUtil.cloneValue(kv);<a name="line.179"></a>
<span class="sourceLineNo">180</span> }<a name="line.180"></a>
<span class="sourceLineNo">181</span> count++;<a name="line.181"></a>
<span class="sourceLineNo">182</span> if (count == 2) {<a name="line.182"></a>
<span class="sourceLineNo">183</span> break;<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> if (firstValue == null) {<a name="line.187"></a>
<span class="sourceLineNo">188</span> throw new NullPointerException(Bytes.toString(r.getRow()) + ": first value is null");<a name="line.188"></a>
<span class="sourceLineNo">189</span> }<a name="line.189"></a>
<span class="sourceLineNo">190</span> String first = Bytes.toString(firstValue);<a name="line.190"></a>
<span class="sourceLineNo">191</span><a name="line.191"></a>
<span class="sourceLineNo">192</span> if (secondValue == null) {<a name="line.192"></a>
<span class="sourceLineNo">193</span> throw new NullPointerException(Bytes.toString(r.getRow()) + ": second value is null");<a name="line.193"></a>
<span class="sourceLineNo">194</span> }<a name="line.194"></a>
<span class="sourceLineNo">195</span> byte[] secondReversed = new byte[secondValue.length];<a name="line.195"></a>
<span class="sourceLineNo">196</span> for (int i = 0, j = secondValue.length - 1; j &gt;= 0; j--, i++) {<a name="line.196"></a>
<span class="sourceLineNo">197</span> secondReversed[i] = secondValue[j];<a name="line.197"></a>
<span class="sourceLineNo">198</span> }<a name="line.198"></a>
<span class="sourceLineNo">199</span> String second = Bytes.toString(secondReversed);<a name="line.199"></a>
<span class="sourceLineNo">200</span><a name="line.200"></a>
<span class="sourceLineNo">201</span> if (first.compareTo(second) != 0) {<a name="line.201"></a>
<span class="sourceLineNo">202</span> if (getLog().isDebugEnabled()) {<a name="line.202"></a>
<span class="sourceLineNo">203</span> getLog().debug(<a name="line.203"></a>
<span class="sourceLineNo">204</span> "second key is not the reverse of first. row=" + Bytes.toStringBinary(r.getRow())<a name="line.204"></a>
<span class="sourceLineNo">205</span> + ", first value=" + first + ", second value=" + second);<a name="line.205"></a>
<span class="sourceLineNo">206</span> }<a name="line.206"></a>
<span class="sourceLineNo">207</span> fail();<a name="line.207"></a>
<span class="sourceLineNo">208</span> }<a name="line.208"></a>
<span class="sourceLineNo">209</span> }<a name="line.209"></a>
<span class="sourceLineNo">210</span> } finally {<a name="line.210"></a>
<span class="sourceLineNo">211</span> scanner.close();<a name="line.211"></a>
<span class="sourceLineNo">212</span> }<a name="line.212"></a>
<span class="sourceLineNo">213</span> }<a name="line.213"></a>
<span class="sourceLineNo">214</span>}<a name="line.214"></a>
</pre>
</div>
</body>
</html>