| <!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.Collections;<a name="line.21"></a> |
| <span class="sourceLineNo">022</span>import java.util.Iterator;<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.conf.Configured;<a name="line.24"></a> |
| <span class="sourceLineNo">025</span>import org.apache.hadoop.fs.FileStatus;<a name="line.25"></a> |
| <span class="sourceLineNo">026</span>import org.apache.hadoop.fs.FileSystem;<a name="line.26"></a> |
| <span class="sourceLineNo">027</span>import org.apache.hadoop.fs.Path;<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.CellBuilderFactory;<a name="line.29"></a> |
| <span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.CellBuilderType;<a name="line.30"></a> |
| <span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.CellComparator;<a name="line.31"></a> |
| <span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.32"></a> |
| <span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.33"></a> |
| <span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.TableName;<a name="line.34"></a> |
| <span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.35"></a> |
| <span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.36"></a> |
| <span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.Delete;<a name="line.37"></a> |
| <span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.client.Mutation;<a name="line.38"></a> |
| <span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.client.Put;<a name="line.39"></a> |
| <span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.client.Result;<a name="line.40"></a> |
| <span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.client.ResultScanner;<a name="line.41"></a> |
| <span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.42"></a> |
| <span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.client.Table;<a name="line.43"></a> |
| <span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.44"></a> |
| <span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.45"></a> |
| <span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.46"></a> |
| <span class="sourceLineNo">047</span>import org.apache.hadoop.mapreduce.Counters;<a name="line.47"></a> |
| <span class="sourceLineNo">048</span>import org.apache.hadoop.mapreduce.Job;<a name="line.48"></a> |
| <span class="sourceLineNo">049</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.49"></a> |
| <span class="sourceLineNo">050</span>import org.apache.hadoop.mapreduce.security.TokenCache;<a name="line.50"></a> |
| <span class="sourceLineNo">051</span>import org.apache.hadoop.util.GenericOptionsParser;<a name="line.51"></a> |
| <span class="sourceLineNo">052</span>import org.apache.hadoop.util.Tool;<a name="line.52"></a> |
| <span class="sourceLineNo">053</span>import org.apache.hadoop.util.ToolRunner;<a name="line.53"></a> |
| <span class="sourceLineNo">054</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.54"></a> |
| <span class="sourceLineNo">055</span>import org.slf4j.Logger;<a name="line.55"></a> |
| <span class="sourceLineNo">056</span>import org.slf4j.LoggerFactory;<a name="line.56"></a> |
| <span class="sourceLineNo">057</span><a name="line.57"></a> |
| <span class="sourceLineNo">058</span>import org.apache.hbase.thirdparty.com.google.common.base.Throwables;<a name="line.58"></a> |
| <span class="sourceLineNo">059</span><a name="line.59"></a> |
| <span class="sourceLineNo">060</span>@InterfaceAudience.Private<a name="line.60"></a> |
| <span class="sourceLineNo">061</span>public class SyncTable extends Configured implements Tool {<a name="line.61"></a> |
| <span class="sourceLineNo">062</span><a name="line.62"></a> |
| <span class="sourceLineNo">063</span> private static final Logger LOG = LoggerFactory.getLogger(SyncTable.class);<a name="line.63"></a> |
| <span class="sourceLineNo">064</span><a name="line.64"></a> |
| <span class="sourceLineNo">065</span> static final String SOURCE_HASH_DIR_CONF_KEY = "sync.table.source.hash.dir";<a name="line.65"></a> |
| <span class="sourceLineNo">066</span> static final String SOURCE_TABLE_CONF_KEY = "sync.table.source.table.name";<a name="line.66"></a> |
| <span class="sourceLineNo">067</span> static final String TARGET_TABLE_CONF_KEY = "sync.table.target.table.name";<a name="line.67"></a> |
| <span class="sourceLineNo">068</span> static final String SOURCE_ZK_CLUSTER_CONF_KEY = "sync.table.source.zk.cluster";<a name="line.68"></a> |
| <span class="sourceLineNo">069</span> static final String TARGET_ZK_CLUSTER_CONF_KEY = "sync.table.target.zk.cluster";<a name="line.69"></a> |
| <span class="sourceLineNo">070</span> static final String DRY_RUN_CONF_KEY = "sync.table.dry.run";<a name="line.70"></a> |
| <span class="sourceLineNo">071</span> static final String DO_DELETES_CONF_KEY = "sync.table.do.deletes";<a name="line.71"></a> |
| <span class="sourceLineNo">072</span> static final String DO_PUTS_CONF_KEY = "sync.table.do.puts";<a name="line.72"></a> |
| <span class="sourceLineNo">073</span> static final String IGNORE_TIMESTAMPS = "sync.table.ignore.timestamps";<a name="line.73"></a> |
| <span class="sourceLineNo">074</span><a name="line.74"></a> |
| <span class="sourceLineNo">075</span> Path sourceHashDir;<a name="line.75"></a> |
| <span class="sourceLineNo">076</span> String sourceTableName;<a name="line.76"></a> |
| <span class="sourceLineNo">077</span> String targetTableName;<a name="line.77"></a> |
| <span class="sourceLineNo">078</span><a name="line.78"></a> |
| <span class="sourceLineNo">079</span> String sourceZkCluster;<a name="line.79"></a> |
| <span class="sourceLineNo">080</span> String targetZkCluster;<a name="line.80"></a> |
| <span class="sourceLineNo">081</span> boolean dryRun;<a name="line.81"></a> |
| <span class="sourceLineNo">082</span> boolean doDeletes = true;<a name="line.82"></a> |
| <span class="sourceLineNo">083</span> boolean doPuts = true;<a name="line.83"></a> |
| <span class="sourceLineNo">084</span> boolean ignoreTimestamps;<a name="line.84"></a> |
| <span class="sourceLineNo">085</span><a name="line.85"></a> |
| <span class="sourceLineNo">086</span> Counters counters;<a name="line.86"></a> |
| <span class="sourceLineNo">087</span><a name="line.87"></a> |
| <span class="sourceLineNo">088</span> public SyncTable(Configuration conf) {<a name="line.88"></a> |
| <span class="sourceLineNo">089</span> super(conf);<a name="line.89"></a> |
| <span class="sourceLineNo">090</span> }<a name="line.90"></a> |
| <span class="sourceLineNo">091</span><a name="line.91"></a> |
| <span class="sourceLineNo">092</span> private void initCredentialsForHBase(String zookeeper, Job job) throws IOException {<a name="line.92"></a> |
| <span class="sourceLineNo">093</span> Configuration peerConf =<a name="line.93"></a> |
| <span class="sourceLineNo">094</span> HBaseConfiguration.createClusterConf(job.getConfiguration(), zookeeper);<a name="line.94"></a> |
| <span class="sourceLineNo">095</span> if ("kerberos".equalsIgnoreCase(peerConf.get("hbase.security.authentication"))) {<a name="line.95"></a> |
| <span class="sourceLineNo">096</span> TableMapReduceUtil.initCredentialsForCluster(job, peerConf);<a name="line.96"></a> |
| <span class="sourceLineNo">097</span> }<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> public Job createSubmittableJob(String[] args) throws IOException {<a name="line.100"></a> |
| <span class="sourceLineNo">101</span> FileSystem fs = sourceHashDir.getFileSystem(getConf());<a name="line.101"></a> |
| <span class="sourceLineNo">102</span> if (!fs.exists(sourceHashDir)) {<a name="line.102"></a> |
| <span class="sourceLineNo">103</span> throw new IOException("Source hash dir not found: " + sourceHashDir);<a name="line.103"></a> |
| <span class="sourceLineNo">104</span> }<a name="line.104"></a> |
| <span class="sourceLineNo">105</span><a name="line.105"></a> |
| <span class="sourceLineNo">106</span> Job job = Job.getInstance(getConf(),<a name="line.106"></a> |
| <span class="sourceLineNo">107</span> getConf().get("mapreduce.job.name", "syncTable_" + sourceTableName + "-" + targetTableName));<a name="line.107"></a> |
| <span class="sourceLineNo">108</span> Configuration jobConf = job.getConfiguration();<a name="line.108"></a> |
| <span class="sourceLineNo">109</span> if ("kerberos".equalsIgnoreCase(jobConf.get("hadoop.security.authentication"))) {<a name="line.109"></a> |
| <span class="sourceLineNo">110</span> TokenCache.obtainTokensForNamenodes(job.getCredentials(), new Path[] { sourceHashDir },<a name="line.110"></a> |
| <span class="sourceLineNo">111</span> getConf());<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> HashTable.TableHash tableHash = HashTable.TableHash.read(getConf(), sourceHashDir);<a name="line.114"></a> |
| <span class="sourceLineNo">115</span> LOG.info("Read source hash manifest: " + tableHash);<a name="line.115"></a> |
| <span class="sourceLineNo">116</span> LOG.info("Read " + tableHash.partitions.size() + " partition keys");<a name="line.116"></a> |
| <span class="sourceLineNo">117</span> if (!tableHash.tableName.equals(sourceTableName)) {<a name="line.117"></a> |
| <span class="sourceLineNo">118</span> LOG.warn("Table name mismatch - manifest indicates hash was taken from: "<a name="line.118"></a> |
| <span class="sourceLineNo">119</span> + tableHash.tableName + " but job is reading from: " + sourceTableName);<a name="line.119"></a> |
| <span class="sourceLineNo">120</span> }<a name="line.120"></a> |
| <span class="sourceLineNo">121</span> if (tableHash.numHashFiles != tableHash.partitions.size() + 1) {<a name="line.121"></a> |
| <span class="sourceLineNo">122</span> throw new RuntimeException("Hash data appears corrupt. The number of of hash files created"<a name="line.122"></a> |
| <span class="sourceLineNo">123</span> + " should be 1 more than the number of partition keys. However, the manifest file "<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> + " says numHashFiles=" + tableHash.numHashFiles + " but the number of partition keys"<a name="line.124"></a> |
| <span class="sourceLineNo">125</span> + " found in the partitions file is " + tableHash.partitions.size());<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> Path dataDir = new Path(sourceHashDir, HashTable.HASH_DATA_DIR);<a name="line.128"></a> |
| <span class="sourceLineNo">129</span> int dataSubdirCount = 0;<a name="line.129"></a> |
| <span class="sourceLineNo">130</span> for (FileStatus file : fs.listStatus(dataDir)) {<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> if (file.getPath().getName().startsWith(HashTable.OUTPUT_DATA_FILE_PREFIX)) {<a name="line.131"></a> |
| <span class="sourceLineNo">132</span> dataSubdirCount++;<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> if (dataSubdirCount != tableHash.numHashFiles) {<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> throw new RuntimeException("Hash data appears corrupt. The number of of hash files created"<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> + " should be 1 more than the number of partition keys. However, the number of data dirs"<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> + " found is " + dataSubdirCount + " but the number of partition keys"<a name="line.139"></a> |
| <span class="sourceLineNo">140</span> + " found in the partitions file is " + tableHash.partitions.size());<a name="line.140"></a> |
| <span class="sourceLineNo">141</span> }<a name="line.141"></a> |
| <span class="sourceLineNo">142</span><a name="line.142"></a> |
| <span class="sourceLineNo">143</span> job.setJarByClass(HashTable.class);<a name="line.143"></a> |
| <span class="sourceLineNo">144</span> jobConf.set(SOURCE_HASH_DIR_CONF_KEY, sourceHashDir.toString());<a name="line.144"></a> |
| <span class="sourceLineNo">145</span> jobConf.set(SOURCE_TABLE_CONF_KEY, sourceTableName);<a name="line.145"></a> |
| <span class="sourceLineNo">146</span> jobConf.set(TARGET_TABLE_CONF_KEY, targetTableName);<a name="line.146"></a> |
| <span class="sourceLineNo">147</span> if (sourceZkCluster != null) {<a name="line.147"></a> |
| <span class="sourceLineNo">148</span> jobConf.set(SOURCE_ZK_CLUSTER_CONF_KEY, sourceZkCluster);<a name="line.148"></a> |
| <span class="sourceLineNo">149</span> initCredentialsForHBase(sourceZkCluster, job);<a name="line.149"></a> |
| <span class="sourceLineNo">150</span> }<a name="line.150"></a> |
| <span class="sourceLineNo">151</span> if (targetZkCluster != null) {<a name="line.151"></a> |
| <span class="sourceLineNo">152</span> jobConf.set(TARGET_ZK_CLUSTER_CONF_KEY, targetZkCluster);<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> initCredentialsForHBase(targetZkCluster, job);<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> }<a name="line.154"></a> |
| <span class="sourceLineNo">155</span> jobConf.setBoolean(DRY_RUN_CONF_KEY, dryRun);<a name="line.155"></a> |
| <span class="sourceLineNo">156</span> jobConf.setBoolean(DO_DELETES_CONF_KEY, doDeletes);<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> jobConf.setBoolean(DO_PUTS_CONF_KEY, doPuts);<a name="line.157"></a> |
| <span class="sourceLineNo">158</span> jobConf.setBoolean(IGNORE_TIMESTAMPS, ignoreTimestamps);<a name="line.158"></a> |
| <span class="sourceLineNo">159</span><a name="line.159"></a> |
| <span class="sourceLineNo">160</span> TableMapReduceUtil.initTableMapperJob(targetTableName, tableHash.initScan(), SyncMapper.class,<a name="line.160"></a> |
| <span class="sourceLineNo">161</span> null, null, job);<a name="line.161"></a> |
| <span class="sourceLineNo">162</span><a name="line.162"></a> |
| <span class="sourceLineNo">163</span> job.setNumReduceTasks(0);<a name="line.163"></a> |
| <span class="sourceLineNo">164</span><a name="line.164"></a> |
| <span class="sourceLineNo">165</span> if (dryRun) {<a name="line.165"></a> |
| <span class="sourceLineNo">166</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> } else {<a name="line.167"></a> |
| <span class="sourceLineNo">168</span> // No reducers. Just write straight to table. Call initTableReducerJob<a name="line.168"></a> |
| <span class="sourceLineNo">169</span> // because it sets up the TableOutputFormat.<a name="line.169"></a> |
| <span class="sourceLineNo">170</span> TableMapReduceUtil.initTableReducerJob(targetTableName, null, job, null, targetZkCluster,<a name="line.170"></a> |
| <span class="sourceLineNo">171</span> null, null);<a name="line.171"></a> |
| <span class="sourceLineNo">172</span><a name="line.172"></a> |
| <span class="sourceLineNo">173</span> // would be nice to add an option for bulk load instead<a name="line.173"></a> |
| <span class="sourceLineNo">174</span> }<a name="line.174"></a> |
| <span class="sourceLineNo">175</span><a name="line.175"></a> |
| <span class="sourceLineNo">176</span> // Obtain an authentication token, for the specified cluster, on behalf of the current user<a name="line.176"></a> |
| <span class="sourceLineNo">177</span> if (sourceZkCluster != null) {<a name="line.177"></a> |
| <span class="sourceLineNo">178</span> Configuration peerConf =<a name="line.178"></a> |
| <span class="sourceLineNo">179</span> HBaseConfiguration.createClusterConf(job.getConfiguration(), sourceZkCluster);<a name="line.179"></a> |
| <span class="sourceLineNo">180</span> TableMapReduceUtil.initCredentialsForCluster(job, peerConf);<a name="line.180"></a> |
| <span class="sourceLineNo">181</span> }<a name="line.181"></a> |
| <span class="sourceLineNo">182</span> return job;<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> public static class SyncMapper extends TableMapper<ImmutableBytesWritable, Mutation> {<a name="line.185"></a> |
| <span class="sourceLineNo">186</span> Path sourceHashDir;<a name="line.186"></a> |
| <span class="sourceLineNo">187</span><a name="line.187"></a> |
| <span class="sourceLineNo">188</span> Connection sourceConnection;<a name="line.188"></a> |
| <span class="sourceLineNo">189</span> Connection targetConnection;<a name="line.189"></a> |
| <span class="sourceLineNo">190</span> Table sourceTable;<a name="line.190"></a> |
| <span class="sourceLineNo">191</span> Table targetTable;<a name="line.191"></a> |
| <span class="sourceLineNo">192</span> boolean dryRun;<a name="line.192"></a> |
| <span class="sourceLineNo">193</span> boolean doDeletes = true;<a name="line.193"></a> |
| <span class="sourceLineNo">194</span> boolean doPuts = true;<a name="line.194"></a> |
| <span class="sourceLineNo">195</span> boolean ignoreTimestamp;<a name="line.195"></a> |
| <span class="sourceLineNo">196</span><a name="line.196"></a> |
| <span class="sourceLineNo">197</span> HashTable.TableHash sourceTableHash;<a name="line.197"></a> |
| <span class="sourceLineNo">198</span> HashTable.TableHash.Reader sourceHashReader;<a name="line.198"></a> |
| <span class="sourceLineNo">199</span> ImmutableBytesWritable currentSourceHash;<a name="line.199"></a> |
| <span class="sourceLineNo">200</span> ImmutableBytesWritable nextSourceKey;<a name="line.200"></a> |
| <span class="sourceLineNo">201</span> HashTable.ResultHasher targetHasher;<a name="line.201"></a> |
| <span class="sourceLineNo">202</span><a name="line.202"></a> |
| <span class="sourceLineNo">203</span> Throwable mapperException;<a name="line.203"></a> |
| <span class="sourceLineNo">204</span><a name="line.204"></a> |
| <span class="sourceLineNo">205</span> public static enum Counter {<a name="line.205"></a> |
| <span class="sourceLineNo">206</span> BATCHES,<a name="line.206"></a> |
| <span class="sourceLineNo">207</span> HASHES_MATCHED,<a name="line.207"></a> |
| <span class="sourceLineNo">208</span> HASHES_NOT_MATCHED,<a name="line.208"></a> |
| <span class="sourceLineNo">209</span> SOURCEMISSINGROWS,<a name="line.209"></a> |
| <span class="sourceLineNo">210</span> SOURCEMISSINGCELLS,<a name="line.210"></a> |
| <span class="sourceLineNo">211</span> TARGETMISSINGROWS,<a name="line.211"></a> |
| <span class="sourceLineNo">212</span> TARGETMISSINGCELLS,<a name="line.212"></a> |
| <span class="sourceLineNo">213</span> ROWSWITHDIFFS,<a name="line.213"></a> |
| <span class="sourceLineNo">214</span> DIFFERENTCELLVALUES,<a name="line.214"></a> |
| <span class="sourceLineNo">215</span> MATCHINGROWS,<a name="line.215"></a> |
| <span class="sourceLineNo">216</span> MATCHINGCELLS,<a name="line.216"></a> |
| <span class="sourceLineNo">217</span> EMPTY_BATCHES,<a name="line.217"></a> |
| <span class="sourceLineNo">218</span> RANGESMATCHED,<a name="line.218"></a> |
| <span class="sourceLineNo">219</span> RANGESNOTMATCHED<a name="line.219"></a> |
| <span class="sourceLineNo">220</span> }<a name="line.220"></a> |
| <span class="sourceLineNo">221</span><a name="line.221"></a> |
| <span class="sourceLineNo">222</span> @Override<a name="line.222"></a> |
| <span class="sourceLineNo">223</span> protected void setup(Context context) throws IOException {<a name="line.223"></a> |
| <span class="sourceLineNo">224</span><a name="line.224"></a> |
| <span class="sourceLineNo">225</span> Configuration conf = context.getConfiguration();<a name="line.225"></a> |
| <span class="sourceLineNo">226</span> sourceHashDir = new Path(conf.get(SOURCE_HASH_DIR_CONF_KEY));<a name="line.226"></a> |
| <span class="sourceLineNo">227</span> sourceConnection = openConnection(conf, SOURCE_ZK_CLUSTER_CONF_KEY, null);<a name="line.227"></a> |
| <span class="sourceLineNo">228</span> targetConnection =<a name="line.228"></a> |
| <span class="sourceLineNo">229</span> openConnection(conf, TARGET_ZK_CLUSTER_CONF_KEY, TableOutputFormat.OUTPUT_CONF_PREFIX);<a name="line.229"></a> |
| <span class="sourceLineNo">230</span> sourceTable = openTable(sourceConnection, conf, SOURCE_TABLE_CONF_KEY);<a name="line.230"></a> |
| <span class="sourceLineNo">231</span> targetTable = openTable(targetConnection, conf, TARGET_TABLE_CONF_KEY);<a name="line.231"></a> |
| <span class="sourceLineNo">232</span> dryRun = conf.getBoolean(DRY_RUN_CONF_KEY, false);<a name="line.232"></a> |
| <span class="sourceLineNo">233</span> doDeletes = conf.getBoolean(DO_DELETES_CONF_KEY, true);<a name="line.233"></a> |
| <span class="sourceLineNo">234</span> doPuts = conf.getBoolean(DO_PUTS_CONF_KEY, true);<a name="line.234"></a> |
| <span class="sourceLineNo">235</span> ignoreTimestamp = conf.getBoolean(IGNORE_TIMESTAMPS, false);<a name="line.235"></a> |
| <span class="sourceLineNo">236</span><a name="line.236"></a> |
| <span class="sourceLineNo">237</span> sourceTableHash = HashTable.TableHash.read(conf, sourceHashDir);<a name="line.237"></a> |
| <span class="sourceLineNo">238</span> LOG.info("Read source hash manifest: " + sourceTableHash);<a name="line.238"></a> |
| <span class="sourceLineNo">239</span> LOG.info("Read " + sourceTableHash.partitions.size() + " partition keys");<a name="line.239"></a> |
| <span class="sourceLineNo">240</span><a name="line.240"></a> |
| <span class="sourceLineNo">241</span> TableSplit split = (TableSplit) context.getInputSplit();<a name="line.241"></a> |
| <span class="sourceLineNo">242</span> ImmutableBytesWritable splitStartKey = new ImmutableBytesWritable(split.getStartRow());<a name="line.242"></a> |
| <span class="sourceLineNo">243</span><a name="line.243"></a> |
| <span class="sourceLineNo">244</span> sourceHashReader = sourceTableHash.newReader(conf, splitStartKey);<a name="line.244"></a> |
| <span class="sourceLineNo">245</span> findNextKeyHashPair();<a name="line.245"></a> |
| <span class="sourceLineNo">246</span><a name="line.246"></a> |
| <span class="sourceLineNo">247</span> // create a hasher, but don't start it right away<a name="line.247"></a> |
| <span class="sourceLineNo">248</span> // instead, find the first hash batch at or after the start row<a name="line.248"></a> |
| <span class="sourceLineNo">249</span> // and skip any rows that come before. they will be caught by the previous task<a name="line.249"></a> |
| <span class="sourceLineNo">250</span> targetHasher = new HashTable.ResultHasher();<a name="line.250"></a> |
| <span class="sourceLineNo">251</span> targetHasher.ignoreTimestamps = ignoreTimestamp;<a name="line.251"></a> |
| <span class="sourceLineNo">252</span> }<a name="line.252"></a> |
| <span class="sourceLineNo">253</span><a name="line.253"></a> |
| <span class="sourceLineNo">254</span> private static Connection openConnection(Configuration conf, String zkClusterConfKey,<a name="line.254"></a> |
| <span class="sourceLineNo">255</span> String configPrefix) throws IOException {<a name="line.255"></a> |
| <span class="sourceLineNo">256</span> String zkCluster = conf.get(zkClusterConfKey);<a name="line.256"></a> |
| <span class="sourceLineNo">257</span> Configuration clusterConf =<a name="line.257"></a> |
| <span class="sourceLineNo">258</span> HBaseConfiguration.createClusterConf(conf, zkCluster, configPrefix);<a name="line.258"></a> |
| <span class="sourceLineNo">259</span> return ConnectionFactory.createConnection(clusterConf);<a name="line.259"></a> |
| <span class="sourceLineNo">260</span> }<a name="line.260"></a> |
| <span class="sourceLineNo">261</span><a name="line.261"></a> |
| <span class="sourceLineNo">262</span> private static Table openTable(Connection connection, Configuration conf,<a name="line.262"></a> |
| <span class="sourceLineNo">263</span> String tableNameConfKey) throws IOException {<a name="line.263"></a> |
| <span class="sourceLineNo">264</span> return connection.getTable(TableName.valueOf(conf.get(tableNameConfKey)));<a name="line.264"></a> |
| <span class="sourceLineNo">265</span> }<a name="line.265"></a> |
| <span class="sourceLineNo">266</span><a name="line.266"></a> |
| <span class="sourceLineNo">267</span> /**<a name="line.267"></a> |
| <span class="sourceLineNo">268</span> * Attempt to read the next source key/hash pair. If there are no more, set nextSourceKey to<a name="line.268"></a> |
| <span class="sourceLineNo">269</span> * null<a name="line.269"></a> |
| <span class="sourceLineNo">270</span> */<a name="line.270"></a> |
| <span class="sourceLineNo">271</span> private void findNextKeyHashPair() throws IOException {<a name="line.271"></a> |
| <span class="sourceLineNo">272</span> boolean hasNext = sourceHashReader.next();<a name="line.272"></a> |
| <span class="sourceLineNo">273</span> if (hasNext) {<a name="line.273"></a> |
| <span class="sourceLineNo">274</span> nextSourceKey = sourceHashReader.getCurrentKey();<a name="line.274"></a> |
| <span class="sourceLineNo">275</span> } else {<a name="line.275"></a> |
| <span class="sourceLineNo">276</span> // no more keys - last hash goes to the end<a name="line.276"></a> |
| <span class="sourceLineNo">277</span> nextSourceKey = null;<a name="line.277"></a> |
| <span class="sourceLineNo">278</span> }<a name="line.278"></a> |
| <span class="sourceLineNo">279</span> }<a name="line.279"></a> |
| <span class="sourceLineNo">280</span><a name="line.280"></a> |
| <span class="sourceLineNo">281</span> @Override<a name="line.281"></a> |
| <span class="sourceLineNo">282</span> protected void map(ImmutableBytesWritable key, Result value, Context context)<a name="line.282"></a> |
| <span class="sourceLineNo">283</span> throws IOException, InterruptedException {<a name="line.283"></a> |
| <span class="sourceLineNo">284</span> try {<a name="line.284"></a> |
| <span class="sourceLineNo">285</span> // first, finish any hash batches that end before the scanned row<a name="line.285"></a> |
| <span class="sourceLineNo">286</span> while (nextSourceKey != null && key.compareTo(nextSourceKey) >= 0) {<a name="line.286"></a> |
| <span class="sourceLineNo">287</span> moveToNextBatch(context);<a name="line.287"></a> |
| <span class="sourceLineNo">288</span> }<a name="line.288"></a> |
| <span class="sourceLineNo">289</span><a name="line.289"></a> |
| <span class="sourceLineNo">290</span> // next, add the scanned row (as long as we've reached the first batch)<a name="line.290"></a> |
| <span class="sourceLineNo">291</span> if (targetHasher.isBatchStarted()) {<a name="line.291"></a> |
| <span class="sourceLineNo">292</span> targetHasher.hashResult(value);<a name="line.292"></a> |
| <span class="sourceLineNo">293</span> }<a name="line.293"></a> |
| <span class="sourceLineNo">294</span> } catch (Throwable t) {<a name="line.294"></a> |
| <span class="sourceLineNo">295</span> mapperException = t;<a name="line.295"></a> |
| <span class="sourceLineNo">296</span> Throwables.propagateIfInstanceOf(t, IOException.class);<a name="line.296"></a> |
| <span class="sourceLineNo">297</span> Throwables.propagateIfInstanceOf(t, InterruptedException.class);<a name="line.297"></a> |
| <span class="sourceLineNo">298</span> Throwables.propagate(t);<a name="line.298"></a> |
| <span class="sourceLineNo">299</span> }<a name="line.299"></a> |
| <span class="sourceLineNo">300</span> }<a name="line.300"></a> |
| <span class="sourceLineNo">301</span><a name="line.301"></a> |
| <span class="sourceLineNo">302</span> /**<a name="line.302"></a> |
| <span class="sourceLineNo">303</span> * If there is an open hash batch, complete it and sync if there are diffs. Start a new batch,<a name="line.303"></a> |
| <span class="sourceLineNo">304</span> * and seek to read the<a name="line.304"></a> |
| <span class="sourceLineNo">305</span> */<a name="line.305"></a> |
| <span class="sourceLineNo">306</span> private void moveToNextBatch(Context context) throws IOException, InterruptedException {<a name="line.306"></a> |
| <span class="sourceLineNo">307</span> if (targetHasher.isBatchStarted()) {<a name="line.307"></a> |
| <span class="sourceLineNo">308</span> finishBatchAndCompareHashes(context);<a name="line.308"></a> |
| <span class="sourceLineNo">309</span> }<a name="line.309"></a> |
| <span class="sourceLineNo">310</span> targetHasher.startBatch(nextSourceKey);<a name="line.310"></a> |
| <span class="sourceLineNo">311</span> currentSourceHash = sourceHashReader.getCurrentHash();<a name="line.311"></a> |
| <span class="sourceLineNo">312</span><a name="line.312"></a> |
| <span class="sourceLineNo">313</span> findNextKeyHashPair();<a name="line.313"></a> |
| <span class="sourceLineNo">314</span> }<a name="line.314"></a> |
| <span class="sourceLineNo">315</span><a name="line.315"></a> |
| <span class="sourceLineNo">316</span> /**<a name="line.316"></a> |
| <span class="sourceLineNo">317</span> * Finish the currently open hash batch. Compare the target hash to the given source hash. If<a name="line.317"></a> |
| <span class="sourceLineNo">318</span> * they do not match, then sync the covered key range.<a name="line.318"></a> |
| <span class="sourceLineNo">319</span> */<a name="line.319"></a> |
| <span class="sourceLineNo">320</span> private void finishBatchAndCompareHashes(Context context)<a name="line.320"></a> |
| <span class="sourceLineNo">321</span> throws IOException, InterruptedException {<a name="line.321"></a> |
| <span class="sourceLineNo">322</span> targetHasher.finishBatch();<a name="line.322"></a> |
| <span class="sourceLineNo">323</span> context.getCounter(Counter.BATCHES).increment(1);<a name="line.323"></a> |
| <span class="sourceLineNo">324</span> if (targetHasher.getBatchSize() == 0) {<a name="line.324"></a> |
| <span class="sourceLineNo">325</span> context.getCounter(Counter.EMPTY_BATCHES).increment(1);<a name="line.325"></a> |
| <span class="sourceLineNo">326</span> }<a name="line.326"></a> |
| <span class="sourceLineNo">327</span> ImmutableBytesWritable targetHash = targetHasher.getBatchHash();<a name="line.327"></a> |
| <span class="sourceLineNo">328</span> if (targetHash.equals(currentSourceHash)) {<a name="line.328"></a> |
| <span class="sourceLineNo">329</span> context.getCounter(Counter.HASHES_MATCHED).increment(1);<a name="line.329"></a> |
| <span class="sourceLineNo">330</span> } else {<a name="line.330"></a> |
| <span class="sourceLineNo">331</span> context.getCounter(Counter.HASHES_NOT_MATCHED).increment(1);<a name="line.331"></a> |
| <span class="sourceLineNo">332</span><a name="line.332"></a> |
| <span class="sourceLineNo">333</span> ImmutableBytesWritable stopRow = nextSourceKey == null<a name="line.333"></a> |
| <span class="sourceLineNo">334</span> ? new ImmutableBytesWritable(sourceTableHash.stopRow)<a name="line.334"></a> |
| <span class="sourceLineNo">335</span> : nextSourceKey;<a name="line.335"></a> |
| <span class="sourceLineNo">336</span><a name="line.336"></a> |
| <span class="sourceLineNo">337</span> if (LOG.isDebugEnabled()) {<a name="line.337"></a> |
| <span class="sourceLineNo">338</span> LOG.debug("Hash mismatch. Key range: " + toHex(targetHasher.getBatchStartKey()) + " to "<a name="line.338"></a> |
| <span class="sourceLineNo">339</span> + toHex(stopRow) + " sourceHash: " + toHex(currentSourceHash) + " targetHash: "<a name="line.339"></a> |
| <span class="sourceLineNo">340</span> + toHex(targetHash));<a name="line.340"></a> |
| <span class="sourceLineNo">341</span> }<a name="line.341"></a> |
| <span class="sourceLineNo">342</span><a name="line.342"></a> |
| <span class="sourceLineNo">343</span> syncRange(context, targetHasher.getBatchStartKey(), stopRow);<a name="line.343"></a> |
| <span class="sourceLineNo">344</span> }<a name="line.344"></a> |
| <span class="sourceLineNo">345</span> }<a name="line.345"></a> |
| <span class="sourceLineNo">346</span><a name="line.346"></a> |
| <span class="sourceLineNo">347</span> private static String toHex(ImmutableBytesWritable bytes) {<a name="line.347"></a> |
| <span class="sourceLineNo">348</span> return Bytes.toHex(bytes.get(), bytes.getOffset(), bytes.getLength());<a name="line.348"></a> |
| <span class="sourceLineNo">349</span> }<a name="line.349"></a> |
| <span class="sourceLineNo">350</span><a name="line.350"></a> |
| <span class="sourceLineNo">351</span> private static final CellScanner EMPTY_CELL_SCANNER =<a name="line.351"></a> |
| <span class="sourceLineNo">352</span> new CellScanner(Collections.<Result> emptyIterator());<a name="line.352"></a> |
| <span class="sourceLineNo">353</span><a name="line.353"></a> |
| <span class="sourceLineNo">354</span> /**<a name="line.354"></a> |
| <span class="sourceLineNo">355</span> * Rescan the given range directly from the source and target tables. Count and log differences,<a name="line.355"></a> |
| <span class="sourceLineNo">356</span> * and if this is not a dry run, output Puts and Deletes to make the target table match the<a name="line.356"></a> |
| <span class="sourceLineNo">357</span> * source table for this range<a name="line.357"></a> |
| <span class="sourceLineNo">358</span> */<a name="line.358"></a> |
| <span class="sourceLineNo">359</span> private void syncRange(Context context, ImmutableBytesWritable startRow,<a name="line.359"></a> |
| <span class="sourceLineNo">360</span> ImmutableBytesWritable stopRow) throws IOException, InterruptedException {<a name="line.360"></a> |
| <span class="sourceLineNo">361</span> Scan scan = sourceTableHash.initScan();<a name="line.361"></a> |
| <span class="sourceLineNo">362</span> scan.withStartRow(startRow.copyBytes());<a name="line.362"></a> |
| <span class="sourceLineNo">363</span> scan.withStopRow(stopRow.copyBytes());<a name="line.363"></a> |
| <span class="sourceLineNo">364</span><a name="line.364"></a> |
| <span class="sourceLineNo">365</span> ResultScanner sourceScanner = sourceTable.getScanner(scan);<a name="line.365"></a> |
| <span class="sourceLineNo">366</span> CellScanner sourceCells = new CellScanner(sourceScanner.iterator());<a name="line.366"></a> |
| <span class="sourceLineNo">367</span><a name="line.367"></a> |
| <span class="sourceLineNo">368</span> ResultScanner targetScanner = targetTable.getScanner(new Scan(scan));<a name="line.368"></a> |
| <span class="sourceLineNo">369</span> CellScanner targetCells = new CellScanner(targetScanner.iterator());<a name="line.369"></a> |
| <span class="sourceLineNo">370</span><a name="line.370"></a> |
| <span class="sourceLineNo">371</span> boolean rangeMatched = true;<a name="line.371"></a> |
| <span class="sourceLineNo">372</span> byte[] nextSourceRow = sourceCells.nextRow();<a name="line.372"></a> |
| <span class="sourceLineNo">373</span> byte[] nextTargetRow = targetCells.nextRow();<a name="line.373"></a> |
| <span class="sourceLineNo">374</span> while (nextSourceRow != null || nextTargetRow != null) {<a name="line.374"></a> |
| <span class="sourceLineNo">375</span> boolean rowMatched;<a name="line.375"></a> |
| <span class="sourceLineNo">376</span> int rowComparison = compareRowKeys(nextSourceRow, nextTargetRow);<a name="line.376"></a> |
| <span class="sourceLineNo">377</span> if (rowComparison < 0) {<a name="line.377"></a> |
| <span class="sourceLineNo">378</span> if (LOG.isDebugEnabled()) {<a name="line.378"></a> |
| <span class="sourceLineNo">379</span> LOG.debug("Target missing row: " + Bytes.toString(nextSourceRow));<a name="line.379"></a> |
| <span class="sourceLineNo">380</span> }<a name="line.380"></a> |
| <span class="sourceLineNo">381</span> context.getCounter(Counter.TARGETMISSINGROWS).increment(1);<a name="line.381"></a> |
| <span class="sourceLineNo">382</span><a name="line.382"></a> |
| <span class="sourceLineNo">383</span> rowMatched = syncRowCells(context, nextSourceRow, sourceCells, EMPTY_CELL_SCANNER);<a name="line.383"></a> |
| <span class="sourceLineNo">384</span> nextSourceRow = sourceCells.nextRow(); // advance only source to next row<a name="line.384"></a> |
| <span class="sourceLineNo">385</span> } else if (rowComparison > 0) {<a name="line.385"></a> |
| <span class="sourceLineNo">386</span> if (LOG.isDebugEnabled()) {<a name="line.386"></a> |
| <span class="sourceLineNo">387</span> LOG.debug("Source missing row: " + Bytes.toString(nextTargetRow));<a name="line.387"></a> |
| <span class="sourceLineNo">388</span> }<a name="line.388"></a> |
| <span class="sourceLineNo">389</span> context.getCounter(Counter.SOURCEMISSINGROWS).increment(1);<a name="line.389"></a> |
| <span class="sourceLineNo">390</span><a name="line.390"></a> |
| <span class="sourceLineNo">391</span> rowMatched = syncRowCells(context, nextTargetRow, EMPTY_CELL_SCANNER, targetCells);<a name="line.391"></a> |
| <span class="sourceLineNo">392</span> nextTargetRow = targetCells.nextRow(); // advance only target to next row<a name="line.392"></a> |
| <span class="sourceLineNo">393</span> } else {<a name="line.393"></a> |
| <span class="sourceLineNo">394</span> // current row is the same on both sides, compare cell by cell<a name="line.394"></a> |
| <span class="sourceLineNo">395</span> rowMatched = syncRowCells(context, nextSourceRow, sourceCells, targetCells);<a name="line.395"></a> |
| <span class="sourceLineNo">396</span> nextSourceRow = sourceCells.nextRow();<a name="line.396"></a> |
| <span class="sourceLineNo">397</span> nextTargetRow = targetCells.nextRow();<a name="line.397"></a> |
| <span class="sourceLineNo">398</span> }<a name="line.398"></a> |
| <span class="sourceLineNo">399</span><a name="line.399"></a> |
| <span class="sourceLineNo">400</span> if (!rowMatched) {<a name="line.400"></a> |
| <span class="sourceLineNo">401</span> rangeMatched = false;<a name="line.401"></a> |
| <span class="sourceLineNo">402</span> }<a name="line.402"></a> |
| <span class="sourceLineNo">403</span> }<a name="line.403"></a> |
| <span class="sourceLineNo">404</span><a name="line.404"></a> |
| <span class="sourceLineNo">405</span> sourceScanner.close();<a name="line.405"></a> |
| <span class="sourceLineNo">406</span> targetScanner.close();<a name="line.406"></a> |
| <span class="sourceLineNo">407</span><a name="line.407"></a> |
| <span class="sourceLineNo">408</span> context.getCounter(rangeMatched ? Counter.RANGESMATCHED : Counter.RANGESNOTMATCHED)<a name="line.408"></a> |
| <span class="sourceLineNo">409</span> .increment(1);<a name="line.409"></a> |
| <span class="sourceLineNo">410</span> }<a name="line.410"></a> |
| <span class="sourceLineNo">411</span><a name="line.411"></a> |
| <span class="sourceLineNo">412</span> private static class CellScanner {<a name="line.412"></a> |
| <span class="sourceLineNo">413</span> private final Iterator<Result> results;<a name="line.413"></a> |
| <span class="sourceLineNo">414</span><a name="line.414"></a> |
| <span class="sourceLineNo">415</span> private byte[] currentRow;<a name="line.415"></a> |
| <span class="sourceLineNo">416</span> private Result currentRowResult;<a name="line.416"></a> |
| <span class="sourceLineNo">417</span> private int nextCellInRow;<a name="line.417"></a> |
| <span class="sourceLineNo">418</span><a name="line.418"></a> |
| <span class="sourceLineNo">419</span> private Result nextRowResult;<a name="line.419"></a> |
| <span class="sourceLineNo">420</span><a name="line.420"></a> |
| <span class="sourceLineNo">421</span> public CellScanner(Iterator<Result> results) {<a name="line.421"></a> |
| <span class="sourceLineNo">422</span> this.results = results;<a name="line.422"></a> |
| <span class="sourceLineNo">423</span> }<a name="line.423"></a> |
| <span class="sourceLineNo">424</span><a name="line.424"></a> |
| <span class="sourceLineNo">425</span> /**<a name="line.425"></a> |
| <span class="sourceLineNo">426</span> * Advance to the next row and return its row key. Returns null iff there are no more rows.<a name="line.426"></a> |
| <span class="sourceLineNo">427</span> */<a name="line.427"></a> |
| <span class="sourceLineNo">428</span> public byte[] nextRow() {<a name="line.428"></a> |
| <span class="sourceLineNo">429</span> if (nextRowResult == null) {<a name="line.429"></a> |
| <span class="sourceLineNo">430</span> // no cached row - check scanner for more<a name="line.430"></a> |
| <span class="sourceLineNo">431</span> while (results.hasNext()) {<a name="line.431"></a> |
| <span class="sourceLineNo">432</span> nextRowResult = results.next();<a name="line.432"></a> |
| <span class="sourceLineNo">433</span> Cell nextCell = nextRowResult.rawCells()[0];<a name="line.433"></a> |
| <span class="sourceLineNo">434</span> if (<a name="line.434"></a> |
| <span class="sourceLineNo">435</span> currentRow == null || !Bytes.equals(currentRow, 0, currentRow.length,<a name="line.435"></a> |
| <span class="sourceLineNo">436</span> nextCell.getRowArray(), nextCell.getRowOffset(), nextCell.getRowLength())<a name="line.436"></a> |
| <span class="sourceLineNo">437</span> ) {<a name="line.437"></a> |
| <span class="sourceLineNo">438</span> // found next row<a name="line.438"></a> |
| <span class="sourceLineNo">439</span> break;<a name="line.439"></a> |
| <span class="sourceLineNo">440</span> } else {<a name="line.440"></a> |
| <span class="sourceLineNo">441</span> // found another result from current row, keep scanning<a name="line.441"></a> |
| <span class="sourceLineNo">442</span> nextRowResult = null;<a name="line.442"></a> |
| <span class="sourceLineNo">443</span> }<a name="line.443"></a> |
| <span class="sourceLineNo">444</span> }<a name="line.444"></a> |
| <span class="sourceLineNo">445</span><a name="line.445"></a> |
| <span class="sourceLineNo">446</span> if (nextRowResult == null) {<a name="line.446"></a> |
| <span class="sourceLineNo">447</span> // end of data, no more rows<a name="line.447"></a> |
| <span class="sourceLineNo">448</span> currentRowResult = null;<a name="line.448"></a> |
| <span class="sourceLineNo">449</span> currentRow = null;<a name="line.449"></a> |
| <span class="sourceLineNo">450</span> return null;<a name="line.450"></a> |
| <span class="sourceLineNo">451</span> }<a name="line.451"></a> |
| <span class="sourceLineNo">452</span> }<a name="line.452"></a> |
| <span class="sourceLineNo">453</span><a name="line.453"></a> |
| <span class="sourceLineNo">454</span> // advance to cached result for next row<a name="line.454"></a> |
| <span class="sourceLineNo">455</span> currentRowResult = nextRowResult;<a name="line.455"></a> |
| <span class="sourceLineNo">456</span> nextCellInRow = 0;<a name="line.456"></a> |
| <span class="sourceLineNo">457</span> currentRow = currentRowResult.getRow();<a name="line.457"></a> |
| <span class="sourceLineNo">458</span> nextRowResult = null;<a name="line.458"></a> |
| <span class="sourceLineNo">459</span> return currentRow;<a name="line.459"></a> |
| <span class="sourceLineNo">460</span> }<a name="line.460"></a> |
| <span class="sourceLineNo">461</span><a name="line.461"></a> |
| <span class="sourceLineNo">462</span> /**<a name="line.462"></a> |
| <span class="sourceLineNo">463</span> * Returns the next Cell in the current row or null iff none remain.<a name="line.463"></a> |
| <span class="sourceLineNo">464</span> */<a name="line.464"></a> |
| <span class="sourceLineNo">465</span> public Cell nextCellInRow() {<a name="line.465"></a> |
| <span class="sourceLineNo">466</span> if (currentRowResult == null) {<a name="line.466"></a> |
| <span class="sourceLineNo">467</span> // nothing left in current row<a name="line.467"></a> |
| <span class="sourceLineNo">468</span> return null;<a name="line.468"></a> |
| <span class="sourceLineNo">469</span> }<a name="line.469"></a> |
| <span class="sourceLineNo">470</span><a name="line.470"></a> |
| <span class="sourceLineNo">471</span> Cell nextCell = currentRowResult.rawCells()[nextCellInRow];<a name="line.471"></a> |
| <span class="sourceLineNo">472</span> nextCellInRow++;<a name="line.472"></a> |
| <span class="sourceLineNo">473</span> if (nextCellInRow == currentRowResult.size()) {<a name="line.473"></a> |
| <span class="sourceLineNo">474</span> if (results.hasNext()) {<a name="line.474"></a> |
| <span class="sourceLineNo">475</span> Result result = results.next();<a name="line.475"></a> |
| <span class="sourceLineNo">476</span> Cell cell = result.rawCells()[0];<a name="line.476"></a> |
| <span class="sourceLineNo">477</span> if (<a name="line.477"></a> |
| <span class="sourceLineNo">478</span> Bytes.equals(currentRow, 0, currentRow.length, cell.getRowArray(),<a name="line.478"></a> |
| <span class="sourceLineNo">479</span> cell.getRowOffset(), cell.getRowLength())<a name="line.479"></a> |
| <span class="sourceLineNo">480</span> ) {<a name="line.480"></a> |
| <span class="sourceLineNo">481</span> // result is part of current row<a name="line.481"></a> |
| <span class="sourceLineNo">482</span> currentRowResult = result;<a name="line.482"></a> |
| <span class="sourceLineNo">483</span> nextCellInRow = 0;<a name="line.483"></a> |
| <span class="sourceLineNo">484</span> } else {<a name="line.484"></a> |
| <span class="sourceLineNo">485</span> // result is part of next row, cache it<a name="line.485"></a> |
| <span class="sourceLineNo">486</span> nextRowResult = result;<a name="line.486"></a> |
| <span class="sourceLineNo">487</span> // current row is complete<a name="line.487"></a> |
| <span class="sourceLineNo">488</span> currentRowResult = null;<a name="line.488"></a> |
| <span class="sourceLineNo">489</span> }<a name="line.489"></a> |
| <span class="sourceLineNo">490</span> } else {<a name="line.490"></a> |
| <span class="sourceLineNo">491</span> // end of data<a name="line.491"></a> |
| <span class="sourceLineNo">492</span> currentRowResult = null;<a name="line.492"></a> |
| <span class="sourceLineNo">493</span> }<a name="line.493"></a> |
| <span class="sourceLineNo">494</span> }<a name="line.494"></a> |
| <span class="sourceLineNo">495</span> return nextCell;<a name="line.495"></a> |
| <span class="sourceLineNo">496</span> }<a name="line.496"></a> |
| <span class="sourceLineNo">497</span> }<a name="line.497"></a> |
| <span class="sourceLineNo">498</span><a name="line.498"></a> |
| <span class="sourceLineNo">499</span> private Cell checkAndResetTimestamp(Cell sourceCell) {<a name="line.499"></a> |
| <span class="sourceLineNo">500</span> if (ignoreTimestamp) {<a name="line.500"></a> |
| <span class="sourceLineNo">501</span> sourceCell =<a name="line.501"></a> |
| <span class="sourceLineNo">502</span> CellBuilderFactory.create(CellBuilderType.SHALLOW_COPY).setType(sourceCell.getType())<a name="line.502"></a> |
| <span class="sourceLineNo">503</span> .setRow(sourceCell.getRowArray(), sourceCell.getRowOffset(), sourceCell.getRowLength())<a name="line.503"></a> |
| <span class="sourceLineNo">504</span> .setFamily(sourceCell.getFamilyArray(), sourceCell.getFamilyOffset(),<a name="line.504"></a> |
| <span class="sourceLineNo">505</span> sourceCell.getFamilyLength())<a name="line.505"></a> |
| <span class="sourceLineNo">506</span> .setQualifier(sourceCell.getQualifierArray(), sourceCell.getQualifierOffset(),<a name="line.506"></a> |
| <span class="sourceLineNo">507</span> sourceCell.getQualifierLength())<a name="line.507"></a> |
| <span class="sourceLineNo">508</span> .setTimestamp(EnvironmentEdgeManager.currentTime()).setValue(sourceCell.getValueArray(),<a name="line.508"></a> |
| <span class="sourceLineNo">509</span> sourceCell.getValueOffset(), sourceCell.getValueLength())<a name="line.509"></a> |
| <span class="sourceLineNo">510</span> .build();<a name="line.510"></a> |
| <span class="sourceLineNo">511</span> }<a name="line.511"></a> |
| <span class="sourceLineNo">512</span> return sourceCell;<a name="line.512"></a> |
| <span class="sourceLineNo">513</span> }<a name="line.513"></a> |
| <span class="sourceLineNo">514</span><a name="line.514"></a> |
| <span class="sourceLineNo">515</span> /**<a name="line.515"></a> |
| <span class="sourceLineNo">516</span> * Compare the cells for the given row from the source and target tables. Count and log any<a name="line.516"></a> |
| <span class="sourceLineNo">517</span> * differences. If not a dry run, output a Put and/or Delete needed to sync the target table to<a name="line.517"></a> |
| <span class="sourceLineNo">518</span> * match the source table.<a name="line.518"></a> |
| <span class="sourceLineNo">519</span> */<a name="line.519"></a> |
| <span class="sourceLineNo">520</span> private boolean syncRowCells(Context context, byte[] rowKey, CellScanner sourceCells,<a name="line.520"></a> |
| <span class="sourceLineNo">521</span> CellScanner targetCells) throws IOException, InterruptedException {<a name="line.521"></a> |
| <span class="sourceLineNo">522</span> Put put = null;<a name="line.522"></a> |
| <span class="sourceLineNo">523</span> Delete delete = null;<a name="line.523"></a> |
| <span class="sourceLineNo">524</span> long matchingCells = 0;<a name="line.524"></a> |
| <span class="sourceLineNo">525</span> boolean matchingRow = true;<a name="line.525"></a> |
| <span class="sourceLineNo">526</span> Cell sourceCell = sourceCells.nextCellInRow();<a name="line.526"></a> |
| <span class="sourceLineNo">527</span> Cell targetCell = targetCells.nextCellInRow();<a name="line.527"></a> |
| <span class="sourceLineNo">528</span> while (sourceCell != null || targetCell != null) {<a name="line.528"></a> |
| <span class="sourceLineNo">529</span><a name="line.529"></a> |
| <span class="sourceLineNo">530</span> int cellKeyComparison = compareCellKeysWithinRow(sourceCell, targetCell);<a name="line.530"></a> |
| <span class="sourceLineNo">531</span> if (cellKeyComparison < 0) {<a name="line.531"></a> |
| <span class="sourceLineNo">532</span> if (LOG.isDebugEnabled()) {<a name="line.532"></a> |
| <span class="sourceLineNo">533</span> LOG.debug("Target missing cell: " + sourceCell);<a name="line.533"></a> |
| <span class="sourceLineNo">534</span> }<a name="line.534"></a> |
| <span class="sourceLineNo">535</span> context.getCounter(Counter.TARGETMISSINGCELLS).increment(1);<a name="line.535"></a> |
| <span class="sourceLineNo">536</span> matchingRow = false;<a name="line.536"></a> |
| <span class="sourceLineNo">537</span><a name="line.537"></a> |
| <span class="sourceLineNo">538</span> if (!dryRun && doPuts) {<a name="line.538"></a> |
| <span class="sourceLineNo">539</span> if (put == null) {<a name="line.539"></a> |
| <span class="sourceLineNo">540</span> put = new Put(rowKey);<a name="line.540"></a> |
| <span class="sourceLineNo">541</span> }<a name="line.541"></a> |
| <span class="sourceLineNo">542</span> sourceCell = checkAndResetTimestamp(sourceCell);<a name="line.542"></a> |
| <span class="sourceLineNo">543</span> put.add(sourceCell);<a name="line.543"></a> |
| <span class="sourceLineNo">544</span> }<a name="line.544"></a> |
| <span class="sourceLineNo">545</span><a name="line.545"></a> |
| <span class="sourceLineNo">546</span> sourceCell = sourceCells.nextCellInRow();<a name="line.546"></a> |
| <span class="sourceLineNo">547</span> } else if (cellKeyComparison > 0) {<a name="line.547"></a> |
| <span class="sourceLineNo">548</span> if (LOG.isDebugEnabled()) {<a name="line.548"></a> |
| <span class="sourceLineNo">549</span> LOG.debug("Source missing cell: " + targetCell);<a name="line.549"></a> |
| <span class="sourceLineNo">550</span> }<a name="line.550"></a> |
| <span class="sourceLineNo">551</span> context.getCounter(Counter.SOURCEMISSINGCELLS).increment(1);<a name="line.551"></a> |
| <span class="sourceLineNo">552</span> matchingRow = false;<a name="line.552"></a> |
| <span class="sourceLineNo">553</span><a name="line.553"></a> |
| <span class="sourceLineNo">554</span> if (!dryRun && doDeletes) {<a name="line.554"></a> |
| <span class="sourceLineNo">555</span> if (delete == null) {<a name="line.555"></a> |
| <span class="sourceLineNo">556</span> delete = new Delete(rowKey);<a name="line.556"></a> |
| <span class="sourceLineNo">557</span> }<a name="line.557"></a> |
| <span class="sourceLineNo">558</span> // add a tombstone to exactly match the target cell that is missing on the source<a name="line.558"></a> |
| <span class="sourceLineNo">559</span> delete.addColumn(CellUtil.cloneFamily(targetCell), CellUtil.cloneQualifier(targetCell),<a name="line.559"></a> |
| <span class="sourceLineNo">560</span> targetCell.getTimestamp());<a name="line.560"></a> |
| <span class="sourceLineNo">561</span> }<a name="line.561"></a> |
| <span class="sourceLineNo">562</span><a name="line.562"></a> |
| <span class="sourceLineNo">563</span> targetCell = targetCells.nextCellInRow();<a name="line.563"></a> |
| <span class="sourceLineNo">564</span> } else {<a name="line.564"></a> |
| <span class="sourceLineNo">565</span> // the cell keys are equal, now check values<a name="line.565"></a> |
| <span class="sourceLineNo">566</span> if (CellUtil.matchingValue(sourceCell, targetCell)) {<a name="line.566"></a> |
| <span class="sourceLineNo">567</span> matchingCells++;<a name="line.567"></a> |
| <span class="sourceLineNo">568</span> } else {<a name="line.568"></a> |
| <span class="sourceLineNo">569</span> if (LOG.isDebugEnabled()) {<a name="line.569"></a> |
| <span class="sourceLineNo">570</span> LOG.debug("Different values: ");<a name="line.570"></a> |
| <span class="sourceLineNo">571</span> LOG.debug(" source cell: " + sourceCell + " value: "<a name="line.571"></a> |
| <span class="sourceLineNo">572</span> + Bytes.toString(sourceCell.getValueArray(), sourceCell.getValueOffset(),<a name="line.572"></a> |
| <span class="sourceLineNo">573</span> sourceCell.getValueLength()));<a name="line.573"></a> |
| <span class="sourceLineNo">574</span> LOG.debug(" target cell: " + targetCell + " value: "<a name="line.574"></a> |
| <span class="sourceLineNo">575</span> + Bytes.toString(targetCell.getValueArray(), targetCell.getValueOffset(),<a name="line.575"></a> |
| <span class="sourceLineNo">576</span> targetCell.getValueLength()));<a name="line.576"></a> |
| <span class="sourceLineNo">577</span> }<a name="line.577"></a> |
| <span class="sourceLineNo">578</span> context.getCounter(Counter.DIFFERENTCELLVALUES).increment(1);<a name="line.578"></a> |
| <span class="sourceLineNo">579</span> matchingRow = false;<a name="line.579"></a> |
| <span class="sourceLineNo">580</span><a name="line.580"></a> |
| <span class="sourceLineNo">581</span> if (!dryRun && doPuts) {<a name="line.581"></a> |
| <span class="sourceLineNo">582</span> // overwrite target cell<a name="line.582"></a> |
| <span class="sourceLineNo">583</span> if (put == null) {<a name="line.583"></a> |
| <span class="sourceLineNo">584</span> put = new Put(rowKey);<a name="line.584"></a> |
| <span class="sourceLineNo">585</span> }<a name="line.585"></a> |
| <span class="sourceLineNo">586</span> sourceCell = checkAndResetTimestamp(sourceCell);<a name="line.586"></a> |
| <span class="sourceLineNo">587</span> put.add(sourceCell);<a name="line.587"></a> |
| <span class="sourceLineNo">588</span> }<a name="line.588"></a> |
| <span class="sourceLineNo">589</span> }<a name="line.589"></a> |
| <span class="sourceLineNo">590</span> sourceCell = sourceCells.nextCellInRow();<a name="line.590"></a> |
| <span class="sourceLineNo">591</span> targetCell = targetCells.nextCellInRow();<a name="line.591"></a> |
| <span class="sourceLineNo">592</span> }<a name="line.592"></a> |
| <span class="sourceLineNo">593</span><a name="line.593"></a> |
| <span class="sourceLineNo">594</span> if (!dryRun && sourceTableHash.scanBatch > 0) {<a name="line.594"></a> |
| <span class="sourceLineNo">595</span> if (put != null && put.size() >= sourceTableHash.scanBatch) {<a name="line.595"></a> |
| <span class="sourceLineNo">596</span> context.write(new ImmutableBytesWritable(rowKey), put);<a name="line.596"></a> |
| <span class="sourceLineNo">597</span> put = null;<a name="line.597"></a> |
| <span class="sourceLineNo">598</span> }<a name="line.598"></a> |
| <span class="sourceLineNo">599</span> if (delete != null && delete.size() >= sourceTableHash.scanBatch) {<a name="line.599"></a> |
| <span class="sourceLineNo">600</span> context.write(new ImmutableBytesWritable(rowKey), delete);<a name="line.600"></a> |
| <span class="sourceLineNo">601</span> delete = null;<a name="line.601"></a> |
| <span class="sourceLineNo">602</span> }<a name="line.602"></a> |
| <span class="sourceLineNo">603</span> }<a name="line.603"></a> |
| <span class="sourceLineNo">604</span> }<a name="line.604"></a> |
| <span class="sourceLineNo">605</span><a name="line.605"></a> |
| <span class="sourceLineNo">606</span> if (!dryRun) {<a name="line.606"></a> |
| <span class="sourceLineNo">607</span> if (put != null) {<a name="line.607"></a> |
| <span class="sourceLineNo">608</span> context.write(new ImmutableBytesWritable(rowKey), put);<a name="line.608"></a> |
| <span class="sourceLineNo">609</span> }<a name="line.609"></a> |
| <span class="sourceLineNo">610</span> if (delete != null) {<a name="line.610"></a> |
| <span class="sourceLineNo">611</span> context.write(new ImmutableBytesWritable(rowKey), delete);<a name="line.611"></a> |
| <span class="sourceLineNo">612</span> }<a name="line.612"></a> |
| <span class="sourceLineNo">613</span> }<a name="line.613"></a> |
| <span class="sourceLineNo">614</span><a name="line.614"></a> |
| <span class="sourceLineNo">615</span> if (matchingCells > 0) {<a name="line.615"></a> |
| <span class="sourceLineNo">616</span> context.getCounter(Counter.MATCHINGCELLS).increment(matchingCells);<a name="line.616"></a> |
| <span class="sourceLineNo">617</span> }<a name="line.617"></a> |
| <span class="sourceLineNo">618</span> if (matchingRow) {<a name="line.618"></a> |
| <span class="sourceLineNo">619</span> context.getCounter(Counter.MATCHINGROWS).increment(1);<a name="line.619"></a> |
| <span class="sourceLineNo">620</span> return true;<a name="line.620"></a> |
| <span class="sourceLineNo">621</span> } else {<a name="line.621"></a> |
| <span class="sourceLineNo">622</span> context.getCounter(Counter.ROWSWITHDIFFS).increment(1);<a name="line.622"></a> |
| <span class="sourceLineNo">623</span> return false;<a name="line.623"></a> |
| <span class="sourceLineNo">624</span> }<a name="line.624"></a> |
| <span class="sourceLineNo">625</span> }<a name="line.625"></a> |
| <span class="sourceLineNo">626</span><a name="line.626"></a> |
| <span class="sourceLineNo">627</span> /**<a name="line.627"></a> |
| <span class="sourceLineNo">628</span> * Compare row keys of the given Result objects. Nulls are after non-nulls<a name="line.628"></a> |
| <span class="sourceLineNo">629</span> */<a name="line.629"></a> |
| <span class="sourceLineNo">630</span> private static int compareRowKeys(byte[] r1, byte[] r2) {<a name="line.630"></a> |
| <span class="sourceLineNo">631</span> if (r1 == null) {<a name="line.631"></a> |
| <span class="sourceLineNo">632</span> return 1; // source missing row<a name="line.632"></a> |
| <span class="sourceLineNo">633</span> } else if (r2 == null) {<a name="line.633"></a> |
| <span class="sourceLineNo">634</span> return -1; // target missing row<a name="line.634"></a> |
| <span class="sourceLineNo">635</span> } else {<a name="line.635"></a> |
| <span class="sourceLineNo">636</span> // Sync on no META tables only. We can directly do what CellComparator is doing inside.<a name="line.636"></a> |
| <span class="sourceLineNo">637</span> // Never the call going to MetaCellComparator.<a name="line.637"></a> |
| <span class="sourceLineNo">638</span> return Bytes.compareTo(r1, 0, r1.length, r2, 0, r2.length);<a name="line.638"></a> |
| <span class="sourceLineNo">639</span> }<a name="line.639"></a> |
| <span class="sourceLineNo">640</span> }<a name="line.640"></a> |
| <span class="sourceLineNo">641</span><a name="line.641"></a> |
| <span class="sourceLineNo">642</span> /**<a name="line.642"></a> |
| <span class="sourceLineNo">643</span> * Compare families, qualifiers, and timestamps of the given Cells. They are assumed to be of<a name="line.643"></a> |
| <span class="sourceLineNo">644</span> * the same row. Nulls are after non-nulls.<a name="line.644"></a> |
| <span class="sourceLineNo">645</span> */<a name="line.645"></a> |
| <span class="sourceLineNo">646</span> private int compareCellKeysWithinRow(Cell c1, Cell c2) {<a name="line.646"></a> |
| <span class="sourceLineNo">647</span> if (c1 == null) {<a name="line.647"></a> |
| <span class="sourceLineNo">648</span> return 1; // source missing cell<a name="line.648"></a> |
| <span class="sourceLineNo">649</span> }<a name="line.649"></a> |
| <span class="sourceLineNo">650</span> if (c2 == null) {<a name="line.650"></a> |
| <span class="sourceLineNo">651</span> return -1; // target missing cell<a name="line.651"></a> |
| <span class="sourceLineNo">652</span> }<a name="line.652"></a> |
| <span class="sourceLineNo">653</span><a name="line.653"></a> |
| <span class="sourceLineNo">654</span> int result = CellComparator.getInstance().compareFamilies(c1, c2);<a name="line.654"></a> |
| <span class="sourceLineNo">655</span> if (result != 0) {<a name="line.655"></a> |
| <span class="sourceLineNo">656</span> return result;<a name="line.656"></a> |
| <span class="sourceLineNo">657</span> }<a name="line.657"></a> |
| <span class="sourceLineNo">658</span><a name="line.658"></a> |
| <span class="sourceLineNo">659</span> result = CellComparator.getInstance().compareQualifiers(c1, c2);<a name="line.659"></a> |
| <span class="sourceLineNo">660</span> if (result != 0) {<a name="line.660"></a> |
| <span class="sourceLineNo">661</span> return result;<a name="line.661"></a> |
| <span class="sourceLineNo">662</span> }<a name="line.662"></a> |
| <span class="sourceLineNo">663</span><a name="line.663"></a> |
| <span class="sourceLineNo">664</span> if (this.ignoreTimestamp) {<a name="line.664"></a> |
| <span class="sourceLineNo">665</span> return 0;<a name="line.665"></a> |
| <span class="sourceLineNo">666</span> } else {<a name="line.666"></a> |
| <span class="sourceLineNo">667</span> // note timestamp comparison is inverted - more recent cells first<a name="line.667"></a> |
| <span class="sourceLineNo">668</span> return CellComparator.getInstance().compareTimestamps(c1, c2);<a name="line.668"></a> |
| <span class="sourceLineNo">669</span> }<a name="line.669"></a> |
| <span class="sourceLineNo">670</span> }<a name="line.670"></a> |
| <span class="sourceLineNo">671</span><a name="line.671"></a> |
| <span class="sourceLineNo">672</span> @Override<a name="line.672"></a> |
| <span class="sourceLineNo">673</span> protected void cleanup(Context context) throws IOException, InterruptedException {<a name="line.673"></a> |
| <span class="sourceLineNo">674</span> if (mapperException == null) {<a name="line.674"></a> |
| <span class="sourceLineNo">675</span> try {<a name="line.675"></a> |
| <span class="sourceLineNo">676</span> finishRemainingHashRanges(context);<a name="line.676"></a> |
| <span class="sourceLineNo">677</span> } catch (Throwable t) {<a name="line.677"></a> |
| <span class="sourceLineNo">678</span> mapperException = t;<a name="line.678"></a> |
| <span class="sourceLineNo">679</span> }<a name="line.679"></a> |
| <span class="sourceLineNo">680</span> }<a name="line.680"></a> |
| <span class="sourceLineNo">681</span><a name="line.681"></a> |
| <span class="sourceLineNo">682</span> try {<a name="line.682"></a> |
| <span class="sourceLineNo">683</span> sourceTable.close();<a name="line.683"></a> |
| <span class="sourceLineNo">684</span> targetTable.close();<a name="line.684"></a> |
| <span class="sourceLineNo">685</span> sourceConnection.close();<a name="line.685"></a> |
| <span class="sourceLineNo">686</span> targetConnection.close();<a name="line.686"></a> |
| <span class="sourceLineNo">687</span> } catch (Throwable t) {<a name="line.687"></a> |
| <span class="sourceLineNo">688</span> if (mapperException == null) {<a name="line.688"></a> |
| <span class="sourceLineNo">689</span> mapperException = t;<a name="line.689"></a> |
| <span class="sourceLineNo">690</span> } else {<a name="line.690"></a> |
| <span class="sourceLineNo">691</span> LOG.error("Suppressing exception from closing tables", t);<a name="line.691"></a> |
| <span class="sourceLineNo">692</span> }<a name="line.692"></a> |
| <span class="sourceLineNo">693</span> }<a name="line.693"></a> |
| <span class="sourceLineNo">694</span><a name="line.694"></a> |
| <span class="sourceLineNo">695</span> // propagate first exception<a name="line.695"></a> |
| <span class="sourceLineNo">696</span> if (mapperException != null) {<a name="line.696"></a> |
| <span class="sourceLineNo">697</span> Throwables.propagateIfInstanceOf(mapperException, IOException.class);<a name="line.697"></a> |
| <span class="sourceLineNo">698</span> Throwables.propagateIfInstanceOf(mapperException, InterruptedException.class);<a name="line.698"></a> |
| <span class="sourceLineNo">699</span> Throwables.propagate(mapperException);<a name="line.699"></a> |
| <span class="sourceLineNo">700</span> }<a name="line.700"></a> |
| <span class="sourceLineNo">701</span> }<a name="line.701"></a> |
| <span class="sourceLineNo">702</span><a name="line.702"></a> |
| <span class="sourceLineNo">703</span> private void finishRemainingHashRanges(Context context)<a name="line.703"></a> |
| <span class="sourceLineNo">704</span> throws IOException, InterruptedException {<a name="line.704"></a> |
| <span class="sourceLineNo">705</span> TableSplit split = (TableSplit) context.getInputSplit();<a name="line.705"></a> |
| <span class="sourceLineNo">706</span> byte[] splitEndRow = split.getEndRow();<a name="line.706"></a> |
| <span class="sourceLineNo">707</span> boolean reachedEndOfTable = HashTable.isTableEndRow(splitEndRow);<a name="line.707"></a> |
| <span class="sourceLineNo">708</span><a name="line.708"></a> |
| <span class="sourceLineNo">709</span> // if there are more hash batches that begin before the end of this split move to them<a name="line.709"></a> |
| <span class="sourceLineNo">710</span> while (<a name="line.710"></a> |
| <span class="sourceLineNo">711</span> nextSourceKey != null && (nextSourceKey.compareTo(splitEndRow) < 0 || reachedEndOfTable)<a name="line.711"></a> |
| <span class="sourceLineNo">712</span> ) {<a name="line.712"></a> |
| <span class="sourceLineNo">713</span> moveToNextBatch(context);<a name="line.713"></a> |
| <span class="sourceLineNo">714</span> }<a name="line.714"></a> |
| <span class="sourceLineNo">715</span><a name="line.715"></a> |
| <span class="sourceLineNo">716</span> if (targetHasher.isBatchStarted()) {<a name="line.716"></a> |
| <span class="sourceLineNo">717</span> // need to complete the final open hash batch<a name="line.717"></a> |
| <span class="sourceLineNo">718</span><a name="line.718"></a> |
| <span class="sourceLineNo">719</span> if (<a name="line.719"></a> |
| <span class="sourceLineNo">720</span> (nextSourceKey != null && nextSourceKey.compareTo(splitEndRow) > 0)<a name="line.720"></a> |
| <span class="sourceLineNo">721</span> || (nextSourceKey == null && !Bytes.equals(splitEndRow, sourceTableHash.stopRow))<a name="line.721"></a> |
| <span class="sourceLineNo">722</span> ) {<a name="line.722"></a> |
| <span class="sourceLineNo">723</span> // the open hash range continues past the end of this region<a name="line.723"></a> |
| <span class="sourceLineNo">724</span> // add a scan to complete the current hash range<a name="line.724"></a> |
| <span class="sourceLineNo">725</span> Scan scan = sourceTableHash.initScan();<a name="line.725"></a> |
| <span class="sourceLineNo">726</span> scan.withStartRow(splitEndRow);<a name="line.726"></a> |
| <span class="sourceLineNo">727</span> if (nextSourceKey == null) {<a name="line.727"></a> |
| <span class="sourceLineNo">728</span> scan.withStopRow(sourceTableHash.stopRow);<a name="line.728"></a> |
| <span class="sourceLineNo">729</span> } else {<a name="line.729"></a> |
| <span class="sourceLineNo">730</span> scan.withStopRow(nextSourceKey.copyBytes());<a name="line.730"></a> |
| <span class="sourceLineNo">731</span> }<a name="line.731"></a> |
| <span class="sourceLineNo">732</span><a name="line.732"></a> |
| <span class="sourceLineNo">733</span> ResultScanner targetScanner = null;<a name="line.733"></a> |
| <span class="sourceLineNo">734</span> try {<a name="line.734"></a> |
| <span class="sourceLineNo">735</span> targetScanner = targetTable.getScanner(scan);<a name="line.735"></a> |
| <span class="sourceLineNo">736</span> for (Result row : targetScanner) {<a name="line.736"></a> |
| <span class="sourceLineNo">737</span> targetHasher.hashResult(row);<a name="line.737"></a> |
| <span class="sourceLineNo">738</span> }<a name="line.738"></a> |
| <span class="sourceLineNo">739</span> } finally {<a name="line.739"></a> |
| <span class="sourceLineNo">740</span> if (targetScanner != null) {<a name="line.740"></a> |
| <span class="sourceLineNo">741</span> targetScanner.close();<a name="line.741"></a> |
| <span class="sourceLineNo">742</span> }<a name="line.742"></a> |
| <span class="sourceLineNo">743</span> }<a name="line.743"></a> |
| <span class="sourceLineNo">744</span> } // else current batch ends exactly at split end row<a name="line.744"></a> |
| <span class="sourceLineNo">745</span><a name="line.745"></a> |
| <span class="sourceLineNo">746</span> finishBatchAndCompareHashes(context);<a name="line.746"></a> |
| <span class="sourceLineNo">747</span> }<a name="line.747"></a> |
| <span class="sourceLineNo">748</span> }<a name="line.748"></a> |
| <span class="sourceLineNo">749</span> }<a name="line.749"></a> |
| <span class="sourceLineNo">750</span><a name="line.750"></a> |
| <span class="sourceLineNo">751</span> private static final int NUM_ARGS = 3;<a name="line.751"></a> |
| <span class="sourceLineNo">752</span><a name="line.752"></a> |
| <span class="sourceLineNo">753</span> private static void printUsage(final String errorMsg) {<a name="line.753"></a> |
| <span class="sourceLineNo">754</span> if (errorMsg != null && errorMsg.length() > 0) {<a name="line.754"></a> |
| <span class="sourceLineNo">755</span> System.err.println("ERROR: " + errorMsg);<a name="line.755"></a> |
| <span class="sourceLineNo">756</span> System.err.println();<a name="line.756"></a> |
| <span class="sourceLineNo">757</span> }<a name="line.757"></a> |
| <span class="sourceLineNo">758</span> System.err.println("Usage: SyncTable [options] <sourcehashdir> <sourcetable> <targettable>");<a name="line.758"></a> |
| <span class="sourceLineNo">759</span> System.err.println();<a name="line.759"></a> |
| <span class="sourceLineNo">760</span> System.err.println("Options:");<a name="line.760"></a> |
| <span class="sourceLineNo">761</span><a name="line.761"></a> |
| <span class="sourceLineNo">762</span> System.err.println(" sourcezkcluster ZK cluster key of the source table");<a name="line.762"></a> |
| <span class="sourceLineNo">763</span> System.err.println(" (defaults to cluster in classpath's config)");<a name="line.763"></a> |
| <span class="sourceLineNo">764</span> System.err.println(" targetzkcluster ZK cluster key of the target table");<a name="line.764"></a> |
| <span class="sourceLineNo">765</span> System.err.println(" (defaults to cluster in classpath's config)");<a name="line.765"></a> |
| <span class="sourceLineNo">766</span> System.err.println(" dryrun if true, output counters but no writes");<a name="line.766"></a> |
| <span class="sourceLineNo">767</span> System.err.println(" (defaults to false)");<a name="line.767"></a> |
| <span class="sourceLineNo">768</span> System.err.println(" doDeletes if false, does not perform deletes");<a name="line.768"></a> |
| <span class="sourceLineNo">769</span> System.err.println(" (defaults to true)");<a name="line.769"></a> |
| <span class="sourceLineNo">770</span> System.err.println(" doPuts if false, does not perform puts");<a name="line.770"></a> |
| <span class="sourceLineNo">771</span> System.err.println(" (defaults to true)");<a name="line.771"></a> |
| <span class="sourceLineNo">772</span> System.err.println(" ignoreTimestamps if true, ignores cells timestamps while comparing ");<a name="line.772"></a> |
| <span class="sourceLineNo">773</span> System.err.println(" cell values. Any missing cell on target then gets");<a name="line.773"></a> |
| <span class="sourceLineNo">774</span> System.err.println(" added with current time as timestamp ");<a name="line.774"></a> |
| <span class="sourceLineNo">775</span> System.err.println(" (defaults to false)");<a name="line.775"></a> |
| <span class="sourceLineNo">776</span> System.err.println();<a name="line.776"></a> |
| <span class="sourceLineNo">777</span> System.err.println("Args:");<a name="line.777"></a> |
| <span class="sourceLineNo">778</span> System.err.println(" sourcehashdir path to HashTable output dir for source table");<a name="line.778"></a> |
| <span class="sourceLineNo">779</span> System.err.println(" (see org.apache.hadoop.hbase.mapreduce.HashTable)");<a name="line.779"></a> |
| <span class="sourceLineNo">780</span> System.err.println(" sourcetable Name of the source table to sync from");<a name="line.780"></a> |
| <span class="sourceLineNo">781</span> System.err.println(" targettable Name of the target table to sync to");<a name="line.781"></a> |
| <span class="sourceLineNo">782</span> System.err.println();<a name="line.782"></a> |
| <span class="sourceLineNo">783</span> System.err.println("Examples:");<a name="line.783"></a> |
| <span class="sourceLineNo">784</span> System.err.println(" For a dry run SyncTable of tableA from a remote source cluster");<a name="line.784"></a> |
| <span class="sourceLineNo">785</span> System.err.println(" to a local target cluster:");<a name="line.785"></a> |
| <span class="sourceLineNo">786</span> System.err.println(" $ hbase " + "org.apache.hadoop.hbase.mapreduce.SyncTable --dryrun=true"<a name="line.786"></a> |
| <span class="sourceLineNo">787</span> + " --sourcezkcluster=zk1.example.com,zk2.example.com,zk3.example.com:2181:/hbase"<a name="line.787"></a> |
| <span class="sourceLineNo">788</span> + " hdfs://nn:9000/hashes/tableA tableA tableA");<a name="line.788"></a> |
| <span class="sourceLineNo">789</span> }<a name="line.789"></a> |
| <span class="sourceLineNo">790</span><a name="line.790"></a> |
| <span class="sourceLineNo">791</span> private boolean doCommandLine(final String[] args) {<a name="line.791"></a> |
| <span class="sourceLineNo">792</span> if (args.length < NUM_ARGS) {<a name="line.792"></a> |
| <span class="sourceLineNo">793</span> printUsage(null);<a name="line.793"></a> |
| <span class="sourceLineNo">794</span> return false;<a name="line.794"></a> |
| <span class="sourceLineNo">795</span> }<a name="line.795"></a> |
| <span class="sourceLineNo">796</span> try {<a name="line.796"></a> |
| <span class="sourceLineNo">797</span> sourceHashDir = new Path(args[args.length - 3]);<a name="line.797"></a> |
| <span class="sourceLineNo">798</span> sourceTableName = args[args.length - 2];<a name="line.798"></a> |
| <span class="sourceLineNo">799</span> targetTableName = args[args.length - 1];<a name="line.799"></a> |
| <span class="sourceLineNo">800</span><a name="line.800"></a> |
| <span class="sourceLineNo">801</span> for (int i = 0; i < args.length - NUM_ARGS; i++) {<a name="line.801"></a> |
| <span class="sourceLineNo">802</span> String cmd = args[i];<a name="line.802"></a> |
| <span class="sourceLineNo">803</span> if (cmd.equals("-h") || cmd.startsWith("--h")) {<a name="line.803"></a> |
| <span class="sourceLineNo">804</span> printUsage(null);<a name="line.804"></a> |
| <span class="sourceLineNo">805</span> return false;<a name="line.805"></a> |
| <span class="sourceLineNo">806</span> }<a name="line.806"></a> |
| <span class="sourceLineNo">807</span><a name="line.807"></a> |
| <span class="sourceLineNo">808</span> final String sourceZkClusterKey = "--sourcezkcluster=";<a name="line.808"></a> |
| <span class="sourceLineNo">809</span> if (cmd.startsWith(sourceZkClusterKey)) {<a name="line.809"></a> |
| <span class="sourceLineNo">810</span> sourceZkCluster = cmd.substring(sourceZkClusterKey.length());<a name="line.810"></a> |
| <span class="sourceLineNo">811</span> continue;<a name="line.811"></a> |
| <span class="sourceLineNo">812</span> }<a name="line.812"></a> |
| <span class="sourceLineNo">813</span><a name="line.813"></a> |
| <span class="sourceLineNo">814</span> final String targetZkClusterKey = "--targetzkcluster=";<a name="line.814"></a> |
| <span class="sourceLineNo">815</span> if (cmd.startsWith(targetZkClusterKey)) {<a name="line.815"></a> |
| <span class="sourceLineNo">816</span> targetZkCluster = cmd.substring(targetZkClusterKey.length());<a name="line.816"></a> |
| <span class="sourceLineNo">817</span> continue;<a name="line.817"></a> |
| <span class="sourceLineNo">818</span> }<a name="line.818"></a> |
| <span class="sourceLineNo">819</span><a name="line.819"></a> |
| <span class="sourceLineNo">820</span> final String dryRunKey = "--dryrun=";<a name="line.820"></a> |
| <span class="sourceLineNo">821</span> if (cmd.startsWith(dryRunKey)) {<a name="line.821"></a> |
| <span class="sourceLineNo">822</span> dryRun = Boolean.parseBoolean(cmd.substring(dryRunKey.length()));<a name="line.822"></a> |
| <span class="sourceLineNo">823</span> continue;<a name="line.823"></a> |
| <span class="sourceLineNo">824</span> }<a name="line.824"></a> |
| <span class="sourceLineNo">825</span><a name="line.825"></a> |
| <span class="sourceLineNo">826</span> final String doDeletesKey = "--doDeletes=";<a name="line.826"></a> |
| <span class="sourceLineNo">827</span> if (cmd.startsWith(doDeletesKey)) {<a name="line.827"></a> |
| <span class="sourceLineNo">828</span> doDeletes = Boolean.parseBoolean(cmd.substring(doDeletesKey.length()));<a name="line.828"></a> |
| <span class="sourceLineNo">829</span> continue;<a name="line.829"></a> |
| <span class="sourceLineNo">830</span> }<a name="line.830"></a> |
| <span class="sourceLineNo">831</span><a name="line.831"></a> |
| <span class="sourceLineNo">832</span> final String doPutsKey = "--doPuts=";<a name="line.832"></a> |
| <span class="sourceLineNo">833</span> if (cmd.startsWith(doPutsKey)) {<a name="line.833"></a> |
| <span class="sourceLineNo">834</span> doPuts = Boolean.parseBoolean(cmd.substring(doPutsKey.length()));<a name="line.834"></a> |
| <span class="sourceLineNo">835</span> continue;<a name="line.835"></a> |
| <span class="sourceLineNo">836</span> }<a name="line.836"></a> |
| <span class="sourceLineNo">837</span><a name="line.837"></a> |
| <span class="sourceLineNo">838</span> final String ignoreTimestampsKey = "--ignoreTimestamps=";<a name="line.838"></a> |
| <span class="sourceLineNo">839</span> if (cmd.startsWith(ignoreTimestampsKey)) {<a name="line.839"></a> |
| <span class="sourceLineNo">840</span> ignoreTimestamps = Boolean.parseBoolean(cmd.substring(ignoreTimestampsKey.length()));<a name="line.840"></a> |
| <span class="sourceLineNo">841</span> continue;<a name="line.841"></a> |
| <span class="sourceLineNo">842</span> }<a name="line.842"></a> |
| <span class="sourceLineNo">843</span><a name="line.843"></a> |
| <span class="sourceLineNo">844</span> printUsage("Invalid argument '" + cmd + "'");<a name="line.844"></a> |
| <span class="sourceLineNo">845</span> return false;<a name="line.845"></a> |
| <span class="sourceLineNo">846</span> }<a name="line.846"></a> |
| <span class="sourceLineNo">847</span><a name="line.847"></a> |
| <span class="sourceLineNo">848</span> } catch (Exception e) {<a name="line.848"></a> |
| <span class="sourceLineNo">849</span> LOG.error("Failed to parse commandLine arguments", e);<a name="line.849"></a> |
| <span class="sourceLineNo">850</span> printUsage("Can't start because " + e.getMessage());<a name="line.850"></a> |
| <span class="sourceLineNo">851</span> return false;<a name="line.851"></a> |
| <span class="sourceLineNo">852</span> }<a name="line.852"></a> |
| <span class="sourceLineNo">853</span> return true;<a name="line.853"></a> |
| <span class="sourceLineNo">854</span> }<a name="line.854"></a> |
| <span class="sourceLineNo">855</span><a name="line.855"></a> |
| <span class="sourceLineNo">856</span> /**<a name="line.856"></a> |
| <span class="sourceLineNo">857</span> * Main entry point.<a name="line.857"></a> |
| <span class="sourceLineNo">858</span> */<a name="line.858"></a> |
| <span class="sourceLineNo">859</span> public static void main(String[] args) throws Exception {<a name="line.859"></a> |
| <span class="sourceLineNo">860</span> int ret = ToolRunner.run(new SyncTable(HBaseConfiguration.create()), args);<a name="line.860"></a> |
| <span class="sourceLineNo">861</span> System.exit(ret);<a name="line.861"></a> |
| <span class="sourceLineNo">862</span> }<a name="line.862"></a> |
| <span class="sourceLineNo">863</span><a name="line.863"></a> |
| <span class="sourceLineNo">864</span> @Override<a name="line.864"></a> |
| <span class="sourceLineNo">865</span> public int run(String[] args) throws Exception {<a name="line.865"></a> |
| <span class="sourceLineNo">866</span> String[] otherArgs = new GenericOptionsParser(getConf(), args).getRemainingArgs();<a name="line.866"></a> |
| <span class="sourceLineNo">867</span> if (!doCommandLine(otherArgs)) {<a name="line.867"></a> |
| <span class="sourceLineNo">868</span> return 1;<a name="line.868"></a> |
| <span class="sourceLineNo">869</span> }<a name="line.869"></a> |
| <span class="sourceLineNo">870</span><a name="line.870"></a> |
| <span class="sourceLineNo">871</span> Job job = createSubmittableJob(otherArgs);<a name="line.871"></a> |
| <span class="sourceLineNo">872</span> if (!job.waitForCompletion(true)) {<a name="line.872"></a> |
| <span class="sourceLineNo">873</span> LOG.info("Map-reduce job failed!");<a name="line.873"></a> |
| <span class="sourceLineNo">874</span> return 1;<a name="line.874"></a> |
| <span class="sourceLineNo">875</span> }<a name="line.875"></a> |
| <span class="sourceLineNo">876</span> counters = job.getCounters();<a name="line.876"></a> |
| <span class="sourceLineNo">877</span> return 0;<a name="line.877"></a> |
| <span class="sourceLineNo">878</span> }<a name="line.878"></a> |
| <span class="sourceLineNo">879</span><a name="line.879"></a> |
| <span class="sourceLineNo">880</span>}<a name="line.880"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |