| <!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.backup.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 org.apache.hadoop.conf.Configuration;<a name="line.21"></a> |
| <span class="sourceLineNo">022</span>import org.apache.hadoop.conf.Configured;<a name="line.22"></a> |
| <span class="sourceLineNo">023</span>import org.apache.hadoop.fs.Path;<a name="line.23"></a> |
| <span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.Cell;<a name="line.24"></a> |
| <span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.25"></a> |
| <span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.26"></a> |
| <span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.TableName;<a name="line.27"></a> |
| <span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.28"></a> |
| <span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.29"></a> |
| <span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.RegionLocator;<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.mapreduce.CellSortReducer;<a name="line.33"></a> |
| <span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.mapreduce.HFileInputFormat;<a name="line.34"></a> |
| <span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.mapreduce.HFileOutputFormat2;<a name="line.35"></a> |
| <span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.mapreduce.TableMapReduceUtil;<a name="line.36"></a> |
| <span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.37"></a> |
| <span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.util.MapReduceExtendedCell;<a name="line.38"></a> |
| <span class="sourceLineNo">039</span>import org.apache.hadoop.io.NullWritable;<a name="line.39"></a> |
| <span class="sourceLineNo">040</span>import org.apache.hadoop.mapreduce.Job;<a name="line.40"></a> |
| <span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.41"></a> |
| <span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.42"></a> |
| <span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.43"></a> |
| <span class="sourceLineNo">044</span>import org.apache.hadoop.util.Tool;<a name="line.44"></a> |
| <span class="sourceLineNo">045</span>import org.apache.hadoop.util.ToolRunner;<a name="line.45"></a> |
| <span class="sourceLineNo">046</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.46"></a> |
| <span class="sourceLineNo">047</span>import org.slf4j.Logger;<a name="line.47"></a> |
| <span class="sourceLineNo">048</span>import org.slf4j.LoggerFactory;<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> * A tool to split HFiles into new region boundaries as a MapReduce job. The tool generates HFiles<a name="line.51"></a> |
| <span class="sourceLineNo">052</span> * for later bulk importing.<a name="line.52"></a> |
| <span class="sourceLineNo">053</span> */<a name="line.53"></a> |
| <span class="sourceLineNo">054</span>@InterfaceAudience.Private<a name="line.54"></a> |
| <span class="sourceLineNo">055</span>public class MapReduceHFileSplitterJob extends Configured implements Tool {<a name="line.55"></a> |
| <span class="sourceLineNo">056</span> private static final Logger LOG = LoggerFactory.getLogger(MapReduceHFileSplitterJob.class);<a name="line.56"></a> |
| <span class="sourceLineNo">057</span> final static String NAME = "HFileSplitterJob";<a name="line.57"></a> |
| <span class="sourceLineNo">058</span> public final static String BULK_OUTPUT_CONF_KEY = "hfile.bulk.output";<a name="line.58"></a> |
| <span class="sourceLineNo">059</span> public final static String TABLES_KEY = "hfile.input.tables";<a name="line.59"></a> |
| <span class="sourceLineNo">060</span> public final static String TABLE_MAP_KEY = "hfile.input.tablesmap";<a name="line.60"></a> |
| <span class="sourceLineNo">061</span> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.61"></a> |
| <span class="sourceLineNo">062</span><a name="line.62"></a> |
| <span class="sourceLineNo">063</span> public MapReduceHFileSplitterJob() {<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> protected MapReduceHFileSplitterJob(final Configuration c) {<a name="line.66"></a> |
| <span class="sourceLineNo">067</span> super(c);<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> /**<a name="line.70"></a> |
| <span class="sourceLineNo">071</span> * A mapper that just writes out cells. This one can be used together with {@link CellSortReducer}<a name="line.71"></a> |
| <span class="sourceLineNo">072</span> */<a name="line.72"></a> |
| <span class="sourceLineNo">073</span> static class HFileCellMapper extends Mapper<NullWritable, Cell, ImmutableBytesWritable, Cell> {<a name="line.73"></a> |
| <span class="sourceLineNo">074</span><a name="line.74"></a> |
| <span class="sourceLineNo">075</span> @Override<a name="line.75"></a> |
| <span class="sourceLineNo">076</span> public void map(NullWritable key, Cell value, Context context)<a name="line.76"></a> |
| <span class="sourceLineNo">077</span> throws IOException, InterruptedException {<a name="line.77"></a> |
| <span class="sourceLineNo">078</span> context.write(new ImmutableBytesWritable(CellUtil.cloneRow(value)),<a name="line.78"></a> |
| <span class="sourceLineNo">079</span> new MapReduceExtendedCell(value));<a name="line.79"></a> |
| <span class="sourceLineNo">080</span> }<a name="line.80"></a> |
| <span class="sourceLineNo">081</span><a name="line.81"></a> |
| <span class="sourceLineNo">082</span> @Override<a name="line.82"></a> |
| <span class="sourceLineNo">083</span> public void setup(Context context) throws IOException {<a name="line.83"></a> |
| <span class="sourceLineNo">084</span> // do nothing<a name="line.84"></a> |
| <span class="sourceLineNo">085</span> }<a name="line.85"></a> |
| <span class="sourceLineNo">086</span> }<a name="line.86"></a> |
| <span class="sourceLineNo">087</span><a name="line.87"></a> |
| <span class="sourceLineNo">088</span> /**<a name="line.88"></a> |
| <span class="sourceLineNo">089</span> * Sets up the actual job.<a name="line.89"></a> |
| <span class="sourceLineNo">090</span> * @param args The command line parameters.<a name="line.90"></a> |
| <span class="sourceLineNo">091</span> * @return The newly created job.<a name="line.91"></a> |
| <span class="sourceLineNo">092</span> * @throws IOException When setting up the job fails.<a name="line.92"></a> |
| <span class="sourceLineNo">093</span> */<a name="line.93"></a> |
| <span class="sourceLineNo">094</span> public Job createSubmittableJob(String[] args) throws IOException {<a name="line.94"></a> |
| <span class="sourceLineNo">095</span> Configuration conf = getConf();<a name="line.95"></a> |
| <span class="sourceLineNo">096</span> String inputDirs = args[0];<a name="line.96"></a> |
| <span class="sourceLineNo">097</span> String tabName = args[1];<a name="line.97"></a> |
| <span class="sourceLineNo">098</span> conf.setStrings(TABLES_KEY, tabName);<a name="line.98"></a> |
| <span class="sourceLineNo">099</span> conf.set(FileInputFormat.INPUT_DIR, inputDirs);<a name="line.99"></a> |
| <span class="sourceLineNo">100</span> Job job = Job.getInstance(conf,<a name="line.100"></a> |
| <span class="sourceLineNo">101</span> conf.get(JOB_NAME_CONF_KEY, NAME + "_" + EnvironmentEdgeManager.currentTime()));<a name="line.101"></a> |
| <span class="sourceLineNo">102</span> job.setJarByClass(MapReduceHFileSplitterJob.class);<a name="line.102"></a> |
| <span class="sourceLineNo">103</span> job.setInputFormatClass(HFileInputFormat.class);<a name="line.103"></a> |
| <span class="sourceLineNo">104</span> job.setMapOutputKeyClass(ImmutableBytesWritable.class);<a name="line.104"></a> |
| <span class="sourceLineNo">105</span> String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);<a name="line.105"></a> |
| <span class="sourceLineNo">106</span> if (hfileOutPath != null) {<a name="line.106"></a> |
| <span class="sourceLineNo">107</span> LOG.debug("add incremental job :" + hfileOutPath + " from " + inputDirs);<a name="line.107"></a> |
| <span class="sourceLineNo">108</span> TableName tableName = TableName.valueOf(tabName);<a name="line.108"></a> |
| <span class="sourceLineNo">109</span> job.setMapperClass(HFileCellMapper.class);<a name="line.109"></a> |
| <span class="sourceLineNo">110</span> job.setReducerClass(CellSortReducer.class);<a name="line.110"></a> |
| <span class="sourceLineNo">111</span> Path outputDir = new Path(hfileOutPath);<a name="line.111"></a> |
| <span class="sourceLineNo">112</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.112"></a> |
| <span class="sourceLineNo">113</span> job.setMapOutputValueClass(MapReduceExtendedCell.class);<a name="line.113"></a> |
| <span class="sourceLineNo">114</span> try (Connection conn = ConnectionFactory.createConnection(conf);<a name="line.114"></a> |
| <span class="sourceLineNo">115</span> Table table = conn.getTable(tableName);<a name="line.115"></a> |
| <span class="sourceLineNo">116</span> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {<a name="line.116"></a> |
| <span class="sourceLineNo">117</span> HFileOutputFormat2.configureIncrementalLoad(job, table.getDescriptor(), regionLocator);<a name="line.117"></a> |
| <span class="sourceLineNo">118</span> }<a name="line.118"></a> |
| <span class="sourceLineNo">119</span> LOG.debug("success configuring load incremental job");<a name="line.119"></a> |
| <span class="sourceLineNo">120</span><a name="line.120"></a> |
| <span class="sourceLineNo">121</span> TableMapReduceUtil.addDependencyJars(job.getConfiguration(),<a name="line.121"></a> |
| <span class="sourceLineNo">122</span> org.apache.hbase.thirdparty.com.google.common.base.Preconditions.class);<a name="line.122"></a> |
| <span class="sourceLineNo">123</span> } else {<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> throw new IOException("No bulk output directory specified");<a name="line.124"></a> |
| <span class="sourceLineNo">125</span> }<a name="line.125"></a> |
| <span class="sourceLineNo">126</span> return job;<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> * Print usage<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> * @param errorMsg Error message. Can be null.<a name="line.131"></a> |
| <span class="sourceLineNo">132</span> */<a name="line.132"></a> |
| <span class="sourceLineNo">133</span> private void usage(final String errorMsg) {<a name="line.133"></a> |
| <span class="sourceLineNo">134</span> if (errorMsg != null && errorMsg.length() > 0) {<a name="line.134"></a> |
| <span class="sourceLineNo">135</span> System.err.println("ERROR: " + errorMsg);<a name="line.135"></a> |
| <span class="sourceLineNo">136</span> }<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> System.err.println("Usage: " + NAME + " [options] <HFile inputdir(s)> <table>");<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> System.err.println("Read all HFile's for <table> and split them to <table> region boundaries.");<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> System.err.println("<table> table to load.\n");<a name="line.139"></a> |
| <span class="sourceLineNo">140</span> System.err.println("To generate HFiles for a bulk data load, pass the option:");<a name="line.140"></a> |
| <span class="sourceLineNo">141</span> System.err.println(" -D" + BULK_OUTPUT_CONF_KEY + "=/path/for/output");<a name="line.141"></a> |
| <span class="sourceLineNo">142</span> System.err.println("Other options:");<a name="line.142"></a> |
| <span class="sourceLineNo">143</span> System.err.println(" -D " + JOB_NAME_CONF_KEY<a name="line.143"></a> |
| <span class="sourceLineNo">144</span> + "=jobName - use the specified mapreduce job name for the HFile splitter");<a name="line.144"></a> |
| <span class="sourceLineNo">145</span> System.err.println("For performance also consider the following options:\n"<a name="line.145"></a> |
| <span class="sourceLineNo">146</span> + " -Dmapreduce.map.speculative=false\n" + " -Dmapreduce.reduce.speculative=false");<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> /**<a name="line.149"></a> |
| <span class="sourceLineNo">150</span> * Main entry point.<a name="line.150"></a> |
| <span class="sourceLineNo">151</span> * @param args The command line parameters.<a name="line.151"></a> |
| <span class="sourceLineNo">152</span> * @throws Exception When running the job fails.<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> */<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> public static void main(String[] args) throws Exception {<a name="line.154"></a> |
| <span class="sourceLineNo">155</span> int ret = ToolRunner.run(new MapReduceHFileSplitterJob(HBaseConfiguration.create()), args);<a name="line.155"></a> |
| <span class="sourceLineNo">156</span> System.exit(ret);<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> }<a name="line.157"></a> |
| <span class="sourceLineNo">158</span><a name="line.158"></a> |
| <span class="sourceLineNo">159</span> @Override<a name="line.159"></a> |
| <span class="sourceLineNo">160</span> public int run(String[] args) throws Exception {<a name="line.160"></a> |
| <span class="sourceLineNo">161</span> if (args.length < 2) {<a name="line.161"></a> |
| <span class="sourceLineNo">162</span> usage("Wrong number of arguments: " + args.length);<a name="line.162"></a> |
| <span class="sourceLineNo">163</span> return -1;<a name="line.163"></a> |
| <span class="sourceLineNo">164</span> }<a name="line.164"></a> |
| <span class="sourceLineNo">165</span> Job job = createSubmittableJob(args);<a name="line.165"></a> |
| <span class="sourceLineNo">166</span> int result = job.waitForCompletion(true) ? 0 : 1;<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> return result;<a name="line.167"></a> |
| <span class="sourceLineNo">168</span> }<a name="line.168"></a> |
| <span class="sourceLineNo">169</span>}<a name="line.169"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |