blob: f2f0bd3226bb50d377330f68be2dcd25f8105dcb [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.util.HashMap;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.Map;<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.TableName;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.client.BufferedMutator;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.client.Delete;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.Durability;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.client.Mutation;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.client.Put;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<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.mapreduce.JobContext;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.mapreduce.OutputCommitter;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.mapreduce.OutputFormat;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.mapreduce.RecordWriter;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.slf4j.Logger;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.slf4j.LoggerFactory;<a name="line.42"></a>
<span class="sourceLineNo">043</span><a name="line.43"></a>
<span class="sourceLineNo">044</span>/**<a name="line.44"></a>
<span class="sourceLineNo">045</span> * &lt;p&gt;<a name="line.45"></a>
<span class="sourceLineNo">046</span> * Hadoop output format that writes to one or more HBase tables. The key is taken to be the table<a name="line.46"></a>
<span class="sourceLineNo">047</span> * name while the output value &lt;em&gt;must&lt;/em&gt; be either a {@link Put} or a {@link Delete} instance.<a name="line.47"></a>
<span class="sourceLineNo">048</span> * All tables must already exist, and all Puts and Deletes must reference only valid column<a name="line.48"></a>
<span class="sourceLineNo">049</span> * families.<a name="line.49"></a>
<span class="sourceLineNo">050</span> * &lt;/p&gt;<a name="line.50"></a>
<span class="sourceLineNo">051</span> * &lt;p&gt;<a name="line.51"></a>
<span class="sourceLineNo">052</span> * Write-ahead logging (WAL) for Puts can be disabled by setting {@link #WAL_PROPERTY} to<a name="line.52"></a>
<span class="sourceLineNo">053</span> * {@link #WAL_OFF}. Default value is {@link #WAL_ON}. Note that disabling write-ahead logging is<a name="line.53"></a>
<span class="sourceLineNo">054</span> * only appropriate for jobs where loss of data due to region server failure can be tolerated (for<a name="line.54"></a>
<span class="sourceLineNo">055</span> * example, because it is easy to rerun a bulk import).<a name="line.55"></a>
<span class="sourceLineNo">056</span> * &lt;/p&gt;<a name="line.56"></a>
<span class="sourceLineNo">057</span> */<a name="line.57"></a>
<span class="sourceLineNo">058</span>@InterfaceAudience.Public<a name="line.58"></a>
<span class="sourceLineNo">059</span>public class MultiTableOutputFormat extends OutputFormat&lt;ImmutableBytesWritable, Mutation&gt; {<a name="line.59"></a>
<span class="sourceLineNo">060</span> /** Set this to {@link #WAL_OFF} to turn off write-ahead logging (WAL) */<a name="line.60"></a>
<span class="sourceLineNo">061</span> public static final String WAL_PROPERTY = "hbase.mapreduce.multitableoutputformat.wal";<a name="line.61"></a>
<span class="sourceLineNo">062</span> /** Property value to use write-ahead logging */<a name="line.62"></a>
<span class="sourceLineNo">063</span> public static final boolean WAL_ON = true;<a name="line.63"></a>
<span class="sourceLineNo">064</span> /** Property value to disable write-ahead logging */<a name="line.64"></a>
<span class="sourceLineNo">065</span> public static final boolean WAL_OFF = false;<a name="line.65"></a>
<span class="sourceLineNo">066</span><a name="line.66"></a>
<span class="sourceLineNo">067</span> /**<a name="line.67"></a>
<span class="sourceLineNo">068</span> * Record writer for outputting to multiple HTables.<a name="line.68"></a>
<span class="sourceLineNo">069</span> */<a name="line.69"></a>
<span class="sourceLineNo">070</span> protected static class MultiTableRecordWriter<a name="line.70"></a>
<span class="sourceLineNo">071</span> extends RecordWriter&lt;ImmutableBytesWritable, Mutation&gt; {<a name="line.71"></a>
<span class="sourceLineNo">072</span> private static final Logger LOG = LoggerFactory.getLogger(MultiTableRecordWriter.class);<a name="line.72"></a>
<span class="sourceLineNo">073</span> Connection connection;<a name="line.73"></a>
<span class="sourceLineNo">074</span> Map&lt;ImmutableBytesWritable, BufferedMutator&gt; mutatorMap = new HashMap&lt;&gt;();<a name="line.74"></a>
<span class="sourceLineNo">075</span> Configuration conf;<a name="line.75"></a>
<span class="sourceLineNo">076</span> boolean useWriteAheadLogging;<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> * HBaseConfiguration to used whether to use write ahead logging. This can be turned off (<a name="line.79"></a>
<span class="sourceLineNo">080</span> * &lt;tt&gt;false&lt;/tt&gt;) to improve performance when bulk loading data.<a name="line.80"></a>
<span class="sourceLineNo">081</span> */<a name="line.81"></a>
<span class="sourceLineNo">082</span> public MultiTableRecordWriter(Configuration conf, boolean useWriteAheadLogging)<a name="line.82"></a>
<span class="sourceLineNo">083</span> throws IOException {<a name="line.83"></a>
<span class="sourceLineNo">084</span> LOG.debug(<a name="line.84"></a>
<span class="sourceLineNo">085</span> "Created new MultiTableRecordReader with WAL " + (useWriteAheadLogging ? "on" : "off"));<a name="line.85"></a>
<span class="sourceLineNo">086</span> this.conf = conf;<a name="line.86"></a>
<span class="sourceLineNo">087</span> this.useWriteAheadLogging = useWriteAheadLogging;<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> /**<a name="line.90"></a>
<span class="sourceLineNo">091</span> * the name of the table, as a string<a name="line.91"></a>
<span class="sourceLineNo">092</span> * @return the named mutator if there is a problem opening a table<a name="line.92"></a>
<span class="sourceLineNo">093</span> */<a name="line.93"></a>
<span class="sourceLineNo">094</span> BufferedMutator getBufferedMutator(ImmutableBytesWritable tableName) throws IOException {<a name="line.94"></a>
<span class="sourceLineNo">095</span> if (this.connection == null) {<a name="line.95"></a>
<span class="sourceLineNo">096</span> this.connection = ConnectionFactory.createConnection(conf);<a name="line.96"></a>
<span class="sourceLineNo">097</span> }<a name="line.97"></a>
<span class="sourceLineNo">098</span> if (!mutatorMap.containsKey(tableName)) {<a name="line.98"></a>
<span class="sourceLineNo">099</span> LOG.debug("Opening HTable \"" + Bytes.toString(tableName.get()) + "\" for writing");<a name="line.99"></a>
<span class="sourceLineNo">100</span><a name="line.100"></a>
<span class="sourceLineNo">101</span> BufferedMutator mutator = connection.getBufferedMutator(TableName.valueOf(tableName.get()));<a name="line.101"></a>
<span class="sourceLineNo">102</span> mutatorMap.put(tableName, mutator);<a name="line.102"></a>
<span class="sourceLineNo">103</span> }<a name="line.103"></a>
<span class="sourceLineNo">104</span> return mutatorMap.get(tableName);<a name="line.104"></a>
<span class="sourceLineNo">105</span> }<a name="line.105"></a>
<span class="sourceLineNo">106</span><a name="line.106"></a>
<span class="sourceLineNo">107</span> @Override<a name="line.107"></a>
<span class="sourceLineNo">108</span> public void close(TaskAttemptContext context) throws IOException {<a name="line.108"></a>
<span class="sourceLineNo">109</span> for (BufferedMutator mutator : mutatorMap.values()) {<a name="line.109"></a>
<span class="sourceLineNo">110</span> mutator.close();<a name="line.110"></a>
<span class="sourceLineNo">111</span> }<a name="line.111"></a>
<span class="sourceLineNo">112</span> if (connection != null) {<a name="line.112"></a>
<span class="sourceLineNo">113</span> connection.close();<a name="line.113"></a>
<span class="sourceLineNo">114</span> }<a name="line.114"></a>
<span class="sourceLineNo">115</span> }<a name="line.115"></a>
<span class="sourceLineNo">116</span><a name="line.116"></a>
<span class="sourceLineNo">117</span> /**<a name="line.117"></a>
<span class="sourceLineNo">118</span> * Writes an action (Put or Delete) to the specified table. the table being updated. the update,<a name="line.118"></a>
<span class="sourceLineNo">119</span> * either a put or a delete. if the action is not a put or a delete.<a name="line.119"></a>
<span class="sourceLineNo">120</span> */<a name="line.120"></a>
<span class="sourceLineNo">121</span> @Override<a name="line.121"></a>
<span class="sourceLineNo">122</span> public void write(ImmutableBytesWritable tableName, Mutation action) throws IOException {<a name="line.122"></a>
<span class="sourceLineNo">123</span> BufferedMutator mutator = getBufferedMutator(tableName);<a name="line.123"></a>
<span class="sourceLineNo">124</span> // The actions are not immutable, so we defensively copy them<a name="line.124"></a>
<span class="sourceLineNo">125</span> if (action instanceof Put) {<a name="line.125"></a>
<span class="sourceLineNo">126</span> Put put = new Put((Put) action);<a name="line.126"></a>
<span class="sourceLineNo">127</span> put.setDurability(useWriteAheadLogging ? Durability.SYNC_WAL : Durability.SKIP_WAL);<a name="line.127"></a>
<span class="sourceLineNo">128</span> mutator.mutate(put);<a name="line.128"></a>
<span class="sourceLineNo">129</span> } else if (action instanceof Delete) {<a name="line.129"></a>
<span class="sourceLineNo">130</span> Delete delete = new Delete((Delete) action);<a name="line.130"></a>
<span class="sourceLineNo">131</span> mutator.mutate(delete);<a name="line.131"></a>
<span class="sourceLineNo">132</span> } else throw new IllegalArgumentException("action must be either Delete or Put");<a name="line.132"></a>
<span class="sourceLineNo">133</span> }<a name="line.133"></a>
<span class="sourceLineNo">134</span> }<a name="line.134"></a>
<span class="sourceLineNo">135</span><a name="line.135"></a>
<span class="sourceLineNo">136</span> @Override<a name="line.136"></a>
<span class="sourceLineNo">137</span> public void checkOutputSpecs(JobContext context) throws IOException, InterruptedException {<a name="line.137"></a>
<span class="sourceLineNo">138</span> // we can't know ahead of time if it's going to blow up when the user<a name="line.138"></a>
<span class="sourceLineNo">139</span> // passes a table name that doesn't exist, so nothing useful here.<a name="line.139"></a>
<span class="sourceLineNo">140</span> }<a name="line.140"></a>
<span class="sourceLineNo">141</span><a name="line.141"></a>
<span class="sourceLineNo">142</span> @Override<a name="line.142"></a>
<span class="sourceLineNo">143</span> public OutputCommitter getOutputCommitter(TaskAttemptContext context)<a name="line.143"></a>
<span class="sourceLineNo">144</span> throws IOException, InterruptedException {<a name="line.144"></a>
<span class="sourceLineNo">145</span> return new TableOutputCommitter();<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> @Override<a name="line.148"></a>
<span class="sourceLineNo">149</span> public RecordWriter&lt;ImmutableBytesWritable, Mutation&gt; getRecordWriter(TaskAttemptContext context)<a name="line.149"></a>
<span class="sourceLineNo">150</span> throws IOException, InterruptedException {<a name="line.150"></a>
<span class="sourceLineNo">151</span> Configuration conf = context.getConfiguration();<a name="line.151"></a>
<span class="sourceLineNo">152</span> return new MultiTableRecordWriter(HBaseConfiguration.create(conf),<a name="line.152"></a>
<span class="sourceLineNo">153</span> conf.getBoolean(WAL_PROPERTY, WAL_ON));<a name="line.153"></a>
<span class="sourceLineNo">154</span> }<a name="line.154"></a>
<span class="sourceLineNo">155</span><a name="line.155"></a>
<span class="sourceLineNo">156</span>}<a name="line.156"></a>
</pre>
</div>
</body>
</html>