blob: 636779ba02862324309251bdb74ad4cbbdbcd55c [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> *<a name="line.2"></a>
<span class="sourceLineNo">003</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.3"></a>
<span class="sourceLineNo">004</span> * or more contributor license agreements. See the NOTICE file<a name="line.4"></a>
<span class="sourceLineNo">005</span> * distributed with this work for additional information<a name="line.5"></a>
<span class="sourceLineNo">006</span> * regarding copyright ownership. The ASF licenses this file<a name="line.6"></a>
<span class="sourceLineNo">007</span> * to you under the Apache License, Version 2.0 (the<a name="line.7"></a>
<span class="sourceLineNo">008</span> * "License"); you may not use this file except in compliance<a name="line.8"></a>
<span class="sourceLineNo">009</span> * with the License. You may obtain a copy of the License at<a name="line.9"></a>
<span class="sourceLineNo">010</span> *<a name="line.10"></a>
<span class="sourceLineNo">011</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.11"></a>
<span class="sourceLineNo">012</span> *<a name="line.12"></a>
<span class="sourceLineNo">013</span> * Unless required by applicable law or agreed to in writing, software<a name="line.13"></a>
<span class="sourceLineNo">014</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.14"></a>
<span class="sourceLineNo">015</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.15"></a>
<span class="sourceLineNo">016</span> * See the License for the specific language governing permissions and<a name="line.16"></a>
<span class="sourceLineNo">017</span> * limitations under the License.<a name="line.17"></a>
<span class="sourceLineNo">018</span> */<a name="line.18"></a>
<span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.19"></a>
<span class="sourceLineNo">020</span><a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import org.apache.hadoop.conf.Configurable;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.conf.Configuration;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.HConstants;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.TableName;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.TableNotEnabledException;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.Delete;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Mutation;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.Put;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.mapreduce.OutputCommitter;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.slf4j.Logger;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.slf4j.LoggerFactory;<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> * Convert Map/Reduce output and write it to an HBase table. The KEY is ignored<a name="line.46"></a>
<span class="sourceLineNo">047</span> * while the output value &lt;u&gt;must&lt;/u&gt; be either a {@link Put} or a<a name="line.47"></a>
<span class="sourceLineNo">048</span> * {@link Delete} instance.<a name="line.48"></a>
<span class="sourceLineNo">049</span> */<a name="line.49"></a>
<span class="sourceLineNo">050</span>@InterfaceAudience.Public<a name="line.50"></a>
<span class="sourceLineNo">051</span>public class TableOutputFormat&lt;KEY&gt; extends OutputFormat&lt;KEY, Mutation&gt;<a name="line.51"></a>
<span class="sourceLineNo">052</span>implements Configurable {<a name="line.52"></a>
<span class="sourceLineNo">053</span><a name="line.53"></a>
<span class="sourceLineNo">054</span> private static final Logger LOG = LoggerFactory.getLogger(TableOutputFormat.class);<a name="line.54"></a>
<span class="sourceLineNo">055</span><a name="line.55"></a>
<span class="sourceLineNo">056</span> /** Job parameter that specifies the output table. */<a name="line.56"></a>
<span class="sourceLineNo">057</span> public static final String OUTPUT_TABLE = "hbase.mapred.outputtable";<a name="line.57"></a>
<span class="sourceLineNo">058</span><a name="line.58"></a>
<span class="sourceLineNo">059</span> /**<a name="line.59"></a>
<span class="sourceLineNo">060</span> * Prefix for configuration property overrides to apply in {@link #setConf(Configuration)}.<a name="line.60"></a>
<span class="sourceLineNo">061</span> * For keys matching this prefix, the prefix is stripped, and the value is set in the<a name="line.61"></a>
<span class="sourceLineNo">062</span> * configuration with the resulting key, ie. the entry "hbase.mapred.output.key1 = value1"<a name="line.62"></a>
<span class="sourceLineNo">063</span> * would be set in the configuration as "key1 = value1". Use this to set properties<a name="line.63"></a>
<span class="sourceLineNo">064</span> * which should only be applied to the {@code TableOutputFormat} configuration and not the<a name="line.64"></a>
<span class="sourceLineNo">065</span> * input configuration.<a name="line.65"></a>
<span class="sourceLineNo">066</span> */<a name="line.66"></a>
<span class="sourceLineNo">067</span> public static final String OUTPUT_CONF_PREFIX = "hbase.mapred.output.";<a name="line.67"></a>
<span class="sourceLineNo">068</span><a name="line.68"></a>
<span class="sourceLineNo">069</span> /**<a name="line.69"></a>
<span class="sourceLineNo">070</span> * Optional job parameter to specify a peer cluster.<a name="line.70"></a>
<span class="sourceLineNo">071</span> * Used specifying remote cluster when copying between hbase clusters (the<a name="line.71"></a>
<span class="sourceLineNo">072</span> * source is picked up from &lt;code&gt;hbase-site.xml&lt;/code&gt;).<a name="line.72"></a>
<span class="sourceLineNo">073</span> * @see TableMapReduceUtil#initTableReducerJob(String, Class, org.apache.hadoop.mapreduce.Job, Class, String, String, String)<a name="line.73"></a>
<span class="sourceLineNo">074</span> */<a name="line.74"></a>
<span class="sourceLineNo">075</span> public static final String QUORUM_ADDRESS = OUTPUT_CONF_PREFIX + "quorum";<a name="line.75"></a>
<span class="sourceLineNo">076</span><a name="line.76"></a>
<span class="sourceLineNo">077</span> /** Optional job parameter to specify peer cluster's ZK client port */<a name="line.77"></a>
<span class="sourceLineNo">078</span> public static final String QUORUM_PORT = OUTPUT_CONF_PREFIX + "quorum.port";<a name="line.78"></a>
<span class="sourceLineNo">079</span><a name="line.79"></a>
<span class="sourceLineNo">080</span> /** Optional specification of the rs class name of the peer cluster */<a name="line.80"></a>
<span class="sourceLineNo">081</span> public static final String<a name="line.81"></a>
<span class="sourceLineNo">082</span> REGION_SERVER_CLASS = OUTPUT_CONF_PREFIX + "rs.class";<a name="line.82"></a>
<span class="sourceLineNo">083</span> /** Optional specification of the rs impl name of the peer cluster */<a name="line.83"></a>
<span class="sourceLineNo">084</span> public static final String<a name="line.84"></a>
<span class="sourceLineNo">085</span> REGION_SERVER_IMPL = OUTPUT_CONF_PREFIX + "rs.impl";<a name="line.85"></a>
<span class="sourceLineNo">086</span><a name="line.86"></a>
<span class="sourceLineNo">087</span> /** The configuration. */<a name="line.87"></a>
<span class="sourceLineNo">088</span> private Configuration conf = null;<a name="line.88"></a>
<span class="sourceLineNo">089</span><a name="line.89"></a>
<span class="sourceLineNo">090</span> /**<a name="line.90"></a>
<span class="sourceLineNo">091</span> * Writes the reducer output to an HBase table.<a name="line.91"></a>
<span class="sourceLineNo">092</span> */<a name="line.92"></a>
<span class="sourceLineNo">093</span> protected class TableRecordWriter<a name="line.93"></a>
<span class="sourceLineNo">094</span> extends RecordWriter&lt;KEY, Mutation&gt; {<a name="line.94"></a>
<span class="sourceLineNo">095</span><a name="line.95"></a>
<span class="sourceLineNo">096</span> private Connection connection;<a name="line.96"></a>
<span class="sourceLineNo">097</span> private BufferedMutator mutator;<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
<span class="sourceLineNo">099</span> /**<a name="line.99"></a>
<span class="sourceLineNo">100</span> * @throws IOException<a name="line.100"></a>
<span class="sourceLineNo">101</span> *<a name="line.101"></a>
<span class="sourceLineNo">102</span> */<a name="line.102"></a>
<span class="sourceLineNo">103</span> public TableRecordWriter() throws IOException {<a name="line.103"></a>
<span class="sourceLineNo">104</span> String tableName = conf.get(OUTPUT_TABLE);<a name="line.104"></a>
<span class="sourceLineNo">105</span> this.connection = ConnectionFactory.createConnection(conf);<a name="line.105"></a>
<span class="sourceLineNo">106</span> this.mutator = connection.getBufferedMutator(TableName.valueOf(tableName));<a name="line.106"></a>
<span class="sourceLineNo">107</span> LOG.info("Created table instance for " + tableName);<a name="line.107"></a>
<span class="sourceLineNo">108</span> }<a name="line.108"></a>
<span class="sourceLineNo">109</span> /**<a name="line.109"></a>
<span class="sourceLineNo">110</span> * Closes the writer, in this case flush table commits.<a name="line.110"></a>
<span class="sourceLineNo">111</span> *<a name="line.111"></a>
<span class="sourceLineNo">112</span> * @param context The context.<a name="line.112"></a>
<span class="sourceLineNo">113</span> * @throws IOException When closing the writer fails.<a name="line.113"></a>
<span class="sourceLineNo">114</span> * @see RecordWriter#close(TaskAttemptContext)<a name="line.114"></a>
<span class="sourceLineNo">115</span> */<a name="line.115"></a>
<span class="sourceLineNo">116</span> @Override<a name="line.116"></a>
<span class="sourceLineNo">117</span> public void close(TaskAttemptContext context) throws IOException {<a name="line.117"></a>
<span class="sourceLineNo">118</span> try {<a name="line.118"></a>
<span class="sourceLineNo">119</span> if (mutator != null) {<a name="line.119"></a>
<span class="sourceLineNo">120</span> mutator.close();<a name="line.120"></a>
<span class="sourceLineNo">121</span> }<a name="line.121"></a>
<span class="sourceLineNo">122</span> } finally {<a name="line.122"></a>
<span class="sourceLineNo">123</span> if (connection != null) {<a name="line.123"></a>
<span class="sourceLineNo">124</span> connection.close();<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> }<a name="line.127"></a>
<span class="sourceLineNo">128</span><a name="line.128"></a>
<span class="sourceLineNo">129</span> /**<a name="line.129"></a>
<span class="sourceLineNo">130</span> * Writes a key/value pair into the table.<a name="line.130"></a>
<span class="sourceLineNo">131</span> *<a name="line.131"></a>
<span class="sourceLineNo">132</span> * @param key The key.<a name="line.132"></a>
<span class="sourceLineNo">133</span> * @param value The value.<a name="line.133"></a>
<span class="sourceLineNo">134</span> * @throws IOException When writing fails.<a name="line.134"></a>
<span class="sourceLineNo">135</span> * @see RecordWriter#write(Object, Object)<a name="line.135"></a>
<span class="sourceLineNo">136</span> */<a name="line.136"></a>
<span class="sourceLineNo">137</span> @Override<a name="line.137"></a>
<span class="sourceLineNo">138</span> public void write(KEY key, Mutation value)<a name="line.138"></a>
<span class="sourceLineNo">139</span> throws IOException {<a name="line.139"></a>
<span class="sourceLineNo">140</span> if (!(value instanceof Put) &amp;&amp; !(value instanceof Delete)) {<a name="line.140"></a>
<span class="sourceLineNo">141</span> throw new IOException("Pass a Delete or a Put");<a name="line.141"></a>
<span class="sourceLineNo">142</span> }<a name="line.142"></a>
<span class="sourceLineNo">143</span> mutator.mutate(value);<a name="line.143"></a>
<span class="sourceLineNo">144</span> }<a name="line.144"></a>
<span class="sourceLineNo">145</span> }<a name="line.145"></a>
<span class="sourceLineNo">146</span><a name="line.146"></a>
<span class="sourceLineNo">147</span> /**<a name="line.147"></a>
<span class="sourceLineNo">148</span> * Creates a new record writer.<a name="line.148"></a>
<span class="sourceLineNo">149</span> *<a name="line.149"></a>
<span class="sourceLineNo">150</span> * Be aware that the baseline javadoc gives the impression that there is a single<a name="line.150"></a>
<span class="sourceLineNo">151</span> * {@link RecordWriter} per job but in HBase, it is more natural if we give you a new<a name="line.151"></a>
<span class="sourceLineNo">152</span> * RecordWriter per call of this method. You must close the returned RecordWriter when done.<a name="line.152"></a>
<span class="sourceLineNo">153</span> * Failure to do so will drop writes.<a name="line.153"></a>
<span class="sourceLineNo">154</span> *<a name="line.154"></a>
<span class="sourceLineNo">155</span> * @param context The current task context.<a name="line.155"></a>
<span class="sourceLineNo">156</span> * @return The newly created writer instance.<a name="line.156"></a>
<span class="sourceLineNo">157</span> * @throws IOException When creating the writer fails.<a name="line.157"></a>
<span class="sourceLineNo">158</span> * @throws InterruptedException When the job is cancelled.<a name="line.158"></a>
<span class="sourceLineNo">159</span> */<a name="line.159"></a>
<span class="sourceLineNo">160</span> @Override<a name="line.160"></a>
<span class="sourceLineNo">161</span> public RecordWriter&lt;KEY, Mutation&gt; getRecordWriter(TaskAttemptContext context)<a name="line.161"></a>
<span class="sourceLineNo">162</span> throws IOException, InterruptedException {<a name="line.162"></a>
<span class="sourceLineNo">163</span> return new TableRecordWriter();<a name="line.163"></a>
<span class="sourceLineNo">164</span> }<a name="line.164"></a>
<span class="sourceLineNo">165</span><a name="line.165"></a>
<span class="sourceLineNo">166</span> /**<a name="line.166"></a>
<span class="sourceLineNo">167</span> * Checks if the output table exists and is enabled.<a name="line.167"></a>
<span class="sourceLineNo">168</span> *<a name="line.168"></a>
<span class="sourceLineNo">169</span> * @param context The current context.<a name="line.169"></a>
<span class="sourceLineNo">170</span> * @throws IOException When the check fails.<a name="line.170"></a>
<span class="sourceLineNo">171</span> * @throws InterruptedException When the job is aborted.<a name="line.171"></a>
<span class="sourceLineNo">172</span> * @see OutputFormat#checkOutputSpecs(JobContext)<a name="line.172"></a>
<span class="sourceLineNo">173</span> */<a name="line.173"></a>
<span class="sourceLineNo">174</span> @Override<a name="line.174"></a>
<span class="sourceLineNo">175</span> public void checkOutputSpecs(JobContext context) throws IOException, InterruptedException {<a name="line.175"></a>
<span class="sourceLineNo">176</span> Configuration hConf = getConf();<a name="line.176"></a>
<span class="sourceLineNo">177</span> if (hConf == null) {<a name="line.177"></a>
<span class="sourceLineNo">178</span> hConf = context.getConfiguration();<a name="line.178"></a>
<span class="sourceLineNo">179</span> }<a name="line.179"></a>
<span class="sourceLineNo">180</span><a name="line.180"></a>
<span class="sourceLineNo">181</span> try (Connection connection = ConnectionFactory.createConnection(hConf);<a name="line.181"></a>
<span class="sourceLineNo">182</span> Admin admin = connection.getAdmin()) {<a name="line.182"></a>
<span class="sourceLineNo">183</span> TableName tableName = TableName.valueOf(hConf.get(OUTPUT_TABLE));<a name="line.183"></a>
<span class="sourceLineNo">184</span> if (!admin.tableExists(tableName)) {<a name="line.184"></a>
<span class="sourceLineNo">185</span> throw new TableNotFoundException("Can't write, table does not exist:" +<a name="line.185"></a>
<span class="sourceLineNo">186</span> tableName.getNameAsString());<a name="line.186"></a>
<span class="sourceLineNo">187</span> }<a name="line.187"></a>
<span class="sourceLineNo">188</span><a name="line.188"></a>
<span class="sourceLineNo">189</span> if (!admin.isTableEnabled(tableName)) {<a name="line.189"></a>
<span class="sourceLineNo">190</span> throw new TableNotEnabledException("Can't write, table is not enabled: " +<a name="line.190"></a>
<span class="sourceLineNo">191</span> tableName.getNameAsString());<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>
<span class="sourceLineNo">194</span> }<a name="line.194"></a>
<span class="sourceLineNo">195</span><a name="line.195"></a>
<span class="sourceLineNo">196</span> /**<a name="line.196"></a>
<span class="sourceLineNo">197</span> * Returns the output committer.<a name="line.197"></a>
<span class="sourceLineNo">198</span> *<a name="line.198"></a>
<span class="sourceLineNo">199</span> * @param context The current context.<a name="line.199"></a>
<span class="sourceLineNo">200</span> * @return The committer.<a name="line.200"></a>
<span class="sourceLineNo">201</span> * @throws IOException When creating the committer fails.<a name="line.201"></a>
<span class="sourceLineNo">202</span> * @throws InterruptedException When the job is aborted.<a name="line.202"></a>
<span class="sourceLineNo">203</span> * @see OutputFormat#getOutputCommitter(TaskAttemptContext)<a name="line.203"></a>
<span class="sourceLineNo">204</span> */<a name="line.204"></a>
<span class="sourceLineNo">205</span> @Override<a name="line.205"></a>
<span class="sourceLineNo">206</span> public OutputCommitter getOutputCommitter(TaskAttemptContext context)<a name="line.206"></a>
<span class="sourceLineNo">207</span> throws IOException, InterruptedException {<a name="line.207"></a>
<span class="sourceLineNo">208</span> return new TableOutputCommitter();<a name="line.208"></a>
<span class="sourceLineNo">209</span> }<a name="line.209"></a>
<span class="sourceLineNo">210</span><a name="line.210"></a>
<span class="sourceLineNo">211</span> @Override<a name="line.211"></a>
<span class="sourceLineNo">212</span> public Configuration getConf() {<a name="line.212"></a>
<span class="sourceLineNo">213</span> return conf;<a name="line.213"></a>
<span class="sourceLineNo">214</span> }<a name="line.214"></a>
<span class="sourceLineNo">215</span><a name="line.215"></a>
<span class="sourceLineNo">216</span> @Override<a name="line.216"></a>
<span class="sourceLineNo">217</span> public void setConf(Configuration otherConf) {<a name="line.217"></a>
<span class="sourceLineNo">218</span> String tableName = otherConf.get(OUTPUT_TABLE);<a name="line.218"></a>
<span class="sourceLineNo">219</span> if(tableName == null || tableName.length() &lt;= 0) {<a name="line.219"></a>
<span class="sourceLineNo">220</span> throw new IllegalArgumentException("Must specify table name");<a name="line.220"></a>
<span class="sourceLineNo">221</span> }<a name="line.221"></a>
<span class="sourceLineNo">222</span><a name="line.222"></a>
<span class="sourceLineNo">223</span> String address = otherConf.get(QUORUM_ADDRESS);<a name="line.223"></a>
<span class="sourceLineNo">224</span> int zkClientPort = otherConf.getInt(QUORUM_PORT, 0);<a name="line.224"></a>
<span class="sourceLineNo">225</span> String serverClass = otherConf.get(REGION_SERVER_CLASS);<a name="line.225"></a>
<span class="sourceLineNo">226</span> String serverImpl = otherConf.get(REGION_SERVER_IMPL);<a name="line.226"></a>
<span class="sourceLineNo">227</span><a name="line.227"></a>
<span class="sourceLineNo">228</span> try {<a name="line.228"></a>
<span class="sourceLineNo">229</span> this.conf = HBaseConfiguration.createClusterConf(otherConf, address, OUTPUT_CONF_PREFIX);<a name="line.229"></a>
<span class="sourceLineNo">230</span><a name="line.230"></a>
<span class="sourceLineNo">231</span> if (serverClass != null) {<a name="line.231"></a>
<span class="sourceLineNo">232</span> this.conf.set(HConstants.REGION_SERVER_IMPL, serverImpl);<a name="line.232"></a>
<span class="sourceLineNo">233</span> }<a name="line.233"></a>
<span class="sourceLineNo">234</span> if (zkClientPort != 0) {<a name="line.234"></a>
<span class="sourceLineNo">235</span> this.conf.setInt(HConstants.ZOOKEEPER_CLIENT_PORT, zkClientPort);<a name="line.235"></a>
<span class="sourceLineNo">236</span> }<a name="line.236"></a>
<span class="sourceLineNo">237</span> } catch(IOException e) {<a name="line.237"></a>
<span class="sourceLineNo">238</span> LOG.error(e.toString(), e);<a name="line.238"></a>
<span class="sourceLineNo">239</span> throw new RuntimeException(e);<a name="line.239"></a>
<span class="sourceLineNo">240</span> }<a name="line.240"></a>
<span class="sourceLineNo">241</span> }<a name="line.241"></a>
<span class="sourceLineNo">242</span>}<a name="line.242"></a>
</pre>
</div>
</body>
</html>