hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From git-site-r...@apache.org
Subject [21/51] [partial] hbase-site git commit: Published site at .
Date Wed, 20 Dec 2017 15:18:31 GMT
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/d0f1a9f6/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.html
index 3c6311d..ec1b7f9 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/SyncTable.html
@@ -178,619 +178,620 @@
 <span class="sourceLineNo">170</span><a name="line.170"></a>
 <span class="sourceLineNo">171</span>    Throwable mapperException;<a name="line.171"></a>
 <span class="sourceLineNo">172</span><a name="line.172"></a>
-<span class="sourceLineNo">173</span>    public static enum Counter {BATCHES, HASHES_MATCHED, HASHES_NOT_MATCHED, SOURCEMISSINGROWS,<a name="line.173"></a>
+<span class="sourceLineNo">173</span>    public static enum Counter { BATCHES, HASHES_MATCHED, HASHES_NOT_MATCHED, SOURCEMISSINGROWS,<a name="line.173"></a>
 <span class="sourceLineNo">174</span>      SOURCEMISSINGCELLS, TARGETMISSINGROWS, TARGETMISSINGCELLS, ROWSWITHDIFFS, DIFFERENTCELLVALUES,<a name="line.174"></a>
-<span class="sourceLineNo">175</span>      MATCHINGROWS, MATCHINGCELLS, EMPTY_BATCHES, RANGESMATCHED, RANGESNOTMATCHED};<a name="line.175"></a>
-<span class="sourceLineNo">176</span><a name="line.176"></a>
-<span class="sourceLineNo">177</span>    @Override<a name="line.177"></a>
-<span class="sourceLineNo">178</span>    protected void setup(Context context) throws IOException {<a name="line.178"></a>
-<span class="sourceLineNo">179</span><a name="line.179"></a>
-<span class="sourceLineNo">180</span>      Configuration conf = context.getConfiguration();<a name="line.180"></a>
-<span class="sourceLineNo">181</span>      sourceHashDir = new Path(conf.get(SOURCE_HASH_DIR_CONF_KEY));<a name="line.181"></a>
-<span class="sourceLineNo">182</span>      sourceConnection = openConnection(conf, SOURCE_ZK_CLUSTER_CONF_KEY, null);<a name="line.182"></a>
-<span class="sourceLineNo">183</span>      targetConnection = openConnection(conf, TARGET_ZK_CLUSTER_CONF_KEY,<a name="line.183"></a>
-<span class="sourceLineNo">184</span>          TableOutputFormat.OUTPUT_CONF_PREFIX);<a name="line.184"></a>
-<span class="sourceLineNo">185</span>      sourceTable = openTable(sourceConnection, conf, SOURCE_TABLE_CONF_KEY);<a name="line.185"></a>
-<span class="sourceLineNo">186</span>      targetTable = openTable(targetConnection, conf, TARGET_TABLE_CONF_KEY);<a name="line.186"></a>
-<span class="sourceLineNo">187</span>      dryRun = conf.getBoolean(SOURCE_TABLE_CONF_KEY, false);<a name="line.187"></a>
-<span class="sourceLineNo">188</span><a name="line.188"></a>
-<span class="sourceLineNo">189</span>      sourceTableHash = HashTable.TableHash.read(conf, sourceHashDir);<a name="line.189"></a>
-<span class="sourceLineNo">190</span>      LOG.info("Read source hash manifest: " + sourceTableHash);<a name="line.190"></a>
-<span class="sourceLineNo">191</span>      LOG.info("Read " + sourceTableHash.partitions.size() + " partition keys");<a name="line.191"></a>
-<span class="sourceLineNo">192</span><a name="line.192"></a>
-<span class="sourceLineNo">193</span>      TableSplit split = (TableSplit) context.getInputSplit();<a name="line.193"></a>
-<span class="sourceLineNo">194</span>      ImmutableBytesWritable splitStartKey = new ImmutableBytesWritable(split.getStartRow());<a name="line.194"></a>
-<span class="sourceLineNo">195</span><a name="line.195"></a>
-<span class="sourceLineNo">196</span>      sourceHashReader = sourceTableHash.newReader(conf, splitStartKey);<a name="line.196"></a>
-<span class="sourceLineNo">197</span>      findNextKeyHashPair();<a name="line.197"></a>
-<span class="sourceLineNo">198</span><a name="line.198"></a>
-<span class="sourceLineNo">199</span>      // create a hasher, but don't start it right away<a name="line.199"></a>
-<span class="sourceLineNo">200</span>      // instead, find the first hash batch at or after the start row<a name="line.200"></a>
-<span class="sourceLineNo">201</span>      // and skip any rows that come before.  they will be caught by the previous task<a name="line.201"></a>
-<span class="sourceLineNo">202</span>      targetHasher = new HashTable.ResultHasher();<a name="line.202"></a>
-<span class="sourceLineNo">203</span>    }<a name="line.203"></a>
-<span class="sourceLineNo">204</span><a name="line.204"></a>
-<span class="sourceLineNo">205</span>    private static Connection openConnection(Configuration conf, String zkClusterConfKey,<a name="line.205"></a>
-<span class="sourceLineNo">206</span>                                             String configPrefix)<a name="line.206"></a>
-<span class="sourceLineNo">207</span>      throws IOException {<a name="line.207"></a>
-<span class="sourceLineNo">208</span>        String zkCluster = conf.get(zkClusterConfKey);<a name="line.208"></a>
-<span class="sourceLineNo">209</span>        Configuration clusterConf = HBaseConfiguration.createClusterConf(conf,<a name="line.209"></a>
-<span class="sourceLineNo">210</span>            zkCluster, configPrefix);<a name="line.210"></a>
-<span class="sourceLineNo">211</span>        return ConnectionFactory.createConnection(clusterConf);<a name="line.211"></a>
-<span class="sourceLineNo">212</span>    }<a name="line.212"></a>
-<span class="sourceLineNo">213</span><a name="line.213"></a>
-<span class="sourceLineNo">214</span>    private static Table openTable(Connection connection, Configuration conf,<a name="line.214"></a>
-<span class="sourceLineNo">215</span>        String tableNameConfKey) throws IOException {<a name="line.215"></a>
-<span class="sourceLineNo">216</span>      return connection.getTable(TableName.valueOf(conf.get(tableNameConfKey)));<a name="line.216"></a>
-<span class="sourceLineNo">217</span>    }<a name="line.217"></a>
-<span class="sourceLineNo">218</span><a name="line.218"></a>
-<span class="sourceLineNo">219</span>    /**<a name="line.219"></a>
-<span class="sourceLineNo">220</span>     * Attempt to read the next source key/hash pair.<a name="line.220"></a>
-<span class="sourceLineNo">221</span>     * If there are no more, set nextSourceKey to null<a name="line.221"></a>
-<span class="sourceLineNo">222</span>     */<a name="line.222"></a>
-<span class="sourceLineNo">223</span>    private void findNextKeyHashPair() throws IOException {<a name="line.223"></a>
-<span class="sourceLineNo">224</span>      boolean hasNext = sourceHashReader.next();<a name="line.224"></a>
-<span class="sourceLineNo">225</span>      if (hasNext) {<a name="line.225"></a>
-<span class="sourceLineNo">226</span>        nextSourceKey = sourceHashReader.getCurrentKey();<a name="line.226"></a>
-<span class="sourceLineNo">227</span>      } else {<a name="line.227"></a>
-<span class="sourceLineNo">228</span>        // no more keys - last hash goes to the end<a name="line.228"></a>
-<span class="sourceLineNo">229</span>        nextSourceKey = null;<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><a name="line.232"></a>
-<span class="sourceLineNo">233</span>    @Override<a name="line.233"></a>
-<span class="sourceLineNo">234</span>    protected void map(ImmutableBytesWritable key, Result value, Context context)<a name="line.234"></a>
-<span class="sourceLineNo">235</span>        throws IOException, InterruptedException {<a name="line.235"></a>
-<span class="sourceLineNo">236</span>      try {<a name="line.236"></a>
-<span class="sourceLineNo">237</span>        // first, finish any hash batches that end before the scanned row<a name="line.237"></a>
-<span class="sourceLineNo">238</span>        while (nextSourceKey != null &amp;&amp; key.compareTo(nextSourceKey) &gt;= 0) {<a name="line.238"></a>
-<span class="sourceLineNo">239</span>          moveToNextBatch(context);<a name="line.239"></a>
-<span class="sourceLineNo">240</span>        }<a name="line.240"></a>
-<span class="sourceLineNo">241</span><a name="line.241"></a>
-<span class="sourceLineNo">242</span>        // next, add the scanned row (as long as we've reached the first batch)<a name="line.242"></a>
-<span class="sourceLineNo">243</span>        if (targetHasher.isBatchStarted()) {<a name="line.243"></a>
-<span class="sourceLineNo">244</span>          targetHasher.hashResult(value);<a name="line.244"></a>
-<span class="sourceLineNo">245</span>        }<a name="line.245"></a>
-<span class="sourceLineNo">246</span>      } catch (Throwable t) {<a name="line.246"></a>
-<span class="sourceLineNo">247</span>        mapperException = t;<a name="line.247"></a>
-<span class="sourceLineNo">248</span>        Throwables.propagateIfInstanceOf(t, IOException.class);<a name="line.248"></a>
-<span class="sourceLineNo">249</span>        Throwables.propagateIfInstanceOf(t, InterruptedException.class);<a name="line.249"></a>
-<span class="sourceLineNo">250</span>        Throwables.propagate(t);<a name="line.250"></a>
-<span class="sourceLineNo">251</span>      }<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>    /**<a name="line.254"></a>
-<span class="sourceLineNo">255</span>     * If there is an open hash batch, complete it and sync if there are diffs.<a name="line.255"></a>
-<span class="sourceLineNo">256</span>     * Start a new batch, and seek to read the<a name="line.256"></a>
-<span class="sourceLineNo">257</span>     */<a name="line.257"></a>
-<span class="sourceLineNo">258</span>    private void moveToNextBatch(Context context) throws IOException, InterruptedException {<a name="line.258"></a>
-<span class="sourceLineNo">259</span>      if (targetHasher.isBatchStarted()) {<a name="line.259"></a>
-<span class="sourceLineNo">260</span>        finishBatchAndCompareHashes(context);<a name="line.260"></a>
-<span class="sourceLineNo">261</span>      }<a name="line.261"></a>
-<span class="sourceLineNo">262</span>      targetHasher.startBatch(nextSourceKey);<a name="line.262"></a>
-<span class="sourceLineNo">263</span>      currentSourceHash = sourceHashReader.getCurrentHash();<a name="line.263"></a>
-<span class="sourceLineNo">264</span><a name="line.264"></a>
-<span class="sourceLineNo">265</span>      findNextKeyHashPair();<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>    /**<a name="line.268"></a>
-<span class="sourceLineNo">269</span>     * Finish the currently open hash batch.<a name="line.269"></a>
-<span class="sourceLineNo">270</span>     * Compare the target hash to the given source hash.<a name="line.270"></a>
-<span class="sourceLineNo">271</span>     * If they do not match, then sync the covered key range.<a name="line.271"></a>
-<span class="sourceLineNo">272</span>     */<a name="line.272"></a>
-<span class="sourceLineNo">273</span>    private void finishBatchAndCompareHashes(Context context)<a name="line.273"></a>
-<span class="sourceLineNo">274</span>        throws IOException, InterruptedException {<a name="line.274"></a>
-<span class="sourceLineNo">275</span>      targetHasher.finishBatch();<a name="line.275"></a>
-<span class="sourceLineNo">276</span>      context.getCounter(Counter.BATCHES).increment(1);<a name="line.276"></a>
-<span class="sourceLineNo">277</span>      if (targetHasher.getBatchSize() == 0) {<a name="line.277"></a>
-<span class="sourceLineNo">278</span>        context.getCounter(Counter.EMPTY_BATCHES).increment(1);<a name="line.278"></a>
-<span class="sourceLineNo">279</span>      }<a name="line.279"></a>
-<span class="sourceLineNo">280</span>      ImmutableBytesWritable targetHash = targetHasher.getBatchHash();<a name="line.280"></a>
-<span class="sourceLineNo">281</span>      if (targetHash.equals(currentSourceHash)) {<a name="line.281"></a>
-<span class="sourceLineNo">282</span>        context.getCounter(Counter.HASHES_MATCHED).increment(1);<a name="line.282"></a>
-<span class="sourceLineNo">283</span>      } else {<a name="line.283"></a>
-<span class="sourceLineNo">284</span>        context.getCounter(Counter.HASHES_NOT_MATCHED).increment(1);<a name="line.284"></a>
-<span class="sourceLineNo">285</span><a name="line.285"></a>
-<span class="sourceLineNo">286</span>        ImmutableBytesWritable stopRow = nextSourceKey == null<a name="line.286"></a>
-<span class="sourceLineNo">287</span>                                          ? new ImmutableBytesWritable(sourceTableHash.stopRow)<a name="line.287"></a>
-<span class="sourceLineNo">288</span>                                          : nextSourceKey;<a name="line.288"></a>
-<span class="sourceLineNo">289</span><a name="line.289"></a>
-<span class="sourceLineNo">290</span>        if (LOG.isDebugEnabled()) {<a name="line.290"></a>
-<span class="sourceLineNo">291</span>          LOG.debug("Hash mismatch.  Key range: " + toHex(targetHasher.getBatchStartKey())<a name="line.291"></a>
-<span class="sourceLineNo">292</span>              + " to " + toHex(stopRow)<a name="line.292"></a>
-<span class="sourceLineNo">293</span>              + " sourceHash: " + toHex(currentSourceHash)<a name="line.293"></a>
-<span class="sourceLineNo">294</span>              + " targetHash: " + toHex(targetHash));<a name="line.294"></a>
-<span class="sourceLineNo">295</span>        }<a name="line.295"></a>
-<span class="sourceLineNo">296</span><a name="line.296"></a>
-<span class="sourceLineNo">297</span>        syncRange(context, targetHasher.getBatchStartKey(), stopRow);<a name="line.297"></a>
-<span class="sourceLineNo">298</span>      }<a name="line.298"></a>
-<span class="sourceLineNo">299</span>    }<a name="line.299"></a>
-<span class="sourceLineNo">300</span>    private static String toHex(ImmutableBytesWritable bytes) {<a name="line.300"></a>
-<span class="sourceLineNo">301</span>      return Bytes.toHex(bytes.get(), bytes.getOffset(), bytes.getLength());<a name="line.301"></a>
-<span class="sourceLineNo">302</span>    }<a name="line.302"></a>
-<span class="sourceLineNo">303</span><a name="line.303"></a>
-<span class="sourceLineNo">304</span>    private static final CellScanner EMPTY_CELL_SCANNER<a name="line.304"></a>
-<span class="sourceLineNo">305</span>      = new CellScanner(Collections.&lt;Result&gt;emptyIterator());<a name="line.305"></a>
-<span class="sourceLineNo">306</span><a name="line.306"></a>
-<span class="sourceLineNo">307</span>    /**<a name="line.307"></a>
-<span class="sourceLineNo">308</span>     * Rescan the given range directly from the source and target tables.<a name="line.308"></a>
-<span class="sourceLineNo">309</span>     * Count and log differences, and if this is not a dry run, output Puts and Deletes<a name="line.309"></a>
-<span class="sourceLineNo">310</span>     * to make the target table match the source table for this range<a name="line.310"></a>
-<span class="sourceLineNo">311</span>     */<a name="line.311"></a>
-<span class="sourceLineNo">312</span>    private void syncRange(Context context, ImmutableBytesWritable startRow,<a name="line.312"></a>
-<span class="sourceLineNo">313</span>        ImmutableBytesWritable stopRow) throws IOException, InterruptedException {<a name="line.313"></a>
-<span class="sourceLineNo">314</span>      Scan scan = sourceTableHash.initScan();<a name="line.314"></a>
-<span class="sourceLineNo">315</span>      scan.setStartRow(startRow.copyBytes());<a name="line.315"></a>
-<span class="sourceLineNo">316</span>      scan.setStopRow(stopRow.copyBytes());<a name="line.316"></a>
-<span class="sourceLineNo">317</span><a name="line.317"></a>
-<span class="sourceLineNo">318</span>      ResultScanner sourceScanner = sourceTable.getScanner(scan);<a name="line.318"></a>
-<span class="sourceLineNo">319</span>      CellScanner sourceCells = new CellScanner(sourceScanner.iterator());<a name="line.319"></a>
-<span class="sourceLineNo">320</span><a name="line.320"></a>
-<span class="sourceLineNo">321</span>      ResultScanner targetScanner = targetTable.getScanner(new Scan(scan));<a name="line.321"></a>
-<span class="sourceLineNo">322</span>      CellScanner targetCells = new CellScanner(targetScanner.iterator());<a name="line.322"></a>
-<span class="sourceLineNo">323</span><a name="line.323"></a>
-<span class="sourceLineNo">324</span>      boolean rangeMatched = true;<a name="line.324"></a>
-<span class="sourceLineNo">325</span>      byte[] nextSourceRow = sourceCells.nextRow();<a name="line.325"></a>
-<span class="sourceLineNo">326</span>      byte[] nextTargetRow = targetCells.nextRow();<a name="line.326"></a>
-<span class="sourceLineNo">327</span>      while(nextSourceRow != null || nextTargetRow != null) {<a name="line.327"></a>
-<span class="sourceLineNo">328</span>        boolean rowMatched;<a name="line.328"></a>
-<span class="sourceLineNo">329</span>        int rowComparison = compareRowKeys(nextSourceRow, nextTargetRow);<a name="line.329"></a>
-<span class="sourceLineNo">330</span>        if (rowComparison &lt; 0) {<a name="line.330"></a>
-<span class="sourceLineNo">331</span>          if (LOG.isInfoEnabled()) {<a name="line.331"></a>
-<span class="sourceLineNo">332</span>            LOG.info("Target missing row: " + Bytes.toHex(nextSourceRow));<a name="line.332"></a>
-<span class="sourceLineNo">333</span>          }<a name="line.333"></a>
-<span class="sourceLineNo">334</span>          context.getCounter(Counter.TARGETMISSINGROWS).increment(1);<a name="line.334"></a>
-<span class="sourceLineNo">335</span><a name="line.335"></a>
-<span class="sourceLineNo">336</span>          rowMatched = syncRowCells(context, nextSourceRow, sourceCells, EMPTY_CELL_SCANNER);<a name="line.336"></a>
-<span class="sourceLineNo">337</span>          nextSourceRow = sourceCells.nextRow();  // advance only source to next row<a name="line.337"></a>
-<span class="sourceLineNo">338</span>        } else if (rowComparison &gt; 0) {<a name="line.338"></a>
-<span class="sourceLineNo">339</span>          if (LOG.isInfoEnabled()) {<a name="line.339"></a>
-<span class="sourceLineNo">340</span>            LOG.info("Source missing row: " + Bytes.toHex(nextTargetRow));<a name="line.340"></a>
-<span class="sourceLineNo">341</span>          }<a name="line.341"></a>
-<span class="sourceLineNo">342</span>          context.getCounter(Counter.SOURCEMISSINGROWS).increment(1);<a name="line.342"></a>
-<span class="sourceLineNo">343</span><a name="line.343"></a>
-<span class="sourceLineNo">344</span>          rowMatched = syncRowCells(context, nextTargetRow, EMPTY_CELL_SCANNER, targetCells);<a name="line.344"></a>
-<span class="sourceLineNo">345</span>          nextTargetRow = targetCells.nextRow();  // advance only target to next row<a name="line.345"></a>
-<span class="sourceLineNo">346</span>        } else {<a name="line.346"></a>
-<span class="sourceLineNo">347</span>          // current row is the same on both sides, compare cell by cell<a name="line.347"></a>
-<span class="sourceLineNo">348</span>          rowMatched = syncRowCells(context, nextSourceRow, sourceCells, targetCells);<a name="line.348"></a>
-<span class="sourceLineNo">349</span>          nextSourceRow = sourceCells.nextRow();<a name="line.349"></a>
-<span class="sourceLineNo">350</span>          nextTargetRow = targetCells.nextRow();<a name="line.350"></a>
-<span class="sourceLineNo">351</span>        }<a name="line.351"></a>
-<span class="sourceLineNo">352</span><a name="line.352"></a>
-<span class="sourceLineNo">353</span>        if (!rowMatched) {<a name="line.353"></a>
-<span class="sourceLineNo">354</span>          rangeMatched = false;<a name="line.354"></a>
-<span class="sourceLineNo">355</span>        }<a name="line.355"></a>
-<span class="sourceLineNo">356</span>      }<a name="line.356"></a>
-<span class="sourceLineNo">357</span><a name="line.357"></a>
-<span class="sourceLineNo">358</span>      sourceScanner.close();<a name="line.358"></a>
-<span class="sourceLineNo">359</span>      targetScanner.close();<a name="line.359"></a>
-<span class="sourceLineNo">360</span><a name="line.360"></a>
-<span class="sourceLineNo">361</span>      context.getCounter(rangeMatched ? Counter.RANGESMATCHED : Counter.RANGESNOTMATCHED)<a name="line.361"></a>
-<span class="sourceLineNo">362</span>        .increment(1);<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>    private static class CellScanner {<a name="line.365"></a>
-<span class="sourceLineNo">366</span>      private final Iterator&lt;Result&gt; results;<a name="line.366"></a>
-<span class="sourceLineNo">367</span><a name="line.367"></a>
-<span class="sourceLineNo">368</span>      private byte[] currentRow;<a name="line.368"></a>
-<span class="sourceLineNo">369</span>      private Result currentRowResult;<a name="line.369"></a>
-<span class="sourceLineNo">370</span>      private int nextCellInRow;<a name="line.370"></a>
-<span class="sourceLineNo">371</span><a name="line.371"></a>
-<span class="sourceLineNo">372</span>      private Result nextRowResult;<a name="line.372"></a>
-<span class="sourceLineNo">373</span><a name="line.373"></a>
-<span class="sourceLineNo">374</span>      public CellScanner(Iterator&lt;Result&gt; results) {<a name="line.374"></a>
-<span class="sourceLineNo">375</span>        this.results = results;<a name="line.375"></a>
-<span class="sourceLineNo">376</span>      }<a name="line.376"></a>
-<span class="sourceLineNo">377</span><a name="line.377"></a>
-<span class="sourceLineNo">378</span>      /**<a name="line.378"></a>
-<span class="sourceLineNo">379</span>       * Advance to the next row and return its row key.<a name="line.379"></a>
-<span class="sourceLineNo">380</span>       * Returns null iff there are no more rows.<a name="line.380"></a>
-<span class="sourceLineNo">381</span>       */<a name="line.381"></a>
-<span class="sourceLineNo">382</span>      public byte[] nextRow() {<a name="line.382"></a>
-<span class="sourceLineNo">383</span>        if (nextRowResult == null) {<a name="line.383"></a>
-<span class="sourceLineNo">384</span>          // no cached row - check scanner for more<a name="line.384"></a>
-<span class="sourceLineNo">385</span>          while (results.hasNext()) {<a name="line.385"></a>
-<span class="sourceLineNo">386</span>            nextRowResult = results.next();<a name="line.386"></a>
-<span class="sourceLineNo">387</span>            Cell nextCell = nextRowResult.rawCells()[0];<a name="line.387"></a>
-<span class="sourceLineNo">388</span>            if (currentRow == null<a name="line.388"></a>
-<span class="sourceLineNo">389</span>                || !Bytes.equals(currentRow, 0, currentRow.length, nextCell.getRowArray(),<a name="line.389"></a>
-<span class="sourceLineNo">390</span>                nextCell.getRowOffset(), nextCell.getRowLength())) {<a name="line.390"></a>
-<span class="sourceLineNo">391</span>              // found next row<a name="line.391"></a>
-<span class="sourceLineNo">392</span>              break;<a name="line.392"></a>
-<span class="sourceLineNo">393</span>            } else {<a name="line.393"></a>
-<span class="sourceLineNo">394</span>              // found another result from current row, keep scanning<a name="line.394"></a>
-<span class="sourceLineNo">395</span>              nextRowResult = null;<a name="line.395"></a>
-<span class="sourceLineNo">396</span>            }<a name="line.396"></a>
-<span class="sourceLineNo">397</span>          }<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span>          if (nextRowResult == null) {<a name="line.399"></a>
-<span class="sourceLineNo">400</span>            // end of data, no more rows<a name="line.400"></a>
-<span class="sourceLineNo">401</span>            currentRowResult = null;<a name="line.401"></a>
-<span class="sourceLineNo">402</span>            currentRow = null;<a name="line.402"></a>
-<span class="sourceLineNo">403</span>            return null;<a name="line.403"></a>
-<span class="sourceLineNo">404</span>          }<a name="line.404"></a>
-<span class="sourceLineNo">405</span>        }<a name="line.405"></a>
-<span class="sourceLineNo">406</span><a name="line.406"></a>
-<span class="sourceLineNo">407</span>        // advance to cached result for next row<a name="line.407"></a>
-<span class="sourceLineNo">408</span>        currentRowResult = nextRowResult;<a name="line.408"></a>
-<span class="sourceLineNo">409</span>        nextCellInRow = 0;<a name="line.409"></a>
-<span class="sourceLineNo">410</span>        currentRow = currentRowResult.getRow();<a name="line.410"></a>
-<span class="sourceLineNo">411</span>        nextRowResult = null;<a name="line.411"></a>
-<span class="sourceLineNo">412</span>        return currentRow;<a name="line.412"></a>
-<span class="sourceLineNo">413</span>      }<a name="line.413"></a>
-<span class="sourceLineNo">414</span><a name="line.414"></a>
-<span class="sourceLineNo">415</span>      /**<a name="line.415"></a>
-<span class="sourceLineNo">416</span>       * Returns the next Cell in the current row or null iff none remain.<a name="line.416"></a>
-<span class="sourceLineNo">417</span>       */<a name="line.417"></a>
-<span class="sourceLineNo">418</span>      public Cell nextCellInRow() {<a name="line.418"></a>
-<span class="sourceLineNo">419</span>        if (currentRowResult == null) {<a name="line.419"></a>
-<span class="sourceLineNo">420</span>          // nothing left in current row<a name="line.420"></a>
-<span class="sourceLineNo">421</span>          return null;<a name="line.421"></a>
-<span class="sourceLineNo">422</span>        }<a name="line.422"></a>
-<span class="sourceLineNo">423</span><a name="line.423"></a>
-<span class="sourceLineNo">424</span>        Cell nextCell = currentRowResult.rawCells()[nextCellInRow];<a name="line.424"></a>
-<span class="sourceLineNo">425</span>        nextCellInRow++;<a name="line.425"></a>
-<span class="sourceLineNo">426</span>        if (nextCellInRow == currentRowResult.size()) {<a name="line.426"></a>
-<span class="sourceLineNo">427</span>          if (results.hasNext()) {<a name="line.427"></a>
-<span class="sourceLineNo">428</span>            Result result = results.next();<a name="line.428"></a>
-<span class="sourceLineNo">429</span>            Cell cell = result.rawCells()[0];<a name="line.429"></a>
-<span class="sourceLineNo">430</span>            if (Bytes.equals(currentRow, 0, currentRow.length, cell.getRowArray(),<a name="line.430"></a>
-<span class="sourceLineNo">431</span>                cell.getRowOffset(), cell.getRowLength())) {<a name="line.431"></a>
-<span class="sourceLineNo">432</span>              // result is part of current row<a name="line.432"></a>
-<span class="sourceLineNo">433</span>              currentRowResult = result;<a name="line.433"></a>
-<span class="sourceLineNo">434</span>              nextCellInRow = 0;<a name="line.434"></a>
-<span class="sourceLineNo">435</span>            } else {<a name="line.435"></a>
-<span class="sourceLineNo">436</span>              // result is part of next row, cache it<a name="line.436"></a>
-<span class="sourceLineNo">437</span>              nextRowResult = result;<a name="line.437"></a>
-<span class="sourceLineNo">438</span>              // current row is complete<a name="line.438"></a>
-<span class="sourceLineNo">439</span>              currentRowResult = null;<a name="line.439"></a>
-<span class="sourceLineNo">440</span>            }<a name="line.440"></a>
-<span class="sourceLineNo">441</span>          } else {<a name="line.441"></a>
-<span class="sourceLineNo">442</span>            // end of data<a name="line.442"></a>
-<span class="sourceLineNo">443</span>            currentRowResult = null;<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>        return nextCell;<a name="line.446"></a>
-<span class="sourceLineNo">447</span>      }<a name="line.447"></a>
-<span class="sourceLineNo">448</span>    }<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>     * Compare the cells for the given row from the source and target tables.<a name="line.451"></a>
-<span class="sourceLineNo">452</span>     * Count and log any differences.<a name="line.452"></a>
-<span class="sourceLineNo">453</span>     * If not a dry run, output a Put and/or Delete needed to sync the target table<a name="line.453"></a>
-<span class="sourceLineNo">454</span>     * to match the source table.<a name="line.454"></a>
-<span class="sourceLineNo">455</span>     */<a name="line.455"></a>
-<span class="sourceLineNo">456</span>    private boolean syncRowCells(Context context, byte[] rowKey, CellScanner sourceCells,<a name="line.456"></a>
-<span class="sourceLineNo">457</span>        CellScanner targetCells) throws IOException, InterruptedException {<a name="line.457"></a>
-<span class="sourceLineNo">458</span>      Put put = null;<a name="line.458"></a>
-<span class="sourceLineNo">459</span>      Delete delete = null;<a name="line.459"></a>
-<span class="sourceLineNo">460</span>      long matchingCells = 0;<a name="line.460"></a>
-<span class="sourceLineNo">461</span>      boolean matchingRow = true;<a name="line.461"></a>
-<span class="sourceLineNo">462</span>      Cell sourceCell = sourceCells.nextCellInRow();<a name="line.462"></a>
-<span class="sourceLineNo">463</span>      Cell targetCell = targetCells.nextCellInRow();<a name="line.463"></a>
-<span class="sourceLineNo">464</span>      while (sourceCell != null || targetCell != null) {<a name="line.464"></a>
-<span class="sourceLineNo">465</span><a name="line.465"></a>
-<span class="sourceLineNo">466</span>        int cellKeyComparison = compareCellKeysWithinRow(sourceCell, targetCell);<a name="line.466"></a>
-<span class="sourceLineNo">467</span>        if (cellKeyComparison &lt; 0) {<a name="line.467"></a>
-<span class="sourceLineNo">468</span>          if (LOG.isDebugEnabled()) {<a name="line.468"></a>
-<span class="sourceLineNo">469</span>            LOG.debug("Target missing cell: " + sourceCell);<a name="line.469"></a>
-<span class="sourceLineNo">470</span>          }<a name="line.470"></a>
-<span class="sourceLineNo">471</span>          context.getCounter(Counter.TARGETMISSINGCELLS).increment(1);<a name="line.471"></a>
-<span class="sourceLineNo">472</span>          matchingRow = false;<a name="line.472"></a>
-<span class="sourceLineNo">473</span><a name="line.473"></a>
-<span class="sourceLineNo">474</span>          if (!dryRun) {<a name="line.474"></a>
-<span class="sourceLineNo">475</span>            if (put == null) {<a name="line.475"></a>
-<span class="sourceLineNo">476</span>              put = new Put(rowKey);<a name="line.476"></a>
-<span class="sourceLineNo">477</span>            }<a name="line.477"></a>
-<span class="sourceLineNo">478</span>            put.add(sourceCell);<a name="line.478"></a>
-<span class="sourceLineNo">479</span>          }<a name="line.479"></a>
-<span class="sourceLineNo">480</span><a name="line.480"></a>
-<span class="sourceLineNo">481</span>          sourceCell = sourceCells.nextCellInRow();<a name="line.481"></a>
-<span class="sourceLineNo">482</span>        } else if (cellKeyComparison &gt; 0) {<a name="line.482"></a>
-<span class="sourceLineNo">483</span>          if (LOG.isDebugEnabled()) {<a name="line.483"></a>
-<span class="sourceLineNo">484</span>            LOG.debug("Source missing cell: " + targetCell);<a name="line.484"></a>
-<span class="sourceLineNo">485</span>          }<a name="line.485"></a>
-<span class="sourceLineNo">486</span>          context.getCounter(Counter.SOURCEMISSINGCELLS).increment(1);<a name="line.486"></a>
-<span class="sourceLineNo">487</span>          matchingRow = false;<a name="line.487"></a>
-<span class="sourceLineNo">488</span><a name="line.488"></a>
-<span class="sourceLineNo">489</span>          if (!dryRun) {<a name="line.489"></a>
-<span class="sourceLineNo">490</span>            if (delete == null) {<a name="line.490"></a>
-<span class="sourceLineNo">491</span>              delete = new Delete(rowKey);<a name="line.491"></a>
-<span class="sourceLineNo">492</span>            }<a name="line.492"></a>
-<span class="sourceLineNo">493</span>            // add a tombstone to exactly match the target cell that is missing on the source<a name="line.493"></a>
-<span class="sourceLineNo">494</span>            delete.addColumn(CellUtil.cloneFamily(targetCell),<a name="line.494"></a>
-<span class="sourceLineNo">495</span>                CellUtil.cloneQualifier(targetCell), targetCell.getTimestamp());<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>          targetCell = targetCells.nextCellInRow();<a name="line.498"></a>
-<span class="sourceLineNo">499</span>        } else {<a name="line.499"></a>
-<span class="sourceLineNo">500</span>          // the cell keys are equal, now check values<a name="line.500"></a>
-<span class="sourceLineNo">501</span>          if (CellUtil.matchingValue(sourceCell, targetCell)) {<a name="line.501"></a>
-<span class="sourceLineNo">502</span>            matchingCells++;<a name="line.502"></a>
-<span class="sourceLineNo">503</span>          } else {<a name="line.503"></a>
-<span class="sourceLineNo">504</span>            if (LOG.isDebugEnabled()) {<a name="line.504"></a>
-<span class="sourceLineNo">505</span>              LOG.debug("Different values: ");<a name="line.505"></a>
-<span class="sourceLineNo">506</span>              LOG.debug("  source cell: " + sourceCell<a name="line.506"></a>
-<span class="sourceLineNo">507</span>                  + " value: " + Bytes.toHex(sourceCell.getValueArray(),<a name="line.507"></a>
-<span class="sourceLineNo">508</span>                      sourceCell.getValueOffset(), sourceCell.getValueLength()));<a name="line.508"></a>
-<span class="sourceLineNo">509</span>              LOG.debug("  target cell: " + targetCell<a name="line.509"></a>
-<span class="sourceLineNo">510</span>                  + " value: " + Bytes.toHex(targetCell.getValueArray(),<a name="line.510"></a>
-<span class="sourceLineNo">511</span>                      targetCell.getValueOffset(), targetCell.getValueLength()));<a name="line.511"></a>
-<span class="sourceLineNo">512</span>            }<a name="line.512"></a>
-<span class="sourceLineNo">513</span>            context.getCounter(Counter.DIFFERENTCELLVALUES).increment(1);<a name="line.513"></a>
-<span class="sourceLineNo">514</span>            matchingRow = false;<a name="line.514"></a>
-<span class="sourceLineNo">515</span><a name="line.515"></a>
-<span class="sourceLineNo">516</span>            if (!dryRun) {<a name="line.516"></a>
-<span class="sourceLineNo">517</span>              // overwrite target cell<a name="line.517"></a>
-<span class="sourceLineNo">518</span>              if (put == null) {<a name="line.518"></a>
-<span class="sourceLineNo">519</span>                put = new Put(rowKey);<a name="line.519"></a>
-<span class="sourceLineNo">520</span>              }<a name="line.520"></a>
-<span class="sourceLineNo">521</span>              put.add(sourceCell);<a name="line.521"></a>
-<span class="sourceLineNo">522</span>            }<a name="line.522"></a>
-<span class="sourceLineNo">523</span>          }<a name="line.523"></a>
-<span class="sourceLineNo">524</span>          sourceCell = sourceCells.nextCellInRow();<a name="line.524"></a>
-<span class="sourceLineNo">525</span>          targetCell = targetCells.nextCellInRow();<a name="line.525"></a>
-<span class="sourceLineNo">526</span>        }<a name="line.526"></a>
-<span class="sourceLineNo">527</span><a name="line.527"></a>
-<span class="sourceLineNo">528</span>        if (!dryRun &amp;&amp; sourceTableHash.scanBatch &gt; 0) {<a name="line.528"></a>
-<span class="sourceLineNo">529</span>          if (put != null &amp;&amp; put.size() &gt;= sourceTableHash.scanBatch) {<a name="line.529"></a>
-<span class="sourceLineNo">530</span>            context.write(new ImmutableBytesWritable(rowKey), put);<a name="line.530"></a>
-<span class="sourceLineNo">531</span>            put = null;<a name="line.531"></a>
-<span class="sourceLineNo">532</span>          }<a name="line.532"></a>
-<span class="sourceLineNo">533</span>          if (delete != null &amp;&amp; delete.size() &gt;= sourceTableHash.scanBatch) {<a name="line.533"></a>
-<span class="sourceLineNo">534</span>            context.write(new ImmutableBytesWritable(rowKey), delete);<a name="line.534"></a>
-<span class="sourceLineNo">535</span>            delete = null;<a name="line.535"></a>
-<span class="sourceLineNo">536</span>          }<a name="line.536"></a>
-<span class="sourceLineNo">537</span>        }<a name="line.537"></a>
-<span class="sourceLineNo">538</span>      }<a name="line.538"></a>
-<span class="sourceLineNo">539</span><a name="line.539"></a>
-<span class="sourceLineNo">540</span>      if (!dryRun) {<a name="line.540"></a>
-<span class="sourceLineNo">541</span>        if (put != null) {<a name="line.541"></a>
-<span class="sourceLineNo">542</span>          context.write(new ImmutableBytesWritable(rowKey), put);<a name="line.542"></a>
-<span class="sourceLineNo">543</span>        }<a name="line.543"></a>
-<span class="sourceLineNo">544</span>        if (delete != null) {<a name="line.544"></a>
-<span class="sourceLineNo">545</span>          context.write(new ImmutableBytesWritable(rowKey), delete);<a name="line.545"></a>
-<span class="sourceLineNo">546</span>        }<a name="line.546"></a>
-<span class="sourceLineNo">547</span>      }<a name="line.547"></a>
-<span class="sourceLineNo">548</span><a name="line.548"></a>
-<span class="sourceLineNo">549</span>      if (matchingCells &gt; 0) {<a name="line.549"></a>
-<span class="sourceLineNo">550</span>        context.getCounter(Counter.MATCHINGCELLS).increment(matchingCells);<a name="line.550"></a>
-<span class="sourceLineNo">551</span>      }<a name="line.551"></a>
-<span class="sourceLineNo">552</span>      if (matchingRow) {<a name="line.552"></a>
-<span class="sourceLineNo">553</span>        context.getCounter(Counter.MATCHINGROWS).increment(1);<a name="line.553"></a>
-<span class="sourceLineNo">554</span>        return true;<a name="line.554"></a>
-<span class="sourceLineNo">555</span>      } else {<a name="line.555"></a>
-<span class="sourceLineNo">556</span>        context.getCounter(Counter.ROWSWITHDIFFS).increment(1);<a name="line.556"></a>
-<span class="sourceLineNo">557</span>        return false;<a name="line.557"></a>
-<span class="sourceLineNo">558</span>      }<a name="line.558"></a>
-<span class="sourceLineNo">559</span>    }<a name="line.559"></a>
-<span class="sourceLineNo">560</span><a name="line.560"></a>
-<span class="sourceLineNo">561</span>    /**<a name="line.561"></a>
-<span class="sourceLineNo">562</span>     * Compare row keys of the given Result objects.<a name="line.562"></a>
-<span class="sourceLineNo">563</span>     * Nulls are after non-nulls<a name="line.563"></a>
-<span class="sourceLineNo">564</span>     */<a name="line.564"></a>
-<span class="sourceLineNo">565</span>    private static int compareRowKeys(byte[] r1, byte[] r2) {<a name="line.565"></a>
-<span class="sourceLineNo">566</span>      if (r1 == null) {<a name="line.566"></a>
-<span class="sourceLineNo">567</span>        return 1;  // source missing row<a name="line.567"></a>
-<span class="sourceLineNo">568</span>      } else if (r2 == null) {<a name="line.568"></a>
-<span class="sourceLineNo">569</span>        return -1; // target missing row<a name="line.569"></a>
-<span class="sourceLineNo">570</span>      } else {<a name="line.570"></a>
-<span class="sourceLineNo">571</span>        // Sync on no META tables only. We can directly do what CellComparator is doing inside.<a name="line.571"></a>
-<span class="sourceLineNo">572</span>        // Never the call going to MetaCellComparator.<a name="line.572"></a>
-<span class="sourceLineNo">573</span>        return Bytes.compareTo(r1, 0, r1.length, r2, 0, r2.length);<a name="line.573"></a>
-<span class="sourceLineNo">574</span>      }<a name="line.574"></a>
-<span class="sourceLineNo">575</span>    }<a name="line.575"></a>
-<span class="sourceLineNo">576</span><a name="line.576"></a>
-<span class="sourceLineNo">577</span>    /**<a name="line.577"></a>
-<span class="sourceLineNo">578</span>     * Compare families, qualifiers, and timestamps of the given Cells.<a name="line.578"></a>
-<span class="sourceLineNo">579</span>     * They are assumed to be of the same row.<a name="line.579"></a>
-<span class="sourceLineNo">580</span>     * Nulls are after non-nulls.<a name="line.580"></a>
-<span class="sourceLineNo">581</span>     */<a name="line.581"></a>
-<span class="sourceLineNo">582</span>     private static int compareCellKeysWithinRow(Cell c1, Cell c2) {<a name="line.582"></a>
-<span class="sourceLineNo">583</span>      if (c1 == null) {<a name="line.583"></a>
-<span class="sourceLineNo">584</span>        return 1; // source missing cell<a name="line.584"></a>
-<span class="sourceLineNo">585</span>      }<a name="line.585"></a>
-<span class="sourceLineNo">586</span>      if (c2 == null) {<a name="line.586"></a>
-<span class="sourceLineNo">587</span>        return -1; // target missing cell<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>      int result = CellComparator.getInstance().compareFamilies(c1, c2);<a name="line.590"></a>
-<span class="sourceLineNo">591</span>      if (result != 0) {<a name="line.591"></a>
-<span class="sourceLineNo">592</span>        return result;<a name="line.592"></a>
-<span class="sourceLineNo">593</span>      }<a name="line.593"></a>
-<span class="sourceLineNo">594</span><a name="line.594"></a>
-<span class="sourceLineNo">595</span>      result = CellComparator.getInstance().compareQualifiers(c1, c2);<a name="line.595"></a>
-<span class="sourceLineNo">596</span>      if (result != 0) {<a name="line.596"></a>
-<span class="sourceLineNo">597</span>        return result;<a name="line.597"></a>
-<span class="sourceLineNo">598</span>      }<a name="line.598"></a>
-<span class="sourceLineNo">599</span><a name="line.599"></a>
-<span class="sourceLineNo">600</span>      // note timestamp comparison is inverted - more recent cells first<a name="line.600"></a>
-<span class="sourceLineNo">601</span>      return CellComparator.getInstance().compareTimestamps(c1, c2);<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>    @Override<a name="line.604"></a>
-<span class="sourceLineNo">605</span>    protected void cleanup(Context context)<a name="line.605"></a>
-<span class="sourceLineNo">606</span>        throws IOException, InterruptedException {<a name="line.606"></a>
-<span class="sourceLineNo">607</span>      if (mapperException == null) {<a name="line.607"></a>
-<span class="sourceLineNo">608</span>        try {<a name="line.608"></a>
-<span class="sourceLineNo">609</span>          finishRemainingHashRanges(context);<a name="line.609"></a>
-<span class="sourceLineNo">610</span>        } catch (Throwable t) {<a name="line.610"></a>
-<span class="sourceLineNo">611</span>          mapperException = t;<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>      try {<a name="line.615"></a>
-<span class="sourceLineNo">616</span>        sourceTable.close();<a name="line.616"></a>
-<span class="sourceLineNo">617</span>        targetTable.close();<a name="line.617"></a>
-<span class="sourceLineNo">618</span>        sourceConnection.close();<a name="line.618"></a>
-<span class="sourceLineNo">619</span>        targetConnection.close();<a name="line.619"></a>
-<span class="sourceLineNo">620</span>      } catch (Throwable t) {<a name="line.620"></a>
-<span class="sourceLineNo">621</span>        if (mapperException == null) {<a name="line.621"></a>
-<span class="sourceLineNo">622</span>          mapperException = t;<a name="line.622"></a>
-<span class="sourceLineNo">623</span>        } else {<a name="line.623"></a>
-<span class="sourceLineNo">624</span>          LOG.error("Suppressing exception from closing tables", t);<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>      // propagate first exception<a name="line.628"></a>
-<span class="sourceLineNo">629</span>      if (mapperException != null) {<a name="line.629"></a>
-<span class="sourceLineNo">630</span>        Throwables.propagateIfInstanceOf(mapperException, IOException.class);<a name="line.630"></a>
-<span class="sourceLineNo">631</span>        Throwables.propagateIfInstanceOf(mapperException, InterruptedException.class);<a name="line.631"></a>
-<span class="sourceLineNo">632</span>        Throwables.propagate(mapperException);<a name="line.632"></a>
-<span class="sourceLineNo">633</span>      }<a name="line.633"></a>
-<span class="sourceLineNo">634</span>    }<a name="line.634"></a>
-<span class="sourceLineNo">635</span><a name="line.635"></a>
-<span class="sourceLineNo">636</span>    private void finishRemainingHashRanges(Context context) throws IOException,<a name="line.636"></a>
-<span class="sourceLineNo">637</span>        InterruptedException {<a name="line.637"></a>
-<span class="sourceLineNo">638</span>      TableSplit split = (TableSplit) context.getInputSplit();<a name="line.638"></a>
-<span class="sourceLineNo">639</span>      byte[] splitEndRow = split.getEndRow();<a name="line.639"></a>
-<span class="sourceLineNo">640</span>      boolean reachedEndOfTable = HashTable.isTableEndRow(splitEndRow);<a name="line.640"></a>
-<span class="sourceLineNo">641</span><a name="line.641"></a>
-<span class="sourceLineNo">642</span>      // if there are more hash batches that begin before the end of this split move to them<a name="line.642"></a>
-<span class="sourceLineNo">643</span>      while (nextSourceKey != null<a name="line.643"></a>
-<span class="sourceLineNo">644</span>          &amp;&amp; (nextSourceKey.compareTo(splitEndRow) &lt; 0 || reachedEndOfTable)) {<a name="line.644"></a>
-<span class="sourceLineNo">645</span>        moveToNextBatch(context);<a name="line.645"></a>
-<span class="sourceLineNo">646</span>      }<a name="line.646"></a>
-<span class="sourceLineNo">647</span><a name="line.647"></a>
-<span class="sourceLineNo">648</span>      if (targetHasher.isBatchStarted()) {<a name="line.648"></a>
-<span class="sourceLineNo">649</span>        // need to complete the final open hash batch<a name="line.649"></a>
-<span class="sourceLineNo">650</span><a name="line.650"></a>
-<span class="sourceLineNo">651</span>        if ((nextSourceKey != null &amp;&amp; nextSourceKey.compareTo(splitEndRow) &gt; 0)<a name="line.651"></a>
-<span class="sourceLineNo">652</span>              || (nextSourceKey == null &amp;&amp; !Bytes.equals(splitEndRow, sourceTableHash.stopRow))) {<a name="line.652"></a>
-<span class="sourceLineNo">653</span>          // the open hash range continues past the end of this region<a name="line.653"></a>
-<span class="sourceLineNo">654</span>          // add a scan to complete the current hash range<a name="line.654"></a>
-<span class="sourceLineNo">655</span>          Scan scan = sourceTableHash.initScan();<a name="line.655"></a>
-<span class="sourceLineNo">656</span>          scan.setStartRow(splitEndRow);<a name="line.656"></a>
-<span class="sourceLineNo">657</span>          if (nextSourceKey == null) {<a name="line.657"></a>
-<span class="sourceLineNo">658</span>            scan.setStopRow(sourceTableHash.stopRow);<a name="line.658"></a>
-<span class="sourceLineNo">659</span>          } else {<a name="line.659"></a>
-<span class="sourceLineNo">660</span>            scan.setStopRow(nextSourceKey.copyBytes());<a name="line.660"></a>
-<span class="sourceLineNo">661</span>          }<a name="line.661"></a>
-<span class="sourceLineNo">662</span><a name="line.662"></a>
-<span class="sourceLineNo">663</span>          ResultScanner targetScanner = null;<a name="line.663"></a>
-<span class="sourceLineNo">664</span>          try {<a name="line.664"></a>
-<span class="sourceLineNo">665</span>            targetScanner = targetTable.getScanner(scan);<a name="line.665"></a>
-<span class="sourceLineNo">666</span>            for (Result row : targetScanner) {<a name="line.666"></a>
-<span class="sourceLineNo">667</span>              targetHasher.hashResult(row);<a name="line.667"></a>
-<span class="sourceLineNo">668</span>            }<a name="line.668"></a>
-<span class="sourceLineNo">669</span>          } finally {<a name="line.669"></a>
-<span class="sourceLineNo">670</span>            if (targetScanner != null) {<a name="line.670"></a>
-<span class="sourceLineNo">671</span>              targetScanner.close();<a name="line.671"></a>
-<span class="sourceLineNo">672</span>            }<a name="line.672"></a>
-<span class="sourceLineNo">673</span>          }<a name="line.673"></a>
-<span class="sourceLineNo">674</span>        } // else current batch ends exactly at split end row<a name="line.674"></a>
-<span class="sourceLineNo">675</span><a name="line.675"></a>
-<span class="sourceLineNo">676</span>        finishBatchAndCompareHashes(context);<a name="line.676"></a>
-<span class="sourceLineNo">677</span>      }<a name="line.677"></a>
-<span class="sourceLineNo">678</span>    }<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>  private static final int NUM_ARGS = 3;<a name="line.681"></a>
-<span class="sourceLineNo">682</span>  private static void printUsage(final String errorMsg) {<a name="line.682"></a>
-<span class="sourceLineNo">683</span>    if (errorMsg != null &amp;&amp; errorMsg.length() &gt; 0) {<a name="line.683"></a>
-<span class="sourceLineNo">684</span>      System.err.println("ERROR: " + errorMsg);<a name="line.684"></a>
-<span class="sourceLineNo">685</span>      System.err.println();<a name="line.685"></a>
-<span class="sourceLineNo">686</span>    }<a name="line.686"></a>
-<span class="sourceLineNo">687</span>    System.err.println("Usage: SyncTable [options] &lt;sourcehashdir&gt; &lt;sourcetable&gt; &lt;targettable&gt;");<a name="line.687"></a>
-<span class="sourceLineNo">688</span>    System.err.println();<a name="line.688"></a>
-<span class="sourceLineNo">689</span>    System.err.println("Options:");<a name="line.689"></a>
-<span class="sourceLineNo">690</span><a name="line.690"></a>
-<span class="sourceLineNo">691</span>    System.err.println(" sourcezkcluster  ZK cluster key of the source table");<a name="line.691"></a>
-<span class="sourceLineNo">692</span>    System.err.println("                  (defaults to cluster in classpath's config)");<a name="line.692"></a>
-<span class="sourceLineNo">693</span>    System.err.println(" targetzkcluster  ZK cluster key of the target table");<a name="line.693"></a>
-<span class="sourceLineNo">694</span>    System.err.println("                  (defaults to cluster in classpath's config)");<a name="line.694"></a>
-<span class="sourceLineNo">695</span>    System.err.println(" dryrun           if true, output counters but no writes");<a name="line.695"></a>
-<span class="sourceLineNo">696</span>    System.err.println("                  (defaults to false)");<a name="line.696"></a>
-<span class="sourceLineNo">697</span>    System.err.println();<a name="line.697"></a>
-<span class="sourceLineNo">698</span>    System.err.println("Args:");<a name="line.698"></a>
-<span class="sourceLineNo">699</span>    System.err.println(" sourcehashdir    path to HashTable output dir for source table");<a name="line.699"></a>
-<span class="sourceLineNo">700</span>    System.err.println("                  (see org.apache.hadoop.hbase.mapreduce.HashTable)");<a name="line.700"></a>
-<span class="sourceLineNo">701</span>    System.err.println(" sourcetable      Name of the source table to sync from");<a name="line.701"></a>
-<span class="sourceLineNo">702</span>    System.err.println(" targettable      Name of the target table to sync to");<a name="line.702"></a>
-<span class="sourceLineNo">703</span>    System.err.println();<a name="line.703"></a>
-<span class="sourceLineNo">704</span>    System.err.println("Examples:");<a name="line.704"></a>
-<span class="sourceLineNo">705</span>    System.err.println(" For a dry run SyncTable of tableA from a remote source cluster");<a name="line.705"></a>
-<span class="sourceLineNo">706</span>    System.err.println(" to a local target cluster:");<a name="line.706"></a>
-<span class="sourceLineNo">707</span>    System.err.println(" $ hbase " +<a name="line.707"></a>
-<span class="sourceLineNo">708</span>        "org.apache.hadoop.hbase.mapreduce.SyncTable --dryrun=true"<a name="line.708"></a>
-<span class="sourceLineNo">709</span>        + " --sourcezkcluster=zk1.example.com,zk2.example.com,zk3.example.com:2181:/hbase"<a name="line.709"></a>
-<span class="sourceLineNo">710</span>        + " hdfs://nn:9000/hashes/tableA tableA tableA");<a name="line.710"></a>
-<span class="sourceLineNo">711</span>  }<a name="line.711"></a>
-<span class="sourceLineNo">712</span><a name="line.712"></a>
-<span class="sourceLineNo">713</span>  private boolean doCommandLine(final String[] args) {<a name="line.713"></a>
-<span class="sourceLineNo">714</span>    if (args.length &lt; NUM_ARGS) {<a name="line.714"></a>
-<span class="sourceLineNo">715</span>      printUsage(null);<a name="line.715"></a>
-<span class="sourceLineNo">716</span>      return false;<a name="line.716"></a>
-<span class="sourceLineNo">717</span>    }<a name="line.717"></a>
-<span class="sourceLineNo">718</span>    try {<a name="line.718"></a>
-<span class="sourceLineNo">719</span>      sourceHashDir = new Path(args[args.length - 3]);<a name="line.719"></a>
-<span class="sourceLineNo">720</span>      sourceTableName = args[args.length - 2];<a name="line.720"></a>
-<span class="sourceLineNo">721</span>      targetTableName = args[args.length - 1];<a name="line.721"></a>
-<span class="sourceLineNo">722</span><a name="line.722"></a>
-<span class="sourceLineNo">723</span>      for (int i = 0; i &lt; args.length - NUM_ARGS; i++) {<a name="line.723"></a>
-<span class="sourceLineNo">724</span>        String cmd = args[i];<a name="line.724"></a>
-<span class="sourceLineNo">725</span>        if (cmd.equals("-h") || cmd.startsWith("--h")) {<a name="line.725"></a>
-<span class="sourceLineNo">726</span>          printUsage(null);<a name="line.726"></a>
-<span class="sourceLineNo">727</span>          return false;<a name="line.727"></a>
-<span class="sourceLineNo">728</span>        }<a name="line.728"></a>
-<span class="sourceLineNo">729</span><a name="line.729"></a>
-<span class="sourceLineNo">730</span>        final String sourceZkClusterKey = "--sourcezkcluster=";<a name="line.730"></a>
-<span class="sourceLineNo">731</span>        if (cmd.startsWith(sourceZkClusterKey)) {<a name="line.731"></a>
-<span class="sourceLineNo">732</span>          sourceZkCluster = cmd.substring(sourceZkClusterKey.length());<a name="line.732"></a>
-<span class="sourceLineNo">733</span>          continue;<a name="line.733"></a>
-<span class="sourceLineNo">734</span>        }<a name="line.734"></a>
-<span class="sourceLineNo">735</span><a name="line.735"></a>
-<span class="sourceLineNo">736</span>        final String targetZkClusterKey = "--targetzkcluster=";<a name="line.736"></a>
-<span class="sourceLineNo">737</span>        if (cmd.startsWith(targetZkClusterKey)) {<a name="line.737"></a>
-<span class="sourceLineNo">738</span>          targetZkCluster = cmd.substring(targetZkClusterKey.length());<a name="line.738"></a>
-<span class="sourceLineNo">739</span>          continue;<a name="line.739"></a>
-<span class="sourceLineNo">740</span>        }<a name="line.740"></a>
-<span class="sourceLineNo">741</span><a name="line.741"></a>
-<span class="sourceLineNo">742</span>        final String dryRunKey = "--dryrun=";<a name="line.742"></a>
-<span class="sourceLineNo">743</span>        if (cmd.startsWith(dryRunKey)) {<a name="line.743"></a>
-<span class="sourceLineNo">744</span>          dryRun = Boolean.parseBoolean(cmd.substring(dryRunKey.length()));<a name="line.744"></a>
-<span class="sourceLineNo">745</span>          continue;<a name="line.745"></a>
-<span class="sourceLineNo">746</span>        }<a name="line.746"></a>
-<span class="sourceLineNo">747</span><a name="line.747"></a>
-<span class="sourceLineNo">748</span>        printUsage("Invalid argument '" + cmd + "'");<a name="line.748"></a>
-<span class="sourceLineNo">749</span>        return false;<a name="line.749"></a>
-<span class="sourceLineNo">750</span>      }<a name="line.750"></a>
-<span class="sourceLineNo">751</span><a name="line.751"></a>
+<span class="sourceLineNo">175</span>      MATCHINGROWS, MATCHINGCELLS, EMPTY_BATCHES, RANGESMATCHED, RANGESNOTMATCHED<a name="line.175"></a>
+<span class="sourceLineNo">176</span>    }<a name="line.176"></a>
+<span class="sourceLineNo">177</span><a name="line.177"></a>
+<span class="sourceLineNo">178</span>    @Override<a name="line.178"></a>
+<span class="sourceLineNo">179</span>    protected void setup(Context context) throws IOException {<a name="line.179"></a>
+<span class="sourceLineNo">180</span><a name="line.180"></a>
+<span class="sourceLineNo">181</span>      Configuration conf = context.getConfiguration();<a name="line.181"></a>
+<span class="sourceLineNo">182</span>      sourceHashDir = new Path(conf.get(SOURCE_HASH_DIR_CONF_KEY));<a name="line.182"></a>
+<span class="sourceLineNo">183</span>      sourceConnection = openConnection(conf, SOURCE_ZK_CLUSTER_CONF_KEY, null);<a name="line.183"></a>
+<span class="sourceLineNo">184</span>      targetConnection = openConnection(conf, TARGET_ZK_CLUSTER_CONF_KEY,<a name="line.184"></a>
+<span class="sourceLineNo">185</span>          TableOutputFormat.OUTPUT_CONF_PREFIX);<a name="line.185"></a>
+<span class="sourceLineNo">186</span>      sourceTable = openTable(sourceConnection, conf, SOURCE_TABLE_CONF_KEY);<a name="line.186"></a>
+<span class="sourceLineNo">187</span>      targetTable = openTable(targetConnection, conf, TARGET_TABLE_CONF_KEY);<a name="line.187"></a>
+<span class="sourceLineNo">188</span>      dryRun = conf.getBoolean(SOURCE_TABLE_CONF_KEY, false);<a name="line.188"></a>
+<span class="sourceLineNo">189</span><a name="line.189"></a>
+<span class="sourceLineNo">190</span>      sourceTableHash = HashTable.TableHash.read(conf, sourceHashDir);<a name="line.190"></a>
+<span class="sourceLineNo">191</span>      LOG.info("Read source hash manifest: " + sourceTableHash);<a name="line.191"></a>
+<span class="sourceLineNo">192</span>      LOG.info("Read " + sourceTableHash.partitions.size() + " partition keys");<a name="line.192"></a>
+<span class="sourceLineNo">193</span><a name="line.193"></a>
+<span class="sourceLineNo">194</span>      TableSplit split = (TableSplit) context.getInputSplit();<a name="line.194"></a>
+<span class="sourceLineNo">195</span>      ImmutableBytesWritable splitStartKey = new ImmutableBytesWritable(split.getStartRow());<a name="line.195"></a>
+<span class="sourceLineNo">196</span><a name="line.196"></a>
+<span class="sourceLineNo">197</span>      sourceHashReader = sourceTableHash.newReader(conf, splitStartKey);<a name="line.197"></a>
+<span class="sourceLineNo">198</span>      findNextKeyHashPair();<a name="line.198"></a>
+<span class="sourceLineNo">199</span><a name="line.199"></a>
+<span class="sourceLineNo">200</span>      // create a hasher, but don't start it right away<a name="line.200"></a>
+<span class="sourceLineNo">201</span>      // instead, find the first hash batch at or after the start row<a name="line.201"></a>
+<span class="sourceLineNo">202</span>      // and skip any rows that come before.  they will be caught by the previous task<a name="line.202"></a>
+<span class="sourceLineNo">203</span>      targetHasher = new HashTable.ResultHasher();<a name="line.203"></a>
+<span class="sourceLineNo">204</span>    }<a name="line.204"></a>
+<span class="sourceLineNo">205</span><a name="line.205"></a>
+<span class="sourceLineNo">206</span>    private static Connection openConnection(Configuration conf, String zkClusterConfKey,<a name="line.206"></a>
+<span class="sourceLineNo">207</span>                                             String configPrefix)<a name="line.207"></a>
+<span class="sourceLineNo">208</span>      throws IOException {<a name="line.208"></a>
+<span class="sourceLineNo">209</span>        String zkCluster = conf.get(zkClusterConfKey);<a name="line.209"></a>
+<span class="sourceLineNo">210</span>        Configuration clusterConf = HBaseConfiguration.createClusterConf(conf,<a name="line.210"></a>
+<span class="sourceLineNo">211</span>            zkCluster, configPrefix);<a name="line.211"></a>
+<span class="sourceLineNo">212</span>        return ConnectionFactory.createConnection(clusterConf);<a name="line.212"></a>
+<span class="sourceLineNo">213</span>    }<a name="line.213"></a>
+<span class="sourceLineNo">214</span><a name="line.214"></a>
+<span class="sourceLineNo">215</span>    private static Table openTable(Connection connection, Configuration conf,<a name="line.215"></a>
+<span class="sourceLineNo">216</span>        String tableNameConfKey) throws IOException {<a name="line.216"></a>
+<span class="sourceLineNo">217</span>      return connection.getTable(TableName.valueOf(conf.get(tableNameConfKey)));<a name="line.217"></a>
+<span class="sourceLineNo">218</span>    }<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>     * Attempt to read the next source key/hash pair.<a name="line.221"></a>
+<span class="sourceLineNo">222</span>     * If there are no more, set nextSourceKey to null<a name="line.222"></a>
+<span class="sourceLineNo">223</span>     */<a name="line.223"></a>
+<span class="sourceLineNo">224</span>    private void findNextKeyHashPair() throws IOException {<a name="line.224"></a>
+<span class="sourceLineNo">225</span>      boolean hasNext = sourceHashReader.next();<a name="line.225"></a>
+<span class="sourceLineNo">226</span>      if (hasNext) {<a name="line.226"></a>
+<span class="sourceLineNo">227</span>        nextSourceKey = sourceHashReader.getCurrentKey();<a name="line.227"></a>
+<span class="sourceLineNo">228</span>      } else {<a name="line.228"></a>
+<span class="sourceLineNo">229</span>        // no more keys - last hash goes to the end<a name="line.229"></a>
+<span class="sourceLineNo">230</span>        nextSourceKey = null;<a name="line.230"></a>
+<span class="sourceLineNo">231</span>      }<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>    @Override<a name="line.234"></a>
+<span class="sourceLineNo">235</span>    protected void map(ImmutableBytesWritable key, Result value, Context context)<a name="line.235"></a>
+<span class="sourceLineNo">236</span>        throws IOException, InterruptedException {<a name="line.236"></a>
+<span class="sourceLineNo">237</span>      try {<a name="line.237"></a>
+<span class="sourceLineNo">238</span>        // first, finish any hash batches that end before the scanned row<a name="line.238"></a>
+<span class="sourceLineNo">239</span>        while (nextSourceKey != null &amp;&amp; key.compareTo(nextSourceKey) &gt;= 0) {<a name="line.239"></a>
+<span class="sourceLineNo">240</span>          moveToNextBatch(context);<a name="line.240"></a>
+<span class="sourceLineNo">241</span>        }<a name="line.241"></a>
+<span class="sourceLineNo">242</span><a name="line.242"></a>
+<span class="sourceLineNo">243</span>        // next, add the scanned row (as long as we've reached the first batch)<a name="line.243"></a>
+<span class="sourceLineNo">244</span>        if (targetHasher.isBatchStarted()) {<a name="line.244"></a>
+<span class="sourceLineNo">245</span>          targetHasher.hashResult(value);<a name="line.245"></a>
+<span class="sourceLineNo">246</span>        }<a name="line.246"></a>
+<span class="sourceLineNo">247</span>      } catch (Throwable t) {<a name="line.247"></a>
+<span class="sourceLineNo">248</span>        mapperException = t;<a name="line.248"></a>
+<span class="sourceLineNo">249</span>        Throwables.propagateIfInstanceOf(t, IOException.class);<a name="line.249"></a>
+<span class="sourceLineNo">250</span>        Throwables.propagateIfInstanceOf(t, InterruptedException.class);<a name="line.250"></a>
+<span class="sourceLineNo">251</span>        Throwables.propagate(t);<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><a name="line.254"></a>
+<span class="sourceLineNo">255</span>    /**<a name="line.255"></a>
+<span class="sourceLineNo">256</span>     * If there is an open hash batch, complete it and sync if there are diffs.<a name="line.256"></a>
+<span class="sourceLineNo">257</span>     * Start a new batch, and seek to read the<a name="line.257"></a>
+<span class="sourceLineNo">258</span>     */<a name="line.258"></a>
+<span class="sourceLineNo">259</span>    private void moveToNextBatch(Context context) throws IOException, InterruptedException {<a name="line.259"></a>
+<span class="sourceLineNo">260</span>      if (targetHasher.isBatchStarted()) {<a name="line.260"></a>
+<span class="sourceLineNo">261</span>        finishBatchAndCompareHashes(context);<a name="line.261"></a>
+<span class="sourceLineNo">262</span>      }<a name="line.262"></a>
+<span class="sourceLineNo">263</span>      targetHasher.startBatch(nextSourceKey);<a name="line.263"></a>
+<span class="sourceLineNo">264</span>      currentSourceHash = sourceHashReader.getCurrentHash();<a name="line.264"></a>
+<span class="sourceLineNo">265</span><a name="line.265"></a>
+<span class="sourceLineNo">266</span>      findNextKeyHashPair();<a name="line.266"></a>
+<span class="sourceLineNo">267</span>    }<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>     * Finish the currently open hash batch.<a name="line.270"></a>
+<span class="sourceLineNo">271</span>     * Compare the target hash to the given source hash.<a name="line.271"></a>
+<span class="sourceLineNo">272</span>     * If they do not match, then sync the covered key range.<a name="line.272"></a>
+<span class="sourceLineNo">273</span>     */<a name="line.273"></a>
+<span class="sourceLineNo">274</span>    private void finishBatchAndCompareHashes(Context context)<a name="line.274"></a>
+<span class="sourceLineNo">275</span>        throws IOException, InterruptedException {<a name="line.275"></a>
+<span class="sourceLineNo">276</span>      targetHasher.finishBatch();<a name="line.276"></a>
+<span class="sourceLineNo">277</span>      context.getCounter(Counter.BATCHES).increment(1);<a name="line.277"></a>
+<span class="sourceLineNo">278</span>      if (targetHasher.getBatchSize() == 0) {<a name="line.278"></a>
+<span class="sourceLineNo">279</span>        context.getCounter(Counter.EMPTY_BATCHES).increment(1);<a name="line.279"></a>
+<span class="sourceLineNo">280</span>      }<a name="line.280"></a>
+<span class="sourceLineNo">281</span>      ImmutableBytesWritable targetHash = targetHasher.getBatchHash();<a name="line.281"></a>
+<span class="sourceLineNo">282</span>      if (targetHash.equals(currentSourceHash)) {<a name="line.282"></a>
+<span class="sourceLineNo">283</span>        context.getCounter(Counter.HASHES_MATCHED).increment(1);<a name="line.283"></a>
+<span class="sourceLineNo">284</span>      } else {<a name="line.284"></a>
+<span class="sourceLineNo">285</span>        context.getCounter(Counter.HASHES_NOT_MATCHED).increment(1);<a name="line.285"></a>
+<span class="sourceLineNo">286</span><a name="line.286"></a>
+<span class="sourceLineNo">287</span>        ImmutableBytesWritable stopRow = nextSourceKey == null<a name="line.287"></a>
+<span class="sourceLineNo">288</span>                                          ? new ImmutableBytesWritable(sourceTableHash.stopRow)<a name="line.288"></a>
+<span class="sourceLineNo">289</span>                                          : nextSourceKey;<a name="line.289"></a>
+<span class="sourceLineNo">290</span><a name="line.290"></a>
+<span class="sourceLineNo">291</span>        if (LOG.isDebugEnabled()) {<a name="line.291"></a>
+<span class="sourceLineNo">292</span>          LOG.debug("Hash mismatch.  Key range: " + toHex(targetHasher.getBatchStartKey())<a name="line.292"></a>
+<span class="sourceLineNo">293</span>              + " to " + toHex(stopRow)<a name="line.293"></a>
+<span class="sourceLineNo">294</span>              + " sourceHash: " + toHex(currentSourceHash)<a name="line.294"></a>
+<span class="sourceLineNo">295</span>              + " targetHash: " + toHex(targetHash));<a name="line.295"></a>
+<span class="sourceLineNo">296</span>        }<a name="line.296"></a>
+<span class="sourceLineNo">297</span><a name="line.297"></a>
+<span class="sourceLineNo">298</span>        syncRange(context, targetHasher.getBatchStartKey(), stopRow);<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>    private static String toHex(ImmutableBytesWritable bytes) {<a name="line.301"></a>
+<span class="sourceLineNo">302</span>      return Bytes.toHex(bytes.get(), bytes.getOffset(), bytes.getLength());<a name="line.302"></a>
+<span class="sourceLineNo">303</span>    }<a name="line.303"></a>
+<span class="sourceLineNo">304</span><a name="line.304"></a>
+<span class="sourceLineNo">305</span>    private static final CellScanner EMPTY_CELL_SCANNER<a name="line.305"></a>
+<span class="sourceLineNo">306</span>      = new CellScanner(Collections.&lt;Result&gt;emptyIterator());<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>     * Rescan the given range directly from the source and target tables.<a name="line.309"></a>
+<span class="sourceLineNo">310</span>     * Count and log differences, and if this is not a dry run, output Puts and Deletes<a name="line.310"></a>
+<span class="sourceLineNo">311</span>     * to make the target table match the source table for this range<a name="line.311"></a>
+<span class="sourceLineNo">312</span>     */<a name="line.312"></a>
+<span class="sourceLineNo">313</span>    private void syncRange(Context context, ImmutableBytesWritable startRow,<a name="line.313"></a>
+<span class="sourceLineNo">314</span>        ImmutableBytesWritable stopRow) throws IOException, InterruptedException {<a name="line.314"></a>
+<span class="sourceLineNo">315</span>      Scan scan = sourceTableHash.initScan();<a name="line.315"></a>
+<span class="sourceLineNo">316</span>      scan.setStartRow(startRow.copyBytes());<a name="line.316"></a>
+<span class="sourceLineNo">317</span>      scan.setStopRow(stopRow.copyBytes());<a name="line.317"></a>
+<span class="sourceLineNo">318</span><a name="line.318"></a>
+<span class="sourceLineNo">319</span>      ResultScanner sourceScanner = sourceTable.getScanner(scan);<a name="line.319"></a>
+<span class="sourceLineNo">320</span>      CellScanner sourceCells = new CellScanner(sourceScanner.iterator());<a name="line.320"></a>
+<span class="sourceLineNo">321</span><a name="line.321"></a>
+<span class="sourceLineNo">322</span>      ResultScanner targetScanner = targetTable.getScanner(new Scan(scan));<a name="line.322"></a>
+<span class="sourceLineNo">323</span>      CellScanner targetCells = new CellScanner(targetScanner.iterator());<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span>      boolean rangeMatched = true;<a name="line.325"></a>
+<span class="sourceLineNo">326</span>      byte[] nextSourceRow = sourceCells.nextRow();<a name="line.326"></a>
+<span class="sourceLineNo">327</span>      byte[] nextTargetRow = targetCells.nextRow();<a name="line.327"></a>
+<span class="sourceLineNo">328</span>      while(nextSourceRow != null || nextTargetRow != null) {<a name="line.328"></a>
+<span class="sourceLineNo">329</span>        boolean rowMatched;<a name="line.329"></a>
+<span class="sourceLineNo">330</span>        int rowComparison = compareRowKeys(nextSourceRow, nextTargetRow);<a name="line.330"></a>
+<span class="sourceLineNo">331</span>        if (rowComparison &lt; 0) {<a name="line.331"></a>
+<span class="sourceLineNo">332</span>          if (LOG.isInfoEnabled()) {<a name="line.332"></a>
+<span class="sourceLineNo">333</span>            LOG.info("Target missing row: " + Bytes.toHex(nextSourceRow));<a name="line.333"></a>
+<span class="sourceLineNo">334</span>          }<a name="line.334"></a>
+<span class="sourceLineNo">335</span>          context.getCounter(Counter.TARGETMISSINGROWS).increment(1);<a name="line.335"></a>
+<span class="sourceLineNo">336</span><a name="line.336"></a>
+<span class="sourceLineNo">337</span>          rowMatched = syncRowCells(context, nextSourceRow, sourceCells, EMPTY_CELL_SCANNER);<a name="line.337"></a>
+<span class="sourceLineNo">338</span>          nextSourceRow = sourceCells.nextRow();  // advance only source to next row<a name="line.338"></a>
+<span class="sourceLineNo">339</span>        } else if (rowComparison &gt; 0) {<a name="line.339"></a>
+<span class="sourceLineNo">340</span>          if (LOG.isInfoEnabled()) {<a name="line.340"></a>
+<span class="sourceLineNo">341</span>            LOG.info("Source missing row: " + Bytes.toHex(nextTargetRow));<a name="line.341"></a>
+<span class="sourceLineNo">342</span>          }<a name="line.342"></a>
+<span class="sourceLineNo">343</span>          context.getCounter(Counter.SOURCEMISSINGROWS).increment(1);<a name="line.343"></a>
+<span class="sourceLineNo">344</span><a name="line.344"></a>
+<span class="sourceLineNo">345</span>          rowMatched = syncRowCells(context, nextTargetRow, EMPTY_CELL_SCANNER, targetCells);<a name="line.345"></a>
+<span class="sourceLineNo">346</span>          nextTargetRow = targetCells.nextRow();  // advance only target to next row<a name="line.346"></a>
+<span class="sourceLineNo">347</span>        } else {<a name="line.347"></a>
+<span class="sourceLineNo">348</span>          // current row is the same on both sides, compare cell by cell<a name="line.348"></a>
+<span class="sourceLineNo">349</span>          rowMatched = syncRowCells(context, nextSourceRow, sourceCells, targetCells);<a name="line.349"></a>
+<span class="sourceLineNo">350</span>          nextSourceRow = sourceCells.nextRow();<a name="line.350"></a>
+<span class="sourceLineNo">351</span>          nextTargetRow = targetCells.nextRow();<a name="line.351"></a>
+<span class="sourceLineNo">352</span>        }<a name="line.352"></a>
+<span class="sourceLineNo">353</span><a name="line.353"></a>
+<span class="sourceLineNo">354</span>        if (!rowMatched) {<a name="line.354"></a>
+<span class="sourceLineNo">355</span>          rangeMatched = false;<a name="line.355"></a>
+<span class="sourceLineNo">356</span>        }<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>      sourceScanner.close();<a name="line.359"></a>
+<span class="sourceLineNo">360</span>      targetScanner.close();<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span>      context.getCounter(rangeMatched ? Counter.RANGESMATCHED : Counter.RANGESNOTMATCHED)<a name="line.362"></a>
+<span class="sourceLineNo">363</span>        .increment(1);<a name="line.363"></a>
+<span class="sourceLineNo">364</span>    }<a name="line.364"></a>
+<span class="sourceLineNo">365</span><a name="line.365"></a>
+<span class="sourceLineNo">366</span>    private static class CellScanner {<a name="line.366"></a>
+<span class="sourceLineNo">367</span>      private final Iterator&lt;Result&gt; results;<a name="line.367"></a>
+<span class="sourceLineNo">368</span><a name="line.368"></a>
+<span class="sourceLineNo">369</span>      private byte[] currentRow;<a name="line.369"></a>
+<span class="sourceLineNo">370</span>      private Result currentRowResult;<a name="line.370"></a>
+<span class="sourceLineNo">371</span>      private int nextCellInRow;<a name="line.371"></a>
+<span class="sourceLineNo">372</span><a name="line.372"></a>
+<span class="sourceLineNo">373</span>      private Result nextRowResult;<a name="line.373"></a>
+<span class="sourceLineNo">374</span><a name="line.374"></a>
+<span class="sourceLineNo">375</span>      public CellScanner(Iterator&lt;Result&gt; results) {<a name="line.375"></a>
+<span class="sourceLineNo">376</span>        this.results = results;<a name="line.376"></a>
+<span class="sourceLineNo">377</span>      }<a name="line.377"></a>
+<span class="sourceLineNo">378</span><a name="line.378"></a>
+<span class="sourceLineNo">379</span>      /**<a name="line.379"></a>
+<span class="sourceLineNo">380</span>       * Advance to the next row and return its row key.<a name="line.380"></a>
+<span class="sourceLineNo">381</span>       * Returns null iff there are no more rows.<a name="line.381"></a>
+<span class="sourceLineNo">382</span>       */<a name="line.382"></a>
+<span class="sourceLineNo">383</span>      public byte[] nextRow() {<a name="line.383"></a>
+<span class="sourceLineNo">384</span>        if (nextRowResult == null) {<a name="line.384"></a>
+<span class="sourceLineNo">385</span>          // no cached row - check scanner for more<a name="line.385"></a>
+<span class="sourceLineNo">386</span>          while (results.hasNext()) {<a name="line.386"></a>
+<span class="sourceLineNo">387</span>            nextRowResult = results.next();<a name="line.387"></a>
+<span class="sourceLineNo">388</span>            Cell nextCell = nextRowResult.rawCells()[0];<a name="line.388"></a>
+<span class="sourceLineNo">389</span>            if (currentRow == null<a name="line.389"></a>
+<span class="sourceLineNo">390</span>                || !Bytes.equals(currentRow, 0, currentRow.length, nextCell.getRowArray(),<a name="line.390"></a>
+<span class="sourceLineNo">391</span>                nextCell.getRowOffset(), nextCell.getRowLength())) {<a name="line.391"></a>
+<span class="sourceLineNo">392</span>              // found next row<a name="line.392"></a>
+<span class="sourceLineNo">393</span>              break;<a name="line.393"></a>
+<span class="sourceLineNo">394</span>            } else {<a name="line.394"></a>
+<span class="sourceLineNo">395</span>              // found another result from current row, keep scanning<a name="line.395"></a>
+<span class="sourceLineNo">396</span>              nextRowResult = null;<a name="line.396"></a>
+<span class="sourceLineNo">397</span>            }<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 (nextRowResult == null) {<a name="line.400"></a>
+<span class="sourceLineNo">401</span>            // end of data, no more rows<a name="line.401"></a>
+<span class="sourceLineNo">402</span>            currentRowResult = null;<a name="line.402"></a>
+<span class="sourceLineNo">403</span>            currentRow = null;<a name="line.403"></a>
+<span class="sourceLineNo">404</span>            return null;<a name="line.404"></a>
+<span class="sourceLineNo">405</span>          }<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>        // advance to cached result for next row<a name="line.408"></a>
+<span class="sourceLineNo">409</span>        currentRowResult = nextRowResult;<a name="line.409"></a>
+<span class="sourceLineNo">410</span>        nextCellInRow = 0;<a name="line.410"></a>
+<span class="sourceLineNo">411</span>        currentRow = currentRowResult.getRow();<a name="line.411"></a>
+<span class="sourceLineNo">412</span>        nextRowResult = null;<a name="line.412"></a>
+<span class="sourceLineNo">413</span>        return currentRow;<a name="line.413"></a>
+<span class="sourceLineNo">414</span>      }<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>       * Returns the next Cell in the current row or null iff none remain.<a name="line.417"></a>
+<span class="sourceLineNo">418</span>       */<a name="line.418"></a>
+<span class="sourceLineNo">419</span>      public Cell nextCellInRow() {<a name="line.419"></a>
+<span class="sourceLineNo">420</span>        if (currentRowResult == null) {<a name="line.420"></a>
+<span class="sourceLineNo">421</span>          // nothing left in current row<a name="line.421"></a>
+<span class="sourceLineNo">422</span>          return null;<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>        Cell nextCell = currentRowResult.rawCells()[nextCellInRow];<a name="line.425"></a>
+<span class="sourceLineNo">426</span>        nextCellInRow++;<a name="line.426"></a>
+<span class="sourceLineNo">427</span>        if (nextCellInRow == currentRowResult.size()) {<a name="line.427"></a>
+<span class="sourceLineNo">428</span>          if (results.hasNext()) {<a name="line.428"></a>
+<span class="sourceLineNo">429</span>            Result result = results.next();<a name="line.429"></a>
+<span class="sourceLineNo">430</span>            Cell cell = result.rawCells()[0];<a name="line.430"></a>
+<span class="sourceLineNo">431</span>            if (Bytes.equals(currentRow, 0, currentRow.length, cell.getRowArray(),<a name="line.431"></a>
+<span class="sourceLineNo">432</span>                cell.getRowOffset(), cell.getRowLength())) {<a name="line.432"></a>
+<span class="sourceLineNo">433</span>              // result is part of current row<a name="line.433"></a>
+<span class="sourceLineNo">434</span>              currentRowResult = result;<a name="line.434"></a>
+<span class="sourceLineNo">435</span>              nextCellInRow = 0;<a name="line.435"></a>
+<span class="sourceLineNo">436</span>            } else {<a name="line.436"></a>
+<span class="sourceLineNo">437</span>              // result is part of next row, cache it<a name="line.437"></a>
+<span class="sourceLineNo">438</span>              nextRowResult = result;<a name="line.438"></a>
+<span class="sourceLineNo">439</span>              // current row is complete<a name="line.439"></a>
+<span class="sourceLineNo">440</span>              currentRowResult = null;<a name="line.440"></a>
+<span class="sourceLineNo">441</span>            }<a name="line.441"></a>
+<span class="sourceLineNo">442</span>          } else {<a name="line.442"></a>
+<span class="sourceLineNo">443</span>            // end of data<a name="line.443"></a>
+<span class="sourceLineNo">444</span>            currentRowResult = null;<a name="line.444"></a>
+<span class="sourceLineNo">445</span>          }<a name="line.445"></a>
+<span class="sourceLineNo">446</span>        }<a name="line.446"></a>
+<span class="sourceLineNo">447</span>        return nextCell;<a name="line.447"></a>
+<span class="sourceLineNo">448</span>      }<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>     * Compare the cells for the given row from the source and target tables.<a name="line.452"></a>
+<span class="sourceLineNo">453</span>     * Count and log any differences.<a name="line.453"></a>
+<span class="sourceLineNo">454</span>     * If not a dry run, output a Put and/or Delete needed to sync the target table<a name="line.454"></a>
+<span class="sourceLineNo">455</span>     * to match the source table.<a name="line.455"></a>
+<span class="sourceLineNo">456</span>     */<a name="line.456"></a>
+<span class="sourceLineNo">457</span>    private boolean syncRowCells(Context context, byte[] rowKey, CellScanner sourceCells,<a name="line.457"></a>
+<span class="sourceLineNo">458</span>        CellScanner targetCells) throws IOException, InterruptedException {<a name="line.458"></a>
+<span class="sourceLineNo">459</span>      Put put = null;<a name="line.459"></a>
+<span class="sourceLineNo">460</span>      Delete delete = null;<a name="line.460"></a>
+<span class="sourceLineNo">461</span>      long matchingCells = 0;<a name="line.461"></a>
+<span class="sourceLineNo">462</span>      boolean matchingRow = true;<a name="line.462"></a>
+<span class="sourceLineNo">463</span>      Cell sourceCell = sourceCells.nextCellInRow();<a name="line.463"></a>
+<span class="sourceLineNo">464</span>      Cell targetCell = targetCells.nextCellInRow();<a name="line.464"></a>
+<span class="sourceLineNo">465</span>      while (sourceCell != null || targetCell != null) {<a name="line.465"></a>
+<span class="sourceLineNo">466</span><a name="line.466"></a>
+<span class="sourceLineNo">467</span>        int cellKeyComparison = compareCellKeysWithinRow(sourceCell, targetCell);<a name="line.467"></a>
+<span class="sourceLineNo">468</span>        if (cellKeyComparison &lt; 0) {<a name="line.468"></a>
+<span class="sourceLineNo">469</span>          if (LOG.isDebugEnabled()) {<a name="line.469"></a>
+<span class="sourceLineNo">470</span>            LOG.debug("Target missing cell: " + sourceCell);<a name="line.470"></a>
+<span class="sourceLineNo">471</span>          }<a name="line.471"></a>
+<span class="sourceLineNo">472</span>          context.getCounter(Counter.TARGETMISSINGCELLS).increment(1);<a name="line.472"></a>
+<span class="sourceLineNo">473</span>          matchingRow = false;<a name="line.473"></a>
+<span class="sourceLineNo">474</span><a name="line.474"></a>
+<span class="sourceLineNo">475</span>          if (!dryRun) {<a name="line.475"></a>
+<span class="sourceLineNo">476</span>            if (put == null) {<a name="line.476"></a>
+<span class="sourceLineNo">477</span>              put = new Put(rowKey);<a name="line.477"></a>
+<span class="sourceLineNo">478</span>            }<a name="line.478"></a>
+<span class="sourceLineNo">479</span>            put.add(sourceCell);<a name="line.479"></a>
+<span class="sourceLineNo">480</span>          }<a name="line.480"></a>
+<span class="sourceLineNo">481</span><a name="line.481"></a>
+<span class="sourceLineNo">482</span>          sourceCell = sourceCells.nextCellInRow();<a name="line.482"></a>
+<span class="sourceLineNo">483</span>        } else if (cellKeyComparison &gt; 0) {<a name="line.483"></a>
+<span class="sourceLineNo">484</span>          if (LOG.isDebugEnabled()) {<a name="line.484"></a>
+<span class="sourceLineNo">485</span>            LOG.debug("Source missing cell: " + targetCell);<a name="line.485"></a>
+<span class="sourceLineNo">486</span>          }<a name="line.486"></a>
+<span class="sourceLineNo">487</span>          context.getCounter(Counter.SOURCEMISSINGCELLS).increment(1);<a name="line.487"></a>
+<span class="sourceLineNo">488</span>          matchingRow = false;<a 

<TRUNCATED>

Mime
View raw message