http://git-wip-us.apache.org/repos/asf/hbase-site/blob/6c67ddd7/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.SyncMapper.html ---------------------------------------------------------------------- diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.SyncMapper.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.SyncMapper.html index 3165a6c..b6817d9 100644 --- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.SyncMapper.html +++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.SyncMapper.html @@ -71,728 +71,754 @@ <span class="sourceLineNo">063</span> static final String TARGET_TABLE_CONF_KEY = "sync.table.target.table.name";<a name="line.63"></a> <span class="sourceLineNo">064</span> static final String SOURCE_ZK_CLUSTER_CONF_KEY = "sync.table.source.zk.cluster";<a name="line.64"></a> <span class="sourceLineNo">065</span> static final String TARGET_ZK_CLUSTER_CONF_KEY = "sync.table.target.zk.cluster";<a name="line.65"></a> -<span class="sourceLineNo">066</span> static final String DRY_RUN_CONF_KEY="sync.table.dry.run";<a name="line.66"></a> -<span class="sourceLineNo">067</span><a name="line.67"></a> -<span class="sourceLineNo">068</span> Path sourceHashDir;<a name="line.68"></a> -<span class="sourceLineNo">069</span> String sourceTableName;<a name="line.69"></a> -<span class="sourceLineNo">070</span> String targetTableName;<a name="line.70"></a> -<span class="sourceLineNo">071</span><a name="line.71"></a> -<span class="sourceLineNo">072</span> String sourceZkCluster;<a name="line.72"></a> -<span class="sourceLineNo">073</span> String targetZkCluster;<a name="line.73"></a> -<span class="sourceLineNo">074</span> boolean dryRun;<a name="line.74"></a> -<span class="sourceLineNo">075</span><a name="line.75"></a> -<span class="sourceLineNo">076</span> Counters counters;<a name="line.76"></a> -<span class="sourceLineNo">077</span><a name="line.77"></a> -<span class="sourceLineNo">078</span> public SyncTable(Configuration conf) {<a name="line.78"></a> -<span class="sourceLineNo">079</span> super(conf);<a name="line.79"></a> -<span class="sourceLineNo">080</span> }<a name="line.80"></a> +<span class="sourceLineNo">066</span> static final String DRY_RUN_CONF_KEY = "sync.table.dry.run";<a name="line.66"></a> +<span class="sourceLineNo">067</span> static final String DO_DELETES_CONF_KEY = "sync.table.do.deletes";<a name="line.67"></a> +<span class="sourceLineNo">068</span> static final String DO_PUTS_CONF_KEY = "sync.table.do.puts";<a name="line.68"></a> +<span class="sourceLineNo">069</span><a name="line.69"></a> +<span class="sourceLineNo">070</span> Path sourceHashDir;<a name="line.70"></a> +<span class="sourceLineNo">071</span> String sourceTableName;<a name="line.71"></a> +<span class="sourceLineNo">072</span> String targetTableName;<a name="line.72"></a> +<span class="sourceLineNo">073</span><a name="line.73"></a> +<span class="sourceLineNo">074</span> String sourceZkCluster;<a name="line.74"></a> +<span class="sourceLineNo">075</span> String targetZkCluster;<a name="line.75"></a> +<span class="sourceLineNo">076</span> boolean dryRun;<a name="line.76"></a> +<span class="sourceLineNo">077</span> boolean doDeletes = true;<a name="line.77"></a> +<span class="sourceLineNo">078</span> boolean doPuts = true;<a name="line.78"></a> +<span class="sourceLineNo">079</span><a name="line.79"></a> +<span class="sourceLineNo">080</span> Counters counters;<a name="line.80"></a> <span class="sourceLineNo">081</span><a name="line.81"></a> -<span class="sourceLineNo">082</span> public Job createSubmittableJob(String[] args) throws IOException {<a name="line.82"></a> -<span class="sourceLineNo">083</span> FileSystem fs = sourceHashDir.getFileSystem(getConf());<a name="line.83"></a> -<span class="sourceLineNo">084</span> if (!fs.exists(sourceHashDir)) {<a name="line.84"></a> -<span class="sourceLineNo">085</span> throw new IOException("Source hash dir not found: " + sourceHashDir);<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> HashTable.TableHash tableHash = HashTable.TableHash.read(getConf(), sourceHashDir);<a name="line.88"></a> -<span class="sourceLineNo">089</span> LOG.info("Read source hash manifest: " + tableHash);<a name="line.89"></a> -<span class="sourceLineNo">090</span> LOG.info("Read " + tableHash.partitions.size() + " partition keys");<a name="line.90"></a> -<span class="sourceLineNo">091</span> if (!tableHash.tableName.equals(sourceTableName)) {<a name="line.91"></a> -<span class="sourceLineNo">092</span> LOG.warn("Table name mismatch - manifest indicates hash was taken from: "<a name="line.92"></a> -<span class="sourceLineNo">093</span> + tableHash.tableName + " but job is reading from: " + sourceTableName);<a name="line.93"></a> -<span class="sourceLineNo">094</span> }<a name="line.94"></a> -<span class="sourceLineNo">095</span> if (tableHash.numHashFiles != tableHash.partitions.size() + 1) {<a name="line.95"></a> -<span class="sourceLineNo">096</span> throw new RuntimeException("Hash data appears corrupt. The number of of hash files created"<a name="line.96"></a> -<span class="sourceLineNo">097</span> + " should be 1 more than the number of partition keys. However, the manifest file "<a name="line.97"></a> -<span class="sourceLineNo">098</span> + " says numHashFiles=" + tableHash.numHashFiles + " but the number of partition keys"<a name="line.98"></a> -<span class="sourceLineNo">099</span> + " found in the partitions file is " + tableHash.partitions.size());<a name="line.99"></a> -<span class="sourceLineNo">100</span> }<a name="line.100"></a> -<span class="sourceLineNo">101</span><a name="line.101"></a> -<span class="sourceLineNo">102</span> Path dataDir = new Path(sourceHashDir, HashTable.HASH_DATA_DIR);<a name="line.102"></a> -<span class="sourceLineNo">103</span> int dataSubdirCount = 0;<a name="line.103"></a> -<span class="sourceLineNo">104</span> for (FileStatus file : fs.listStatus(dataDir)) {<a name="line.104"></a> -<span class="sourceLineNo">105</span> if (file.getPath().getName().startsWith(HashTable.OUTPUT_DATA_FILE_PREFIX)) {<a name="line.105"></a> -<span class="sourceLineNo">106</span> dataSubdirCount++;<a name="line.106"></a> -<span class="sourceLineNo">107</span> }<a name="line.107"></a> -<span class="sourceLineNo">108</span> }<a name="line.108"></a> -<span class="sourceLineNo">109</span><a name="line.109"></a> -<span class="sourceLineNo">110</span> if (dataSubdirCount != tableHash.numHashFiles) {<a name="line.110"></a> -<span class="sourceLineNo">111</span> throw new RuntimeException("Hash data appears corrupt. The number of of hash files created"<a name="line.111"></a> -<span class="sourceLineNo">112</span> + " should be 1 more than the number of partition keys. However, the number of data dirs"<a name="line.112"></a> -<span class="sourceLineNo">113</span> + " found is " + dataSubdirCount + " but the number of partition keys"<a name="line.113"></a> -<span class="sourceLineNo">114</span> + " found in the partitions file is " + tableHash.partitions.size());<a name="line.114"></a> -<span class="sourceLineNo">115</span> }<a name="line.115"></a> -<span class="sourceLineNo">116</span><a name="line.116"></a> -<span class="sourceLineNo">117</span> Job job = Job.getInstance(getConf(),getConf().get("mapreduce.job.name",<a name="line.117"></a> -<span class="sourceLineNo">118</span> "syncTable_" + sourceTableName + "-" + targetTableName));<a name="line.118"></a> -<span class="sourceLineNo">119</span> Configuration jobConf = job.getConfiguration();<a name="line.119"></a> -<span class="sourceLineNo">120</span> job.setJarByClass(HashTable.class);<a name="line.120"></a> -<span class="sourceLineNo">121</span> jobConf.set(SOURCE_HASH_DIR_CONF_KEY, sourceHashDir.toString());<a name="line.121"></a> -<span class="sourceLineNo">122</span> jobConf.set(SOURCE_TABLE_CONF_KEY, sourceTableName);<a name="line.122"></a> -<span class="sourceLineNo">123</span> jobConf.set(TARGET_TABLE_CONF_KEY, targetTableName);<a name="line.123"></a> -<span class="sourceLineNo">124</span> if (sourceZkCluster != null) {<a name="line.124"></a> -<span class="sourceLineNo">125</span> jobConf.set(SOURCE_ZK_CLUSTER_CONF_KEY, sourceZkCluster);<a name="line.125"></a> -<span class="sourceLineNo">126</span> }<a name="line.126"></a> -<span class="sourceLineNo">127</span> if (targetZkCluster != null) {<a name="line.127"></a> -<span class="sourceLineNo">128</span> jobConf.set(TARGET_ZK_CLUSTER_CONF_KEY, targetZkCluster);<a name="line.128"></a> -<span class="sourceLineNo">129</span> }<a name="line.129"></a> -<span class="sourceLineNo">130</span> jobConf.setBoolean(DRY_RUN_CONF_KEY, dryRun);<a name="line.130"></a> -<span class="sourceLineNo">131</span><a name="line.131"></a> -<span class="sourceLineNo">132</span> TableMapReduceUtil.initTableMapperJob(targetTableName, tableHash.initScan(),<a name="line.132"></a> -<span class="sourceLineNo">133</span> SyncMapper.class, null, null, job);<a name="line.133"></a> -<span class="sourceLineNo">134</span><a name="line.134"></a> -<span class="sourceLineNo">135</span> job.setNumReduceTasks(0);<a name="line.135"></a> -<span class="sourceLineNo">136</span><a name="line.136"></a> -<span class="sourceLineNo">137</span> if (dryRun) {<a name="line.137"></a> -<span class="sourceLineNo">138</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.138"></a> -<span class="sourceLineNo">139</span> } else {<a name="line.139"></a> -<span class="sourceLineNo">140</span> // No reducers. Just write straight to table. Call initTableReducerJob<a name="line.140"></a> -<span class="sourceLineNo">141</span> // because it sets up the TableOutputFormat.<a name="line.141"></a> -<span class="sourceLineNo">142</span> TableMapReduceUtil.initTableReducerJob(targetTableName, null, job, null,<a name="line.142"></a> -<span class="sourceLineNo">143</span> targetZkCluster, null, null);<a name="line.143"></a> -<span class="sourceLineNo">144</span><a name="line.144"></a> -<span class="sourceLineNo">145</span> // would be nice to add an option for bulk load instead<a name="line.145"></a> -<span class="sourceLineNo">146</span> }<a name="line.146"></a> -<span class="sourceLineNo">147</span><a name="line.147"></a> -<span class="sourceLineNo">148</span> // Obtain an authentication token, for the specified cluster, on behalf of the current user<a name="line.148"></a> -<span class="sourceLineNo">149</span> if (sourceZkCluster != null) {<a name="line.149"></a> -<span class="sourceLineNo">150</span> Configuration peerConf =<a name="line.150"></a> -<span class="sourceLineNo">151</span> HBaseConfiguration.createClusterConf(job.getConfiguration(), sourceZkCluster);<a name="line.151"></a> -<span class="sourceLineNo">152</span> TableMapReduceUtil.initCredentialsForCluster(job, peerConf);<a name="line.152"></a> -<span class="sourceLineNo">153</span> }<a name="line.153"></a> -<span class="sourceLineNo">154</span> return job;<a name="line.154"></a> -<span class="sourceLineNo">155</span> }<a name="line.155"></a> -<span class="sourceLineNo">156</span><a name="line.156"></a> -<span class="sourceLineNo">157</span> public static class SyncMapper extends TableMapper<ImmutableBytesWritable, Mutation> {<a name="line.157"></a> -<span class="sourceLineNo">158</span> Path sourceHashDir;<a name="line.158"></a> -<span class="sourceLineNo">159</span><a name="line.159"></a> -<span class="sourceLineNo">160</span> Connection sourceConnection;<a name="line.160"></a> -<span class="sourceLineNo">161</span> Connection targetConnection;<a name="line.161"></a> -<span class="sourceLineNo">162</span> Table sourceTable;<a name="line.162"></a> -<span class="sourceLineNo">163</span> Table targetTable;<a name="line.163"></a> -<span class="sourceLineNo">164</span> boolean dryRun;<a name="line.164"></a> +<span class="sourceLineNo">082</span> public SyncTable(Configuration conf) {<a name="line.82"></a> +<span class="sourceLineNo">083</span> super(conf);<a name="line.83"></a> +<span class="sourceLineNo">084</span> }<a name="line.84"></a> +<span class="sourceLineNo">085</span><a name="line.85"></a> +<span class="sourceLineNo">086</span> public Job createSubmittableJob(String[] args) throws IOException {<a name="line.86"></a> +<span class="sourceLineNo">087</span> FileSystem fs = sourceHashDir.getFileSystem(getConf());<a name="line.87"></a> +<span class="sourceLineNo">088</span> if (!fs.exists(sourceHashDir)) {<a name="line.88"></a> +<span class="sourceLineNo">089</span> throw new IOException("Source hash dir not found: " + sourceHashDir);<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> HashTable.TableHash tableHash = HashTable.TableHash.read(getConf(), sourceHashDir);<a name="line.92"></a> +<span class="sourceLineNo">093</span> LOG.info("Read source hash manifest: " + tableHash);<a name="line.93"></a> +<span class="sourceLineNo">094</span> LOG.info("Read " + tableHash.partitions.size() + " partition keys");<a name="line.94"></a> +<span class="sourceLineNo">095</span> if (!tableHash.tableName.equals(sourceTableName)) {<a name="line.95"></a> +<span class="sourceLineNo">096</span> LOG.warn("Table name mismatch - manifest indicates hash was taken from: "<a name="line.96"></a> +<span class="sourceLineNo">097</span> + tableHash.tableName + " but job is reading from: " + sourceTableName);<a name="line.97"></a> +<span class="sourceLineNo">098</span> }<a name="line.98"></a> +<span class="sourceLineNo">099</span> if (tableHash.numHashFiles != tableHash.partitions.size() + 1) {<a name="line.99"></a> +<span class="sourceLineNo">100</span> throw new RuntimeException("Hash data appears corrupt. The number of of hash files created"<a name="line.100"></a> +<span class="sourceLineNo">101</span> + " should be 1 more than the number of partition keys. However, the manifest file "<a name="line.101"></a> +<span class="sourceLineNo">102</span> + " says numHashFiles=" + tableHash.numHashFiles + " but the number of partition keys"<a name="line.102"></a> +<span class="sourceLineNo">103</span> + " found in the partitions file is " + tableHash.partitions.size());<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> Path dataDir = new Path(sourceHashDir, HashTable.HASH_DATA_DIR);<a name="line.106"></a> +<span class="sourceLineNo">107</span> int dataSubdirCount = 0;<a name="line.107"></a> +<span class="sourceLineNo">108</span> for (FileStatus file : fs.listStatus(dataDir)) {<a name="line.108"></a> +<span class="sourceLineNo">109</span> if (file.getPath().getName().startsWith(HashTable.OUTPUT_DATA_FILE_PREFIX)) {<a name="line.109"></a> +<span class="sourceLineNo">110</span> dataSubdirCount++;<a name="line.110"></a> +<span class="sourceLineNo">111</span> }<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> if (dataSubdirCount != tableHash.numHashFiles) {<a name="line.114"></a> +<span class="sourceLineNo">115</span> throw new RuntimeException("Hash data appears corrupt. The number of of hash files created"<a name="line.115"></a> +<span class="sourceLineNo">116</span> + " should be 1 more than the number of partition keys. However, the number of data dirs"<a name="line.116"></a> +<span class="sourceLineNo">117</span> + " found is " + dataSubdirCount + " but the number of partition keys"<a name="line.117"></a> +<span class="sourceLineNo">118</span> + " found in the partitions file is " + tableHash.partitions.size());<a name="line.118"></a> +<span class="sourceLineNo">119</span> }<a name="line.119"></a> +<span class="sourceLineNo">120</span><a name="line.120"></a> +<span class="sourceLineNo">121</span> Job job = Job.getInstance(getConf(),getConf().get("mapreduce.job.name",<a name="line.121"></a> +<span class="sourceLineNo">122</span> "syncTable_" + sourceTableName + "-" + targetTableName));<a name="line.122"></a> +<span class="sourceLineNo">123</span> Configuration jobConf = job.getConfiguration();<a name="line.123"></a> +<span class="sourceLineNo">124</span> job.setJarByClass(HashTable.class);<a name="line.124"></a> +<span class="sourceLineNo">125</span> jobConf.set(SOURCE_HASH_DIR_CONF_KEY, sourceHashDir.toString());<a name="line.125"></a> +<span class="sourceLineNo">126</span> jobConf.set(SOURCE_TABLE_CONF_KEY, sourceTableName);<a name="line.126"></a> +<span class="sourceLineNo">127</span> jobConf.set(TARGET_TABLE_CONF_KEY, targetTableName);<a name="line.127"></a> +<span class="sourceLineNo">128</span> if (sourceZkCluster != null) {<a name="line.128"></a> +<span class="sourceLineNo">129</span> jobConf.set(SOURCE_ZK_CLUSTER_CONF_KEY, sourceZkCluster);<a name="line.129"></a> +<span class="sourceLineNo">130</span> }<a name="line.130"></a> +<span class="sourceLineNo">131</span> if (targetZkCluster != null) {<a name="line.131"></a> +<span class="sourceLineNo">132</span> jobConf.set(TARGET_ZK_CLUSTER_CONF_KEY, targetZkCluster);<a name="line.132"></a> +<span class="sourceLineNo">133</span> }<a name="line.133"></a> +<span class="sourceLineNo">134</span> jobConf.setBoolean(DRY_RUN_CONF_KEY, dryRun);<a name="line.134"></a> +<span class="sourceLineNo">135</span> jobConf.setBoolean(DO_DELETES_CONF_KEY, doDeletes);<a name="line.135"></a> +<span class="sourceLineNo">136</span> jobConf.setBoolean(DO_PUTS_CONF_KEY, doPuts);<a name="line.136"></a> +<span class="sourceLineNo">137</span><a name="line.137"></a> +<span class="sourceLineNo">138</span> TableMapReduceUtil.initTableMapperJob(targetTableName, tableHash.initScan(),<a name="line.138"></a> +<span class="sourceLineNo">139</span> SyncMapper.class, null, null, job);<a name="line.139"></a> +<span class="sourceLineNo">140</span><a name="line.140"></a> +<span class="sourceLineNo">141</span> job.setNumReduceTasks(0);<a name="line.141"></a> +<span class="sourceLineNo">142</span><a name="line.142"></a> +<span class="sourceLineNo">143</span> if (dryRun) {<a name="line.143"></a> +<span class="sourceLineNo">144</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.144"></a> +<span class="sourceLineNo">145</span> } else {<a name="line.145"></a> +<span class="sourceLineNo">146</span> // No reducers. Just write straight to table. Call initTableReducerJob<a name="line.146"></a> +<span class="sourceLineNo">147</span> // because it sets up the TableOutputFormat.<a name="line.147"></a> +<span class="sourceLineNo">148</span> TableMapReduceUtil.initTableReducerJob(targetTableName, null, job, null,<a name="line.148"></a> +<span class="sourceLineNo">149</span> targetZkCluster, null, null);<a name="line.149"></a> +<span class="sourceLineNo">150</span><a name="line.150"></a> +<span class="sourceLineNo">151</span> // would be nice to add an option for bulk load instead<a name="line.151"></a> +<span class="sourceLineNo">152</span> }<a name="line.152"></a> +<span class="sourceLineNo">153</span><a name="line.153"></a> +<span class="sourceLineNo">154</span> // Obtain an authentication token, for the specified cluster, on behalf of the current user<a name="line.154"></a> +<span class="sourceLineNo">155</span> if (sourceZkCluster != null) {<a name="line.155"></a> +<span class="sourceLineNo">156</span> Configuration peerConf =<a name="line.156"></a> +<span class="sourceLineNo">157</span> HBaseConfiguration.createClusterConf(job.getConfiguration(), sourceZkCluster);<a name="line.157"></a> +<span class="sourceLineNo">158</span> TableMapReduceUtil.initCredentialsForCluster(job, peerConf);<a name="line.158"></a> +<span class="sourceLineNo">159</span> }<a name="line.159"></a> +<span class="sourceLineNo">160</span> return job;<a name="line.160"></a> +<span class="sourceLineNo">161</span> }<a name="line.161"></a> +<span class="sourceLineNo">162</span><a name="line.162"></a> +<span class="sourceLineNo">163</span> public static class SyncMapper extends TableMapper<ImmutableBytesWritable, Mutation> {<a name="line.163"></a> +<span class="sourceLineNo">164</span> Path sourceHashDir;<a name="line.164"></a> <span class="sourceLineNo">165</span><a name="line.165"></a> -<span class="sourceLineNo">166</span> HashTable.TableHash sourceTableHash;<a name="line.166"></a> -<span class="sourceLineNo">167</span> HashTable.TableHash.Reader sourceHashReader;<a name="line.167"></a> -<span class="sourceLineNo">168</span> ImmutableBytesWritable currentSourceHash;<a name="line.168"></a> -<span class="sourceLineNo">169</span> ImmutableBytesWritable nextSourceKey;<a name="line.169"></a> -<span class="sourceLineNo">170</span> HashTable.ResultHasher targetHasher;<a name="line.170"></a> -<span class="sourceLineNo">171</span><a name="line.171"></a> -<span class="sourceLineNo">172</span> Throwable mapperException;<a name="line.172"></a> +<span class="sourceLineNo">166</span> Connection sourceConnection;<a name="line.166"></a> +<span class="sourceLineNo">167</span> Connection targetConnection;<a name="line.167"></a> +<span class="sourceLineNo">168</span> Table sourceTable;<a name="line.168"></a> +<span class="sourceLineNo">169</span> Table targetTable;<a name="line.169"></a> +<span class="sourceLineNo">170</span> boolean dryRun;<a name="line.170"></a> +<span class="sourceLineNo">171</span> boolean doDeletes = true;<a name="line.171"></a> +<span class="sourceLineNo">172</span> boolean doPuts = true;<a name="line.172"></a> <span class="sourceLineNo">173</span><a name="line.173"></a> -<span class="sourceLineNo">174</span> public static enum Counter { BATCHES, HASHES_MATCHED, HASHES_NOT_MATCHED, SOURCEMISSINGROWS,<a name="line.174"></a> -<span class="sourceLineNo">175</span> SOURCEMISSINGCELLS, TARGETMISSINGROWS, TARGETMISSINGCELLS, ROWSWITHDIFFS, DIFFERENTCELLVALUES,<a name="line.175"></a> -<span class="sourceLineNo">176</span> MATCHINGROWS, MATCHINGCELLS, EMPTY_BATCHES, RANGESMATCHED, RANGESNOTMATCHED<a name="line.176"></a> -<span class="sourceLineNo">177</span> }<a name="line.177"></a> -<span class="sourceLineNo">178</span><a name="line.178"></a> -<span class="sourceLineNo">179</span> @Override<a name="line.179"></a> -<span class="sourceLineNo">180</span> protected void setup(Context context) throws IOException {<a name="line.180"></a> +<span class="sourceLineNo">174</span> HashTable.TableHash sourceTableHash;<a name="line.174"></a> +<span class="sourceLineNo">175</span> HashTable.TableHash.Reader sourceHashReader;<a name="line.175"></a> +<span class="sourceLineNo">176</span> ImmutableBytesWritable currentSourceHash;<a name="line.176"></a> +<span class="sourceLineNo">177</span> ImmutableBytesWritable nextSourceKey;<a name="line.177"></a> +<span class="sourceLineNo">178</span> HashTable.ResultHasher targetHasher;<a name="line.178"></a> +<span class="sourceLineNo">179</span><a name="line.179"></a> +<span class="sourceLineNo">180</span> Throwable mapperException;<a name="line.180"></a> <span class="sourceLineNo">181</span><a name="line.181"></a> -<span class="sourceLineNo">182</span> Configuration conf = context.getConfiguration();<a name="line.182"></a> -<span class="sourceLineNo">183</span> sourceHashDir = new Path(conf.get(SOURCE_HASH_DIR_CONF_KEY));<a name="line.183"></a> -<span class="sourceLineNo">184</span> sourceConnection = openConnection(conf, SOURCE_ZK_CLUSTER_CONF_KEY, null);<a name="line.184"></a> -<span class="sourceLineNo">185</span> targetConnection = openConnection(conf, TARGET_ZK_CLUSTER_CONF_KEY,<a name="line.185"></a> -<span class="sourceLineNo">186</span> TableOutputFormat.OUTPUT_CONF_PREFIX);<a name="line.186"></a> -<span class="sourceLineNo">187</span> sourceTable = openTable(sourceConnection, conf, SOURCE_TABLE_CONF_KEY);<a name="line.187"></a> -<span class="sourceLineNo">188</span> targetTable = openTable(targetConnection, conf, TARGET_TABLE_CONF_KEY);<a name="line.188"></a> -<span class="sourceLineNo">189</span> dryRun = conf.getBoolean(SOURCE_TABLE_CONF_KEY, false);<a name="line.189"></a> -<span class="sourceLineNo">190</span><a name="line.190"></a> -<span class="sourceLineNo">191</span> sourceTableHash = HashTable.TableHash.read(conf, sourceHashDir);<a name="line.191"></a> -<span class="sourceLineNo">192</span> LOG.info("Read source hash manifest: " + sourceTableHash);<a name="line.192"></a> -<span class="sourceLineNo">193</span> LOG.info("Read " + sourceTableHash.partitions.size() + " partition keys");<a name="line.193"></a> -<span class="sourceLineNo">194</span><a name="line.194"></a> -<span class="sourceLineNo">195</span> TableSplit split = (TableSplit) context.getInputSplit();<a name="line.195"></a> -<span class="sourceLineNo">196</span> ImmutableBytesWritable splitStartKey = new ImmutableBytesWritable(split.getStartRow());<a name="line.196"></a> -<span class="sourceLineNo">197</span><a name="line.197"></a> -<span class="sourceLineNo">198</span> sourceHashReader = sourceTableHash.newReader(conf, splitStartKey);<a name="line.198"></a> -<span class="sourceLineNo">199</span> findNextKeyHashPair();<a name="line.199"></a> +<span class="sourceLineNo">182</span> public static enum Counter { BATCHES, HASHES_MATCHED, HASHES_NOT_MATCHED, SOURCEMISSINGROWS,<a name="line.182"></a> +<span class="sourceLineNo">183</span> SOURCEMISSINGCELLS, TARGETMISSINGROWS, TARGETMISSINGCELLS, ROWSWITHDIFFS, DIFFERENTCELLVALUES,<a name="line.183"></a> +<span class="sourceLineNo">184</span> MATCHINGROWS, MATCHINGCELLS, EMPTY_BATCHES, RANGESMATCHED, RANGESNOTMATCHED<a name="line.184"></a> +<span class="sourceLineNo">185</span> }<a name="line.185"></a> +<span class="sourceLineNo">186</span><a name="line.186"></a> +<span class="sourceLineNo">187</span> @Override<a name="line.187"></a> +<span class="sourceLineNo">188</span> protected void setup(Context context) throws IOException {<a name="line.188"></a> +<span class="sourceLineNo">189</span><a name="line.189"></a> +<span class="sourceLineNo">190</span> Configuration conf = context.getConfiguration();<a name="line.190"></a> +<span class="sourceLineNo">191</span> sourceHashDir = new Path(conf.get(SOURCE_HASH_DIR_CONF_KEY));<a name="line.191"></a> +<span class="sourceLineNo">192</span> sourceConnection = openConnection(conf, SOURCE_ZK_CLUSTER_CONF_KEY, null);<a name="line.192"></a> +<span class="sourceLineNo">193</span> targetConnection = openConnection(conf, TARGET_ZK_CLUSTER_CONF_KEY,<a name="line.193"></a> +<span class="sourceLineNo">194</span> TableOutputFormat.OUTPUT_CONF_PREFIX);<a name="line.194"></a> +<span class="sourceLineNo">195</span> sourceTable = openTable(sourceConnection, conf, SOURCE_TABLE_CONF_KEY);<a name="line.195"></a> +<span class="sourceLineNo">196</span> targetTable = openTable(targetConnection, conf, TARGET_TABLE_CONF_KEY);<a name="line.196"></a> +<span class="sourceLineNo">197</span> dryRun = conf.getBoolean(DRY_RUN_CONF_KEY, false);<a name="line.197"></a> +<span class="sourceLineNo">198</span> doDeletes = conf.getBoolean(DO_DELETES_CONF_KEY, true);<a name="line.198"></a> +<span class="sourceLineNo">199</span> doPuts = conf.getBoolean(DO_PUTS_CONF_KEY, true);<a name="line.199"></a> <span class="sourceLineNo">200</span><a name="line.200"></a> -<span class="sourceLineNo">201</span> // create a hasher, but don't start it right away<a name="line.201"></a> -<span class="sourceLineNo">202</span> // instead, find the first hash batch at or after the start row<a name="line.202"></a> -<span class="sourceLineNo">203</span> // and skip any rows that come before. they will be caught by the previous task<a name="line.203"></a> -<span class="sourceLineNo">204</span> targetHasher = new HashTable.ResultHasher();<a name="line.204"></a> -<span class="sourceLineNo">205</span> }<a name="line.205"></a> -<span class="sourceLineNo">206</span><a name="line.206"></a> -<span class="sourceLineNo">207</span> private static Connection openConnection(Configuration conf, String zkClusterConfKey,<a name="line.207"></a> -<span class="sourceLineNo">208</span> String configPrefix)<a name="line.208"></a> -<span class="sourceLineNo">209</span> throws IOException {<a name="line.209"></a> -<span class="sourceLineNo">210</span> String zkCluster = conf.get(zkClusterConfKey);<a name="line.210"></a> -<span class="sourceLineNo">211</span> Configuration clusterConf = HBaseConfiguration.createClusterConf(conf,<a name="line.211"></a> -<span class="sourceLineNo">212</span> zkCluster, configPrefix);<a name="line.212"></a> -<span class="sourceLineNo">213</span> return ConnectionFactory.createConnection(clusterConf);<a name="line.213"></a> -<span class="sourceLineNo">214</span> }<a name="line.214"></a> -<span class="sourceLineNo">215</span><a name="line.215"></a> -<span class="sourceLineNo">216</span> private static Table openTable(Connection connection, Configuration conf,<a name="line.216"></a> -<span class="sourceLineNo">217</span> String tableNameConfKey) throws IOException {<a name="line.217"></a> -<span class="sourceLineNo">218</span> return connection.getTable(TableName.valueOf(conf.get(tableNameConfKey)));<a name="line.218"></a> -<span class="sourceLineNo">219</span> }<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> * Attempt to read the next source key/hash pair.<a name="line.222"></a> -<span class="sourceLineNo">223</span> * If there are no more, set nextSourceKey to null<a name="line.223"></a> -<span class="sourceLineNo">224</span> */<a name="line.224"></a> -<span class="sourceLineNo">225</span> private void findNextKeyHashPair() throws IOException {<a name="line.225"></a> -<span class="sourceLineNo">226</span> boolean hasNext = sourceHashReader.next();<a name="line.226"></a> -<span class="sourceLineNo">227</span> if (hasNext) {<a name="line.227"></a> -<span class="sourceLineNo">228</span> nextSourceKey = sourceHashReader.getCurrentKey();<a name="line.228"></a> -<span class="sourceLineNo">229</span> } else {<a name="line.229"></a> -<span class="sourceLineNo">230</span> // no more keys - last hash goes to the end<a name="line.230"></a> -<span class="sourceLineNo">231</span> nextSourceKey = null;<a name="line.231"></a> -<span class="sourceLineNo">232</span> }<a name="line.232"></a> -<span class="sourceLineNo">233</span> }<a name="line.233"></a> -<span class="sourceLineNo">234</span><a name="line.234"></a> -<span class="sourceLineNo">235</span> @Override<a name="line.235"></a> -<span class="sourceLineNo">236</span> protected void map(ImmutableBytesWritable key, Result value, Context context)<a name="line.236"></a> -<span class="sourceLineNo">237</span> throws IOException, InterruptedException {<a name="line.237"></a> -<span class="sourceLineNo">238</span> try {<a name="line.238"></a> -<span class="sourceLineNo">239</span> // first, finish any hash batches that end before the scanned row<a name="line.239"></a> -<span class="sourceLineNo">240</span> while (nextSourceKey != null && key.compareTo(nextSourceKey) >= 0) {<a name="line.240"></a> -<span class="sourceLineNo">241</span> moveToNextBatch(context);<a name="line.241"></a> -<span class="sourceLineNo">242</span> }<a name="line.242"></a> -<span class="sourceLineNo">243</span><a name="line.243"></a> -<span class="sourceLineNo">244</span> // next, add the scanned row (as long as we've reached the first batch)<a name="line.244"></a> -<span class="sourceLineNo">245</span> if (targetHasher.isBatchStarted()) {<a name="line.245"></a> -<span class="sourceLineNo">246</span> targetHasher.hashResult(value);<a name="line.246"></a> -<span class="sourceLineNo">247</span> }<a name="line.247"></a> -<span class="sourceLineNo">248</span> } catch (Throwable t) {<a name="line.248"></a> -<span class="sourceLineNo">249</span> mapperException = t;<a name="line.249"></a> -<span class="sourceLineNo">250</span> Throwables.propagateIfInstanceOf(t, IOException.class);<a name="line.250"></a> -<span class="sourceLineNo">251</span> Throwables.propagateIfInstanceOf(t, InterruptedException.class);<a name="line.251"></a> -<span class="sourceLineNo">252</span> Throwables.propagate(t);<a name="line.252"></a> -<span class="sourceLineNo">253</span> }<a name="line.253"></a> -<span class="sourceLineNo">254</span> }<a name="line.254"></a> -<span class="sourceLineNo">255</span><a name="line.255"></a> -<span class="sourceLineNo">256</span> /**<a name="line.256"></a> -<span class="sourceLineNo">257</span> * If there is an open hash batch, complete it and sync if there are diffs.<a name="line.257"></a> -<span class="sourceLineNo">258</span> * Start a new batch, and seek to read the<a name="line.258"></a> -<span class="sourceLineNo">259</span> */<a name="line.259"></a> -<span class="sourceLineNo">260</span> private void moveToNextBatch(Context context) throws IOException, InterruptedException {<a name="line.260"></a> -<span class="sourceLineNo">261</span> if (targetHasher.isBatchStarted()) {<a name="line.261"></a> -<span class="sourceLineNo">262</span> finishBatchAndCompareHashes(context);<a name="line.262"></a> +<span class="sourceLineNo">201</span> sourceTableHash = HashTable.TableHash.read(conf, sourceHashDir);<a name="line.201"></a> +<span class="sourceLineNo">202</span> LOG.info("Read source hash manifest: " + sourceTableHash);<a name="line.202"></a> +<span class="sourceLineNo">203</span> LOG.info("Read " + sourceTableHash.partitions.size() + " partition keys");<a name="line.203"></a> +<span class="sourceLineNo">204</span><a name="line.204"></a> +<span class="sourceLineNo">205</span> TableSplit split = (TableSplit) context.getInputSplit();<a name="line.205"></a> +<span class="sourceLineNo">206</span> ImmutableBytesWritable splitStartKey = new ImmutableBytesWritable(split.getStartRow());<a name="line.206"></a> +<span class="sourceLineNo">207</span><a name="line.207"></a> +<span class="sourceLineNo">208</span> sourceHashReader = sourceTableHash.newReader(conf, splitStartKey);<a name="line.208"></a> +<span class="sourceLineNo">209</span> findNextKeyHashPair();<a name="line.209"></a> +<span class="sourceLineNo">210</span><a name="line.210"></a> +<span class="sourceLineNo">211</span> // create a hasher, but don't start it right away<a name="line.211"></a> +<span class="sourceLineNo">212</span> // instead, find the first hash batch at or after the start row<a name="line.212"></a> +<span class="sourceLineNo">213</span> // and skip any rows that come before. they will be caught by the previous task<a name="line.213"></a> +<span class="sourceLineNo">214</span> targetHasher = new HashTable.ResultHasher();<a name="line.214"></a> +<span class="sourceLineNo">215</span> }<a name="line.215"></a> +<span class="sourceLineNo">216</span><a name="line.216"></a> +<span class="sourceLineNo">217</span> private static Connection openConnection(Configuration conf, String zkClusterConfKey,<a name="line.217"></a> +<span class="sourceLineNo">218</span> String configPrefix)<a name="line.218"></a> +<span class="sourceLineNo">219</span> throws IOException {<a name="line.219"></a> +<span class="sourceLineNo">220</span> String zkCluster = conf.get(zkClusterConfKey);<a name="line.220"></a> +<span class="sourceLineNo">221</span> Configuration clusterConf = HBaseConfiguration.createClusterConf(conf,<a name="line.221"></a> +<span class="sourceLineNo">222</span> zkCluster, configPrefix);<a name="line.222"></a> +<span class="sourceLineNo">223</span> return ConnectionFactory.createConnection(clusterConf);<a name="line.223"></a> +<span class="sourceLineNo">224</span> }<a name="line.224"></a> +<span class="sourceLineNo">225</span><a name="line.225"></a> +<span class="sourceLineNo">226</span> private static Table openTable(Connection connection, Configuration conf,<a name="line.226"></a> +<span class="sourceLineNo">227</span> String tableNameConfKey) throws IOException {<a name="line.227"></a> +<span class="sourceLineNo">228</span> return connection.getTable(TableName.valueOf(conf.get(tableNameConfKey)));<a name="line.228"></a> +<span class="sourceLineNo">229</span> }<a name="line.229"></a> +<span class="sourceLineNo">230</span><a name="line.230"></a> +<span class="sourceLineNo">231</span> /**<a name="line.231"></a> +<span class="sourceLineNo">232</span> * Attempt to read the next source key/hash pair.<a name="line.232"></a> +<span class="sourceLineNo">233</span> * If there are no more, set nextSourceKey to null<a name="line.233"></a> +<span class="sourceLineNo">234</span> */<a name="line.234"></a> +<span class="sourceLineNo">235</span> private void findNextKeyHashPair() throws IOException {<a name="line.235"></a> +<span class="sourceLineNo">236</span> boolean hasNext = sourceHashReader.next();<a name="line.236"></a> +<span class="sourceLineNo">237</span> if (hasNext) {<a name="line.237"></a> +<span class="sourceLineNo">238</span> nextSourceKey = sourceHashReader.getCurrentKey();<a name="line.238"></a> +<span class="sourceLineNo">239</span> } else {<a name="line.239"></a> +<span class="sourceLineNo">240</span> // no more keys - last hash goes to the end<a name="line.240"></a> +<span class="sourceLineNo">241</span> nextSourceKey = null;<a name="line.241"></a> +<span class="sourceLineNo">242</span> }<a name="line.242"></a> +<span class="sourceLineNo">243</span> }<a name="line.243"></a> +<span class="sourceLineNo">244</span><a name="line.244"></a> +<span class="sourceLineNo">245</span> @Override<a name="line.245"></a> +<span class="sourceLineNo">246</span> protected void map(ImmutableBytesWritable key, Result value, Context context)<a name="line.246"></a> +<span class="sourceLineNo">247</span> throws IOException, InterruptedException {<a name="line.247"></a> +<span class="sourceLineNo">248</span> try {<a name="line.248"></a> +<span class="sourceLineNo">249</span> // first, finish any hash batches that end before the scanned row<a name="line.249"></a> +<span class="sourceLineNo">250</span> while (nextSourceKey != null && key.compareTo(nextSourceKey) >= 0) {<a name="line.250"></a> +<span class="sourceLineNo">251</span> moveToNextBatch(context);<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> // next, add the scanned row (as long as we've reached the first batch)<a name="line.254"></a> +<span class="sourceLineNo">255</span> if (targetHasher.isBatchStarted()) {<a name="line.255"></a> +<span class="sourceLineNo">256</span> targetHasher.hashResult(value);<a name="line.256"></a> +<span class="sourceLineNo">257</span> }<a name="line.257"></a> +<span class="sourceLineNo">258</span> } catch (Throwable t) {<a name="line.258"></a> +<span class="sourceLineNo">259</span> mapperException = t;<a name="line.259"></a> +<span class="sourceLineNo">260</span> Throwables.propagateIfInstanceOf(t, IOException.class);<a name="line.260"></a> +<span class="sourceLineNo">261</span> Throwables.propagateIfInstanceOf(t, InterruptedException.class);<a name="line.261"></a> +<span class="sourceLineNo">262</span> Throwables.propagate(t);<a name="line.262"></a> <span class="sourceLineNo">263</span> }<a name="line.263"></a> -<span class="sourceLineNo">264</span> targetHasher.startBatch(nextSourceKey);<a name="line.264"></a> -<span class="sourceLineNo">265</span> currentSourceHash = sourceHashReader.getCurrentHash();<a name="line.265"></a> -<span class="sourceLineNo">266</span><a name="line.266"></a> -<span class="sourceLineNo">267</span> findNextKeyHashPair();<a name="line.267"></a> -<span class="sourceLineNo">268</span> }<a name="line.268"></a> -<span class="sourceLineNo">269</span><a name="line.269"></a> -<span class="sourceLineNo">270</span> /**<a name="line.270"></a> -<span class="sourceLineNo">271</span> * Finish the currently open hash batch.<a name="line.271"></a> -<span class="sourceLineNo">272</span> * Compare the target hash to the given source hash.<a name="line.272"></a> -<span class="sourceLineNo">273</span> * If they do not match, then sync the covered key range.<a name="line.273"></a> -<span class="sourceLineNo">274</span> */<a name="line.274"></a> -<span class="sourceLineNo">275</span> private void finishBatchAndCompareHashes(Context context)<a name="line.275"></a> -<span class="sourceLineNo">276</span> throws IOException, InterruptedException {<a name="line.276"></a> -<span class="sourceLineNo">277</span> targetHasher.finishBatch();<a name="line.277"></a> -<span class="sourceLineNo">278</span> context.getCounter(Counter.BATCHES).increment(1);<a name="line.278"></a> -<span class="sourceLineNo">279</span> if (targetHasher.getBatchSize() == 0) {<a name="line.279"></a> -<span class="sourceLineNo">280</span> context.getCounter(Counter.EMPTY_BATCHES).increment(1);<a name="line.280"></a> -<span class="sourceLineNo">281</span> }<a name="line.281"></a> -<span class="sourceLineNo">282</span> ImmutableBytesWritable targetHash = targetHasher.getBatchHash();<a name="line.282"></a> -<span class="sourceLineNo">283</span> if (targetHash.equals(currentSourceHash)) {<a name="line.283"></a> -<span class="sourceLineNo">284</span> context.getCounter(Counter.HASHES_MATCHED).increment(1);<a name="line.284"></a> -<span class="sourceLineNo">285</span> } else {<a name="line.285"></a> -<span class="sourceLineNo">286</span> context.getCounter(Counter.HASHES_NOT_MATCHED).increment(1);<a name="line.286"></a> -<span class="sourceLineNo">287</span><a name="line.287"></a> -<span class="sourceLineNo">288</span> ImmutableBytesWritable stopRow = nextSourceKey == null<a name="line.288"></a> -<span class="sourceLineNo">289</span> ? new ImmutableBytesWritable(sourceTableHash.stopRow)<a name="line.289"></a> -<span class="sourceLineNo">290</span> : nextSourceKey;<a name="line.290"></a> -<span class="sourceLineNo">291</span><a name="line.291"></a> -<span class="sourceLineNo">292</span> if (LOG.isDebugEnabled()) {<a name="line.292"></a> -<span class="sourceLineNo">293</span> LOG.debug("Hash mismatch. Key range: " + toHex(targetHasher.getBatchStartKey())<a name="line.293"></a> -<span class="sourceLineNo">294</span> + " to " + toHex(stopRow)<a name="line.294"></a> -<span class="sourceLineNo">295</span> + " sourceHash: " + toHex(currentSourceHash)<a name="line.295"></a> -<span class="sourceLineNo">296</span> + " targetHash: " + toHex(targetHash));<a name="line.296"></a> -<span class="sourceLineNo">297</span> }<a name="line.297"></a> -<span class="sourceLineNo">298</span><a name="line.298"></a> -<span class="sourceLineNo">299</span> syncRange(context, targetHasher.getBatchStartKey(), stopRow);<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> private static String toHex(ImmutableBytesWritable bytes) {<a name="line.302"></a> -<span class="sourceLineNo">303</span> return Bytes.toHex(bytes.get(), bytes.getOffset(), bytes.getLength());<a name="line.303"></a> -<span class="sourceLineNo">304</span> }<a name="line.304"></a> -<span class="sourceLineNo">305</span><a name="line.305"></a> -<span class="sourceLineNo">306</span> private static final CellScanner EMPTY_CELL_SCANNER<a name="line.306"></a> -<span class="sourceLineNo">307</span> = new CellScanner(Collections.<Result>emptyIterator());<a name="line.307"></a> +<span class="sourceLineNo">264</span> }<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> * If there is an open hash batch, complete it and sync if there are diffs.<a name="line.267"></a> +<span class="sourceLineNo">268</span> * Start a new batch, and seek to read the<a name="line.268"></a> +<span class="sourceLineNo">269</span> */<a name="line.269"></a> +<span class="sourceLineNo">270</span> private void moveToNextBatch(Context context) throws IOException, InterruptedException {<a name="line.270"></a> +<span class="sourceLineNo">271</span> if (targetHasher.isBatchStarted()) {<a name="line.271"></a> +<span class="sourceLineNo">272</span> finishBatchAndCompareHashes(context);<a name="line.272"></a> +<span class="sourceLineNo">273</span> }<a name="line.273"></a> +<span class="sourceLineNo">274</span> targetHasher.startBatch(nextSourceKey);<a name="line.274"></a> +<span class="sourceLineNo">275</span> currentSourceHash = sourceHashReader.getCurrentHash();<a name="line.275"></a> +<span class="sourceLineNo">276</span><a name="line.276"></a> +<span class="sourceLineNo">277</span> findNextKeyHashPair();<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> * Finish the currently open hash batch.<a name="line.281"></a> +<span class="sourceLineNo">282</span> * Compare the target hash to the given source hash.<a name="line.282"></a> +<span class="sourceLineNo">283</span> * If they do not match, then sync the covered key range.<a name="line.283"></a> +<span class="sourceLineNo">284</span> */<a name="line.284"></a> +<span class="sourceLineNo">285</span> private void finishBatchAndCompareHashes(Context context)<a name="line.285"></a> +<span class="sourceLineNo">286</span> throws IOException, InterruptedException {<a name="line.286"></a> +<span class="sourceLineNo">287</span> targetHasher.finishBatch();<a name="line.287"></a> +<span class="sourceLineNo">288</span> context.getCounter(Counter.BATCHES).increment(1);<a name="line.288"></a> +<span class="sourceLineNo">289</span> if (targetHasher.getBatchSize() == 0) {<a name="line.289"></a> +<span class="sourceLineNo">290</span> context.getCounter(Counter.EMPTY_BATCHES).increment(1);<a name="line.290"></a> +<span class="sourceLineNo">291</span> }<a name="line.291"></a> +<span class="sourceLineNo">292</span> ImmutableBytesWritable targetHash = targetHasher.getBatchHash();<a name="line.292"></a> +<span class="sourceLineNo">293</span> if (targetHash.equals(currentSourceHash)) {<a name="line.293"></a> +<span class="sourceLineNo">294</span> context.getCounter(Counter.HASHES_MATCHED).increment(1);<a name="line.294"></a> +<span class="sourceLineNo">295</span> } else {<a name="line.295"></a> +<span class="sourceLineNo">296</span> context.getCounter(Counter.HASHES_NOT_MATCHED).increment(1);<a name="line.296"></a> +<span class="sourceLineNo">297</span><a name="line.297"></a> +<span class="sourceLineNo">298</span> ImmutableBytesWritable stopRow = nextSourceKey == null<a name="line.298"></a> +<span class="sourceLineNo">299</span> ? new ImmutableBytesWritable(sourceTableHash.stopRow)<a name="line.299"></a> +<span class="sourceLineNo">300</span> : nextSourceKey;<a name="line.300"></a> +<span class="sourceLineNo">301</span><a name="line.301"></a> +<span class="sourceLineNo">302</span> if (LOG.isDebugEnabled()) {<a name="line.302"></a> +<span class="sourceLineNo">303</span> LOG.debug("Hash mismatch. Key range: " + toHex(targetHasher.getBatchStartKey())<a name="line.303"></a> +<span class="sourceLineNo">304</span> + " to " + toHex(stopRow)<a name="line.304"></a> +<span class="sourceLineNo">305</span> + " sourceHash: " + toHex(currentSourceHash)<a name="line.305"></a> +<span class="sourceLineNo">306</span> + " targetHash: " + toHex(targetHash));<a name="line.306"></a> +<span class="sourceLineNo">307</span> }<a name="line.307"></a> <span class="sourceLineNo">308</span><a name="line.308"></a> -<span class="sourceLineNo">309</span> /**<a name="line.309"></a> -<span class="sourceLineNo">310</span> * Rescan the given range directly from the source and target tables.<a name="line.310"></a> -<span class="sourceLineNo">311</span> * Count and log differences, and if this is not a dry run, output Puts and Deletes<a name="line.311"></a> -<span class="sourceLineNo">312</span> * to make the target table match the source table for this range<a name="line.312"></a> -<span class="sourceLineNo">313</span> */<a name="line.313"></a> -<span class="sourceLineNo">314</span> private void syncRange(Context context, ImmutableBytesWritable startRow,<a name="line.314"></a> -<span class="sourceLineNo">315</span> ImmutableBytesWritable stopRow) throws IOException, InterruptedException {<a name="line.315"></a> -<span class="sourceLineNo">316</span> Scan scan = sourceTableHash.initScan();<a name="line.316"></a> -<span class="sourceLineNo">317</span> scan.setStartRow(startRow.copyBytes());<a name="line.317"></a> -<span class="sourceLineNo">318</span> scan.setStopRow(stopRow.copyBytes());<a name="line.318"></a> -<span class="sourceLineNo">319</span><a name="line.319"></a> -<span class="sourceLineNo">320</span> ResultScanner sourceScanner = sourceTable.getScanner(scan);<a name="line.320"></a> -<span class="sourceLineNo">321</span> CellScanner sourceCells = new CellScanner(sourceScanner.iterator());<a name="line.321"></a> -<span class="sourceLineNo">322</span><a name="line.322"></a> -<span class="sourceLineNo">323</span> ResultScanner targetScanner = targetTable.getScanner(new Scan(scan));<a name="line.323"></a> -<span class="sourceLineNo">324</span> CellScanner targetCells = new CellScanner(targetScanner.iterator());<a name="line.324"></a> -<span class="sourceLineNo">325</span><a name="line.325"></a> -<span class="sourceLineNo">326</span> boolean rangeMatched = true;<a name="line.326"></a> -<span class="sourceLineNo">327</span> byte[] nextSourceRow = sourceCells.nextRow();<a name="line.327"></a> -<span class="sourceLineNo">328</span> byte[] nextTargetRow = targetCells.nextRow();<a name="line.328"></a> -<span class="sourceLineNo">329</span> while(nextSourceRow != null || nextTargetRow != null) {<a name="line.329"></a> -<span class="sourceLineNo">330</span> boolean rowMatched;<a name="line.330"></a> -<span class="sourceLineNo">331</span> int rowComparison = compareRowKeys(nextSourceRow, nextTargetRow);<a name="line.331"></a> -<span class="sourceLineNo">332</span> if (rowComparison < 0) {<a name="line.332"></a> -<span class="sourceLineNo">333</span> if (LOG.isInfoEnabled()) {<a name="line.333"></a> -<span class="sourceLineNo">334</span> LOG.info("Target missing row: " + Bytes.toHex(nextSourceRow));<a name="line.334"></a> -<span class="sourceLineNo">335</span> }<a name="line.335"></a> -<span class="sourceLineNo">336</span> context.getCounter(Counter.TARGETMISSINGROWS).increment(1);<a name="line.336"></a> -<span class="sourceLineNo">337</span><a name="line.337"></a> -<span class="sourceLineNo">338</span> rowMatched = syncRowCells(context, nextSourceRow, sourceCells, EMPTY_CELL_SCANNER);<a name="line.338"></a> -<span class="sourceLineNo">339</span> nextSourceRow = sourceCells.nextRow(); // advance only source to next row<a name="line.339"></a> -<span class="sourceLineNo">340</span> } else if (rowComparison > 0) {<a name="line.340"></a> -<span class="sourceLineNo">341</span> if (LOG.isInfoEnabled()) {<a name="line.341"></a> -<span class="sourceLineNo">342</span> LOG.info("Source missing row: " + Bytes.toHex(nextTargetRow));<a name="line.342"></a> -<span class="sourceLineNo">343</span> }<a name="line.343"></a> -<span class="sourceLineNo">344</span> context.getCounter(Counter.SOURCEMISSINGROWS).increment(1);<a name="line.344"></a> -<span class="sourceLineNo">345</span><a name="line.345"></a> -<span class="sourceLineNo">346</span> rowMatched = syncRowCells(context, nextTargetRow, EMPTY_CELL_SCANNER, targetCells);<a name="line.346"></a> -<span class="sourceLineNo">347</span> nextTargetRow = targetCells.nextRow(); // advance only target to next row<a name="line.347"></a> -<span class="sourceLineNo">348</span> } else {<a name="line.348"></a> -<span class="sourceLineNo">349</span> // current row is the same on both sides, compare cell by cell<a name="line.349"></a> -<span class="sourceLineNo">350</span> rowMatched = syncRowCells(context, nextSourceRow, sourceCells, targetCells);<a name="line.350"></a> -<span class="sourceLineNo">351</span> nextSourceRow = sourceCells.nextRow();<a name="line.351"></a> -<span class="sourceLineNo">352</span> nextTargetRow = targetCells.nextRow();<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> if (!rowMatched) {<a name="line.355"></a> -<span class="sourceLineNo">356</span> rangeMatched = false;<a name="line.356"></a> -<span class="sourceLineNo">357</span> }<a name="line.357"></a> -<span class="sourceLineNo">358</span> }<a name="line.358"></a> -<span class="sourceLineNo">359</span><a name="line.359"></a> -<span class="sourceLineNo">360</span> sourceScanner.close();<a name="line.360"></a> -<span class="sourceLineNo">361</span> targetScanner.close();<a name="line.361"></a> -<span class="sourceLineNo">362</span><a name="line.362"></a> -<span class="sourceLineNo">363</span> context.getCounter(rangeMatched ? Counter.RANGESMATCHED : Counter.RANGESNOTMATCHED)<a name="line.363"></a> -<span class="sourceLineNo">364</span> .increment(1);<a name="line.364"></a> -<span class="sourceLineNo">365</span> }<a name="line.365"></a> -<span class="sourceLineNo">366</span><a name="line.366"></a> -<span class="sourceLineNo">367</span> private static class CellScanner {<a name="line.367"></a> -<span class="sourceLineNo">368</span> private final Iterator<Result> results;<a name="line.368"></a> +<span class="sourceLineNo">309</span> syncRange(context, targetHasher.getBatchStartKey(), stopRow);<a name="line.309"></a> +<span class="sourceLineNo">310</span> }<a name="line.310"></a> +<span class="sourceLineNo">311</span> }<a name="line.311"></a> +<span class="sourceLineNo">312</span> private static String toHex(ImmutableBytesWritable bytes) {<a name="line.312"></a> +<span class="sourceLineNo">313</span> return Bytes.toHex(bytes.get(), bytes.getOffset(), bytes.getLength());<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> private static final CellScanner EMPTY_CELL_SCANNER<a name="line.316"></a> +<span class="sourceLineNo">317</span> = new CellScanner(Collections.<Result>emptyIterator());<a name="line.317"></a> +<span class="sourceLineNo">318</span><a name="line.318"></a> +<span class="sourceLineNo">319</span> /**<a name="line.319"></a> +<span class="sourceLineNo">320</span> * Rescan the given range directly from the source and target tables.<a name="line.320"></a> +<span class="sourceLineNo">321</span> * Count and log differences, and if this is not a dry run, output Puts and Deletes<a name="line.321"></a> +<span class="sourceLineNo">322</span> * to make the target table match the source table for this range<a name="line.322"></a> +<span class="sourceLineNo">323</span> */<a name="line.323"></a> +<span class="sourceLineNo">324</span> private void syncRange(Context context, ImmutableBytesWritable startRow,<a name="line.324"></a> +<span class="sourceLineNo">325</span> ImmutableBytesWritable stopRow) throws IOException, InterruptedException {<a name="line.325"></a> +<span class="sourceLineNo">326</span> Scan scan = sourceTableHash.initScan();<a name="line.326"></a> +<span class="sourceLineNo">327</span> scan.setStartRow(startRow.copyBytes());<a name="line.327"></a> +<span class="sourceLineNo">328</span> scan.setStopRow(stopRow.copyBytes());<a name="line.328"></a> +<span class="sourceLineNo">329</span><a name="line.329"></a> +<span class="sourceLineNo">330</span> ResultScanner sourceScanner = sourceTable.getScanner(scan);<a name="line.330"></a> +<span class="sourceLineNo">331</span> CellScanner sourceCells = new CellScanner(sourceScanner.iterator());<a name="line.331"></a> +<span class="sourceLineNo">332</span><a name="line.332"></a> +<span class="sourceLineNo">333</span> ResultScanner targetScanner = targetTable.getScanner(new Scan(scan));<a name="line.333"></a> +<span class="sourceLineNo">334</span> CellScanner targetCells = new CellScanner(targetScanner.iterator());<a name="line.334"></a> +<span class="sourceLineNo">335</span><a name="line.335"></a> +<span class="sourceLineNo">336</span> boolean rangeMatched = true;<a name="line.336"></a> +<span class="sourceLineNo">337</span> byte[] nextSourceRow = sourceCells.nextRow();<a name="line.337"></a> +<span class="sourceLineNo">338</span> byte[] nextTargetRow = targetCells.nextRow();<a name="line.338"></a> +<span class="sourceLineNo">339</span> while(nextSourceRow != null || nextTargetRow != null) {<a name="line.339"></a> +<span class="sourceLineNo">340</span> boolean rowMatched;<a name="line.340"></a> +<span class="sourceLineNo">341</span> int rowComparison = compareRowKeys(nextSourceRow, nextTargetRow);<a name="line.341"></a> +<span class="sourceLineNo">342</span> if (rowComparison < 0) {<a name="line.342"></a> +<span class="sourceLineNo">343</span> if (LOG.isInfoEnabled()) {<a name="line.343"></a> +<span class="sourceLineNo">344</span> LOG.info("Target missing row: " + Bytes.toHex(nextSourceRow));<a name="line.344"></a> +<span class="sourceLineNo">345</span> }<a name="line.345"></a> +<span class="sourceLineNo">346</span> context.getCounter(Counter.TARGETMISSINGROWS).increment(1);<a name="line.346"></a> +<span class="sourceLineNo">347</span><a name="line.347"></a> +<span class="sourceLineNo">348</span> rowMatched = syncRowCells(context, nextSourceRow, sourceCells, EMPTY_CELL_SCANNER);<a name="line.348"></a> +<span class="sourceLineNo">349</span> nextSourceRow = sourceCells.nextRow(); // advance only source to next row<a name="line.349"></a> +<span class="sourceLineNo">350</span> } else if (rowComparison > 0) {<a name="line.350"></a> +<span class="sourceLineNo">351</span> if (LOG.isInfoEnabled()) {<a name="line.351"></a> +<span class="sourceLineNo">352</span> LOG.info("Source missing row: " + Bytes.toHex(nextTargetRow));<a name="line.352"></a> +<span class="sourceLineNo">353</span> }<a name="line.353"></a> +<span class="sourceLineNo">354</span> context.getCounter(Counter.SOURCEMISSINGROWS).increment(1);<a name="line.354"></a> +<span class="sourceLineNo">355</span><a name="line.355"></a> +<span class="sourceLineNo">356</span> rowMatched = syncRowCells(context, nextTargetRow, EMPTY_CELL_SCANNER, targetCells);<a name="line.356"></a> +<span class="sourceLineNo">357</span> nextTargetRow = targetCells.nextRow(); // advance only target to next row<a name="line.357"></a> +<span class="sourceLineNo">358</span> } else {<a name="line.358"></a> +<span class="sourceLineNo">359</span> // current row is the same on both sides, compare cell by cell<a name="line.359"></a> +<span class="sourceLineNo">360</span> rowMatched = syncRowCells(context, nextSourceRow, sourceCells, targetCells);<a name="line.360"></a> +<span class="sourceLineNo">361</span> nextSourceRow = sourceCells.nextRow();<a name="line.361"></a> +<span class="sourceLineNo">362</span> nextTargetRow = targetCells.nextRow();<a name="line.362"></a> +<span class="sourceLineNo">363</span> }<a name="line.363"></a> +<span class="sourceLineNo">364</span><a name="line.364"></a> +<span class="sourceLineNo">365</span> if (!rowMatched) {<a name="line.365"></a> +<span class="sourceLineNo">366</span> rangeMatched = false;<a name="line.366"></a> +<span class="sourceLineNo">367</span> }<a name="line.367"></a> +<span class="sourceLineNo">368</span> }<a name="line.368"></a> <span class="sourceLineNo">369</span><a name="line.369"></a> -<span class="sourceLineNo">370</span> private byte[] currentRow;<a name="line.370"></a> -<span class="sourceLineNo">371</span> private Result currentRowResult;<a name="line.371"></a> -<span class="sourceLineNo">372</span> private int nextCellInRow;<a name="line.372"></a> -<span class="sourceLineNo">373</span><a name="line.373"></a> -<span class="sourceLineNo">374</span> private Result nextRowResult;<a name="line.374"></a> -<span class="sourceLineNo">375</span><a name="line.375"></a> -<span class="sourceLineNo">376</span> public CellScanner(Iterator<Result> results) {<a name="line.376"></a> -<span class="sourceLineNo">377</span> this.results = results;<a name="line.377"></a> -<span class="sourceLineNo">378</span> }<a name="line.378"></a> +<span class="sourceLineNo">370</span> sourceScanner.close();<a name="line.370"></a> +<span class="sourceLineNo">371</span> targetScanner.close();<a name="line.371"></a> +<span class="sourceLineNo">372</span><a name="line.372"></a> +<span class="sourceLineNo">373</span> context.getCounter(rangeMatched ? Counter.RANGESMATCHED : Counter.RANGESNOTMATCHED)<a name="line.373"></a> +<span class="sourceLineNo">374</span> .increment(1);<a name="line.374"></a> +<span class="sourceLineNo">375</span> }<a name="line.375"></a> +<span class="sourceLineNo">376</span><a name="line.376"></a> +<span class="sourceLineNo">377</span> private static class CellScanner {<a name="line.377"></a> +<span class="sourceLineNo">378</span> private final Iterator<Result> results;<a name="line.378"></a> <span class="sourceLineNo">379</span><a name="line.379"></a> -<span class="sourceLineNo">380</span> /**<a name="line.380"></a> -<span class="sourceLineNo">381</span> * Advance to the next row and return its row key.<a name="line.381"></a> -<span class="sourceLineNo">382</span> * Returns null iff there are no more rows.<a name="line.382"></a> -<span class="sourceLineNo">383</span> */<a name="line.383"></a> -<span class="sourceLineNo">384</span> public byte[] nextRow() {<a name="line.384"></a> -<span class="sourceLineNo">385</span> if (nextRowResult == null) {<a name="line.385"></a> -<span class="sourceLineNo">386</span> // no cached row - check scanner for more<a name="line.386"></a> -<span class="sourceLineNo">387</span> while (results.hasNext()) {<a name="line.387"></a> -<span class="sourceLineNo">388</span> nextRowResult = results.next();<a name="line.388"></a> -<span class="sourceLineNo">389</span> Cell nextCell = nextRowResult.rawCells()[0];<a name="line.389"></a> -<span class="sourceLineNo">390</span> if (currentRow == null<a name="line.390"></a> -<span class="sourceLineNo">391</span> || !Bytes.equals(currentRow, 0, currentRow.length, nextCell.getRowArray(),<a name="line.391"></a> -<span class="sourceLineNo">392</span> nextCell.getRowOffset(), nextCell.getRowLength())) {<a name="line.392"></a> -<span class="sourceLineNo">393</span> // found next row<a name="line.393"></a> -<span class="sourceLineNo">394</span> break;<a name="line.394"></a> -<span class="sourceLineNo">395</span> } else {<a name="line.395"></a> -<span class="sourceLineNo">396</span> // found another result from current row, keep scanning<a name="line.396"></a> -<span class="sourceLineNo">397</span> nextRowResult = null;<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><a name="line.400"></a> -<span class="sourceLineNo">401</span> if (nextRowResult == null) {<a name="line.401"></a> -<span class="sourceLineNo">402</span> // end of data, no more rows<a name="line.402"></a> -<span class="sourceLineNo">403</span> currentRowResult = null;<a name="line.403"></a> -<span class="sourceLineNo">404</span> currentRow = null;<a name="line.404"></a> -<span class="sourceLineNo">405</span> return null;<a name="line.405"></a> -<span class="sourceLineNo">406</span> }<a name="line.406"></a> -<span class="sourceLineNo">407</span> }<a name="line.407"></a> -<span class="sourceLineNo">408</span><a name="line.408"></a> -<span class="sourceLineNo">409</span> // advance to cached result for next row<a name="line.409"></a> -<span class="sourceLineNo">410</span> currentRowResult = nextRowResult;<a name="line.410"></a> -<span class="sourceLineNo">411</span> nextCellInRow = 0;<a name="line.411"></a> -<span class="sourceLineNo">412</span> currentRow = currentRowResult.getRow();<a name="line.412"></a> -<span class="sourceLineNo">413</span> nextRowResult = null;<a name="line.413"></a> -<span class="sourceLineNo">414</span> return currentRow;<a name="line.414"></a> -<span class="sourceLineNo">415</span> }<a name="line.415"></a> -<span class="sourceLineNo">416</span><a name="line.416"></a> -<span class="sourceLineNo">417</span> /**<a name="line.417"></a> -<span class="sourceLineNo">418</span> * Returns the next Cell in the current row or null iff none remain.<a name="line.418"></a> -<span class="sourceLineNo">419</span> */<a name="line.419"></a> -<span class="sourceLineNo">420</span> public Cell nextCellInRow() {<a name="line.420"></a> -<span class="sourceLineNo">421</span> if (currentRowResult == null) {<a name="line.421"></a> -<span class="sourceLineNo">422</span> // nothing left in current row<a name="line.422"></a> -<span class="sourceLineNo">423</span> return null;<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> Cell nextCell = currentRowResult.rawCells()[nextCellInRow];<a name="line.426"></a> -<span class="sourceLineNo">427</span> nextCellInRow++;<a name="line.427"></a> -<span class="sourceLineNo">428</span> if (nextCellInRow == currentRowResult.size()) {<a name="line.428"></a> -<span class="sourceLineNo">429</span> if (results.hasNext()) {<a name="line.429"></a> -<span class="sourceLineNo">430</span> Result result = results.next();<a name="line.430"></a> -<span class="sourceLineNo">431</span> Cell cell = result.rawCells()[0];<a name="line.431"></a> -<span class="sourceLineNo">432</span> if (Bytes.equals(currentRow, 0, currentRow.length, cell.getRowArray(),<a name="line.432"></a> -<span class="sourceLineNo">433</span> cell.getRowOffset(), cell.getRowLength())) {<a name="line.433"></a> -<span class="sourceLineNo">434</span> // result is part of current row<a name="line.434"></a> -<span class="sourceLineNo">435</span> currentRowResult = result;<a name="line.435"></a> -<span class="sourceLineNo">436</span> nextCellInRow = 0;<a name="line.436"></a> -<span class="sourceLineNo">437</span> } else {<a name="line.437"></a> -<span class="sourceLineNo">438</span> // result is part of next row, cache it<a name="line.438"></a> -<span class="sourceLineNo">439</span> nextRowResult = result;<a name="line.439"></a> -<span class="sourceLineNo">440</span> // current row is complete<a name="line.440"></a> -<span class="sourceLineNo">441</span> currentRowResult = null;<a name="line.441"></a> -<span class="sourceLineNo">442</span> }<a name="line.442"></a> -<span class="sourceLineNo">443</span> } else {<a name="line.443"></a> -<span class="sourceLineNo">444</span> // end of data<a name="line.444"></a> -<span class="sourceLineNo">445</span> currentRowResult = null;<a name="line.445"></a> -<span class="sourceLineNo">446</span> }<a name="line.446"></a> -<span class="sourceLineNo">447</span> }<a name="line.447"></a> -<span class="sourceLineNo">448</span> return nextCell;<a name="line.448"></a> -<span class="sourceLineNo">449</span> }<a name="line.449"></a> -<span class="sourceLineNo">450</span> }<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> * Compare the cells for the given row from the source and target tables.<a name="line.453"></a> -<span class="sourceLineNo">454</span> * Count and log any differences.<a name="line.454"></a> -<span class="sourceLineNo">455</span> * If not a dry run, output a Put and/or Delete needed to sync the target table<a name="line.455"></a> -<span class="sourceLineNo">456</span> * to match the source table.<a name="line.456"></a> -<span class="sourceLineNo">457</span> */<a name="line.457"></a> -<span class="sourceLineNo">458</span> private boolean syncRowCells(Context context, byte[] rowKey, CellScanner sourceCells,<a name="line.458"></a> -<span class="sourceLineNo">459</span> CellScanner targetCells) throws IOException, InterruptedException {<a name="line.459"></a> -<span class="sourceLineNo">460</span> Put put = null;<a name="line.460"></a> -<span class="sourceLineNo">461</span> Delete delete = null;<a name="line.461"></a> -<span class="sourceLineNo">462</span> long matchingCells = 0;<a name="line.462"></a> -<span class="sourceLineNo">463</span> boolean matchingRow = true;<a name="line.463"></a> -<span class="sourceLineNo">464</span> Cell sourceCell = sourceCells.nextCellInRow();<a name="line.464"></a> -<span class="sourceLineNo">465</span> Cell targetCell = targetCells.nextCellInRow();<a name="line.465"></a> -<span class="sourceLineNo">466</span> while (sourceCell != null || targetCell != null) {<a name="line.466"></a> -<span class="sourceLineNo">467</span><a name="line.467"></a> -<span class="sourceLineNo">468</span> int cellKeyComparison = compareCellKeysWithinRow(sourceCell, targetCell);<a name="line.468"></a> -<span class="sourceLineNo">469</span> if (cellKeyComparison < 0) {<a name="line.469"></a> -<span class="sourceLineNo">470</span> if (LOG.isDebugEnabled()) {<a name="line.470"></a> -<span class="sourceLineNo">471</span> LOG.debug("Target missing cell: " + sourceCell);<a name="line.471"></a> -<span class="sourceLineNo">472</span> }<a name="line.472"></a> -<span class="sourceLineNo">473</span> context.getCounter(Counter.TARGETMISSINGCELLS).increment(1);<a name="line.473"></a> -<span class="sourceLineNo">474</span> matchingRow = false;<a name="line.474"></a> -<span class="sourceLineNo">475</span><a name="line.475"></a> -<span class="sourceLineNo">476</span> if (!dryRun) {<a name="line.476"></a> -<span class="sourceLineNo">477</span> if (put == null) {<a name="line.477"></a> -<span class="sourceLineNo">478</span> put = new Put(rowKey);<a name="line.478"></a> -<span class="sourceLineNo">479</span> }<a name="line.479"></a> -<span class="sourceLineNo">480</span> put.add(sourceCell);<a name="line.480"></a> -<span class="sourceLineNo">481</span> }<a name="line.481"></a> -<span class="sourceLineNo">482</span><a name="line.482"></a> -<span class="sourceLineNo">483</span> sourceCell = sourceCells.nextCellInRow();<a name="line.483"></a> -<span class="sourceLineNo">484</span> } else if (cellKeyComparison > 0) {<a name="line.484"></a> -<span class="sourceLineNo">485</span> if (LOG.isDebugEnabled()) {<a name="line.485"></a> -<span class="sourceLineNo">486</span> LOG.debug("Source missing cell: " + targetCell);<a name="line.486"></a> -<span class="sourceLineNo">487</span> }<a name="line.487"></a> -<span class="sourceLineNo">488</span> context.getCounter(Counter.SOURCEMISSINGCELLS).increment(1);<a name="line.488"></a> -<span class="sourceLineNo">489</span> matchingRow = false;<a name="line.489"></a> -<span class="sourceLineNo">490</span><a name="line.490"></a> -<span class="sourceLineNo">491</span> if (!dryRun) {<a name="line.491"></a> -<span class="sourceLineNo">492</span> if (delete == null) {<a name="line.492"></a> -<span class="sourceLineNo">493</span> delete = new Delete(rowKey);<a name="line.493"></a> -<span class="sourceLineNo">494</span> }<a name="line.494"></a> -<span class="sourceLineNo">495</span> // add a tombstone to exactly match the target cell that is missing on the source<a name="line.495"></a> -<span class="sourceLineNo">496</span> delete.addColumn(CellUtil.cloneFamily(targetCell),<a name="line.496"></a> -<span class="sourceLineNo">497</span> CellUtil.cloneQualifier(targetCell), targetCell.getTimestamp());<a name="line.497"></a> -<span class="sourceLineNo">498</span> }<a name="line.498"></a> -<span class="sourceLineNo">499</span><a name="line.499"></a> -<span class="sourceLineNo">500</span> targetCell = targetCells.nextCellInRow();<a name="line.500"></a> -<span class="sourceLineNo">501</span> } else {<a name="line.501"></a> -<span class="sourceLineNo">502</span> // the cell keys are equal, now check values<a name="line.502"></a> -<span class="sourceLineNo">503</span> if (CellUtil.matchingValue(sourceCell, targetCell)) {<a name="line.503"></a> -<span class="sourceLineNo">504</span> matchingCells++;<a name="line.504"></a> -<span class="sourceLineNo">505</span> } else {<a name="line.505"></a> -<span class="sourceLineNo">506</span> if (LOG.isDebugEnabled()) {<a name="line.506"></a> -<span class="sourceLineNo">507</span> LOG.debug("Different values: ");<a name="line.507"></a> -<span class="sourceLineNo">508</span> LOG.debug(" source cell: " + sourceCell<a name="line.508"></a> -<span class="sourceLineNo">509</span> + " value: " + Bytes.toHex(sourceCell.getValueArray(),<a name="line.509"></a> -<span class="sourceLineNo">510</span> sourceCell.getValueOffset(), sourceCell.getValueLength()));<a name="line.510"></a> -<span class="sourceLineNo">511</span> LOG.debug(" target cell: " + targetCell<a name="line.511"></a> -<span class="sourceLineNo">512</span> + " value: " + Bytes.toHex(targetCell.getValueArray(),<a name="line.512"></a> -<span class="sourceLineNo">513</span> targetCell.getValueOffset(), targetCell.getValueLength()));<a name="line.513"></a> -<span class="sourceLineNo">514</span> }<a name="line.514"></a> -<span class="sourceLineNo">515</span> context.getCounter(Counter.DIFFERENTCELLVALUES).increment(1);<a name="line.515"></a> -<span class="sourceLineNo">516</span> matchingRow = false;<a name="line.516"></a> -<span class="sourceLineNo">517</span><a name="line.517"></a> -<span class="sourceLineNo">518</span> if (!dryRun) {<a name="line.518"></a> -<span class="sourceLineNo">519</span> // overwrite target cell<a name="line.519"></a> -<span class="sourceLineNo">520</span> if (put == null) {<a name="line.520"></a> -<span class="sourceLineNo">521</span> put = new Put(rowKey);<a name="line.521"></a> -<span class="sourceLineNo">522</span> }<a name="line.522"></a> -<span class="sourceLineNo">523</span> put.add(sourceCell);<a name="line.523"></a> +<span class="sourceLineNo">380</span> private byte[] currentRow;<a name="line.380"></a> +<span class="sourceLineNo">381</span> private Result currentRowResult;<a name="line.381"></a> +<span class="sourceLineNo">382</span> private int nextCellInRow;<a name="line.382"></a> +<span class="sourceLineNo">383</span><a name="line.383"></a> +<span class="sourceLineNo">384</span> private Result nextRowResult;<a name="line.384"></a> +<span class="sourceLineNo">385</span><a name="line.385"></a> +<span class="sourceLineNo">386</span> public CellScanner(Iterator<Result> results) {<a name="line.386"></a> +<span class="sourceLineNo">387</span> this.results = results;<a name="line.387"></a> +<span class="sourceLineNo">388</span> }<a name="line.388"></a> +<span class="sourceLineNo">389</span><a name="line.389"></a> +<span class="sourceLineNo">390</span> /**<a name="line.390"></a> +<span class="sourceLineNo">391</span> * Advance to the next row and return its row key.<a name="line.391"></a> +<span class="sourceLineNo">392</span> * Returns null iff there are no more rows.<a name="line.392"></a> +<span class="sourceLineNo">393</span> */<a name="line.393"></a> +<span class="sourceLineNo">394</span> public byte[] nextRow() {<a name="line.394"></a> +<span class="sourceLineNo">395</span> if (nextRowResult == null) {<a name="line.395"></a> +<span class="sourceLineNo">396</span> // no cached row - check scanner for more<a name="line.396"></a> +<span class="sourceLineNo">397</span> while (results.hasNext()) {<a name="line.397"></a> +<span class="sourceLineNo">398</span> nextRowResult = results.next();<a name="line.398"></a> +<span class="sourceLineNo">399</span> Cell nextCell = nextRowResult.rawCells()[0];<a name="line.399"></a> +<span class="sourceLineNo">400</span> if (currentRow == null<a name="line.400"></a> +<span class="sourceLineNo">401</span> || !Bytes.equals(currentRow, 0, currentRow.length, nextCell.getRowArray(),<a name="line.401"></a> +<span class="sourceLineNo">402</span> nextCell.getRowOffset(), nextCell.getRowLength())) {<a name="line.402"></a> +<span class="sourceLineNo">403</span> // found next row<a name="line.403"></a> +<span class="sourceLineNo">404</span> break;<a name="line.404"></a> +<span class="sourceLineNo">405</span> } else {<a name="line.405"></a> +<span class="sourceLineNo">406</span> // found another result from current row, keep scanning<a name="line.406"></a> +<span class="sourceLineNo">407</span> nextRowResult = null;<a name="line.407"></a> +<span class="sourceLineNo">408</span> }<a name="line.408"></a> +<span class="sourceLineNo">409</span> }<a name="line.409"></a> +<span class="sourceLineNo">410</span><a name="line.410"></a> +<span class="sourceLineNo">411</span> if (nextRowResult == null) {<a name="line.411"></a> +<span class="sourceLineNo">412</span> // end of data, no more rows<a name="line.412"></a> +<span class="sourceLineNo">413</span> currentRowResult = null;<a name="line.413"></a> +<span class="sourceLineNo">414</span> currentRow = null;<a name="line.414"></a> +<span class="sourceLineNo">415</span> return null;<a name="line.415"></a> +<span class="sourceLineNo">416</span> }<a name="line.416"></a> +<span class="sourceLineNo">417</span> }<a name="line.417"></a> +<span class="sourceLineNo">418</span><a name="line.418"></a> +<span class="sourceLineNo">419</span> // advance to cached result for next row<a name="line.419"></a> +<span class="sourceLineNo">420</span> currentRowResult = nextRowResult;<a name="line.420"></a> +<span class="sourceLineNo">421</span> nextCellInRow = 0;<a name="line.421"></a> +<span class="sourceLineNo">422</span> currentRow = currentRowResult.getRow();<a name="line.422"></a> +<span class="sourceLineNo">423</span> nextRowResult = null;<a name="line.423"></a> +<span class="sourceLineNo">424</span> return currentRow;<a name="line.424"></a> +<span class="sourceLineNo">425</span> }<a name="line.425"></a> +<span class="sourceLineNo">426</span><a name="line.426"></a> +<span class="sourceLineNo">427</span> /**<a name="line.427"></a> +<span class="sourceLineNo">428</span> * Returns the next Cell in the current row or null iff none remain.<a name="line.428"></a> +<span class="sourceLineNo">429</span> */<a name="line.429"></a> +<span class="sourceLineNo">430</span> public Cell nextCellInRow() {<a name="line.430"></a> +<span class="sourceLineNo">431</span> if (currentRowResult == null) {<a name="line.431"></a> +<span class="sourceLineNo">432</span> // nothing left in current row<a name="line.432"></a> +<span class="sourceLineNo">433</span> return null;<a name="line.433"></a> +<span class="sourceLineNo">434</span> }<a name="line.434"></a> +<span class="sourceLineNo">435</span><a name="line.435"></a> +<span class="sourceLineNo">436</span> Cell nextCell = currentRowResult.rawCells()[nextCellInRow];<a name="line.436"></a> +<span class="sourceLineNo">437</span> nextCellInRow++;<a name="line.437"></a> +<span class="sourceLineNo">438</span> if (nextCellInRow == currentRowResult.size()) {<a name="line.438"></a> +<span class="sourceLineNo">439</span> if (results.hasNext()) {<a name="line.439"></a> +<span class="sourceLineNo">440</span> Result result = results.next();<a name="line.440"></a> +<span class="sourceLineNo">441</span> Cell cell = result.rawCells()[0];<a name="line.441"></a> +<span class="sourceLineNo">442</span> if (Bytes.equals(currentRow, 0, currentRow.length, cell.getRowArray(),<a name="line.442"></a> +<span class="sourceLineNo">443</span> cell.getRowOffset(), cell.getRowLength())) {<a name="line.443"></a> +<span class="sourceLineNo">444</span> // result is part of current row<a name="line.444"></a> +<span class="sourceLineNo">445</span> currentRowResult = result;<a name="line.445"></a> +<span class="sourceLineNo">446</span> nextCellInRow = 0;<a name="line.446"></a> +<span class="sourceLineNo">447</span> } else {<a name="line.447"></a> +<span class="sourceLineNo">448</span> // result is part of next row, cache it<a name="line.448"></a> +<span class="sourceLineNo">449</span> nextRowResult = result;<a name="line.449"></a> +<span class="sourceLineNo">450</span> // current row is complete<a name="line.450"></a> +<span class="sourceLineNo">451</span> currentRowResult = null;<a name="line.451"></a> +<span class="sourceLineNo">452</span> }<a name="line.452"></a> +<span class="sourceLineNo">453</span> } else {<a name="line.453"></a> +<span class="sourceLineNo">454</span> // end of data<a name="line.454"></a> +<span class="sourceLineNo">455</span> currentRowResult = null;<a name="line.455"></a> +<span class="sourceLineNo">456</span> }<a name="line.456"></a> +<span class="sourceLineNo">457</span> }<a name="line.457"></a> +<span class="sourceLineNo">458</span> return nextCell;<a name="line.458"></a> +<span class="sourceLineNo">459</span> }<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> * Compare the cells for the given row from the source and target tables.<a name="line.463"></a> +<span class="sourceLineNo">464</span> * Count and log any differences.<a name="line.464"></a> +<span class="sourceLineNo">465</span> * If not a dry run, output a Put and/or Delete needed to sync the target table<a name="line.465"></a> +<span class="sourceLineNo">466</span> * to match the source table.<a name="line.466"></a> +<span class="sourceLineNo">467</span> */<a name="line.467"></a> +<span class="sourceLineNo">468</span> private boolean syncRowCells(Context context, byte[] rowKey, CellScanner sourceCells,<a name="line.468"></a> +<span class="sourceLineNo">469</span> CellScanner targetCells) throws IOException, InterruptedException {<a name="line.469"></a> +<span class="sourceLineNo">470</span> Put put = null;<a name="line.470"></a> +<span class="sourceLineNo">471</span> Delete delete = null;<a name="line.471"></a> +<span class="sourceLineNo">472</span> long matchingCells = 0;<a name="line.472"></a> +<span class="sourceLineNo">473</span> boolean matchingRow = true;<a name="line.473"></a> +<span class="sourceLineNo">474</span> Cell sourceCell = sourceCells.nextCellInRow();<a name="line.474"></a> +<span class="sourceLineNo">475</span> Cell targetCell = targetCells.nextCellInRow();<a name="line.475"></a> +<span class="sourceLineNo">476</span> while (sourceCell != null || targetCell != null) {<a name="line.476"></a> +<span class="sourceLineNo">477</span><a name="line.477"></a> +<span class="sourceLineNo">478</span> int cellKeyComparison = compareCellKeysWithinRow(sourceCell, targetCell);<a name="line.478"></a> +<span class="sourceLineNo">479</span> if (cellKeyComparison < 0) {<a name="line.479"></a> +<span class="sourceLineNo">480</span> if (LOG.isDebugEnabled()) {<a name="line.480"></a> +<span class="sourceLineNo">481</span> LOG.debug("Target missing cell: " + sourceCell);<a name="line.481"></a> +<span class="sourceLineNo">482</span> }<a name="line.482"></a> +<span class="sourceLineNo">483</span> context.getCounter(Counter.TARGETMISSINGCELLS).increment(1);<a name="line.483"></a> +<span class="sourceLineNo">484</span> matchingRow = false;<a name="line.484"></a> +<span class="sourceLineNo">485</span><a name="line.485"></a> +<span class="sourceLineNo">486</span> if (!dryRun && doPuts) {<a name="line.486"></a> +<span class="sourceLineNo">487</span> if (put == null) {<a name="line.487"></a> +<span class="sourceLineNo">488</span> put = new Put(rowKey);<a name="line.488"></a> +<span class="sourceLineNo">489</span> }<a name="line.489"></a> +<span class="sourceLineNo">490</span> put.add(sourceCell);<a name="line.490"></a> +<span class="sourceLineNo">491</span> }<a name="line.491"></a> +<span class="sourceLineNo">492</span><a name="line.492"></a> +<span class="sourceLineNo">493</span> sourceCell = sourceCells.nextCellInRow();<a name="line.493"></a> +<span class="sourceLineNo">494</span> } else if (cellKeyComparison > 0) {<a name="line.494"></a> +<span class="sourceLineNo">495</span> if (LOG.isDebugEnabled()) {<a name="line.495"></a> +<span class="sourceLineNo">496</span> LOG.debug("Source missing ce
<TRUNCATED>
