hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From bus...@apache.org
Subject [14/51] [partial] hbase-site git commit: Published site at ce318a2906817058ae7b2fce6e9b54d9d6230f9b.
Date Wed, 27 Apr 2016 17:06:22 GMT
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/4131cace/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
index 9671728..3d4d3d3 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.html
@@ -309,495 +309,500 @@
 <span class="sourceLineNo">301</span>                                    .withChecksumType(HStore.getChecksumType(conf))<a name="line.301"></a>
 <span class="sourceLineNo">302</span>                                    .withBytesPerCheckSum(HStore.getBytesPerChecksum(conf))<a name="line.302"></a>
 <span class="sourceLineNo">303</span>                                    .withBlockSize(blockSize);<a name="line.303"></a>
-<span class="sourceLineNo">304</span>        contextBuilder.withDataBlockEncoding(encoding);<a name="line.304"></a>
-<span class="sourceLineNo">305</span>        HFileContext hFileContext = contextBuilder.build();<a name="line.305"></a>
-<span class="sourceLineNo">306</span>                                    <a name="line.306"></a>
-<span class="sourceLineNo">307</span>        if (null == favoredNodes) {<a name="line.307"></a>
-<span class="sourceLineNo">308</span>          wl.writer =<a name="line.308"></a>
-<span class="sourceLineNo">309</span>              new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), fs)<a name="line.309"></a>
-<span class="sourceLineNo">310</span>                  .withOutputDir(familydir).withBloomType(bloomType)<a name="line.310"></a>
-<span class="sourceLineNo">311</span>                  .withComparator(CellComparator.COMPARATOR).withFileContext(hFileContext).build();<a name="line.311"></a>
-<span class="sourceLineNo">312</span>        } else {<a name="line.312"></a>
+<span class="sourceLineNo">304</span><a name="line.304"></a>
+<span class="sourceLineNo">305</span>        if (HFile.getFormatVersion(conf) &gt;= HFile.MIN_FORMAT_VERSION_WITH_TAGS) {<a name="line.305"></a>
+<span class="sourceLineNo">306</span>          contextBuilder.withIncludesTags(true);<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>        contextBuilder.withDataBlockEncoding(encoding);<a name="line.309"></a>
+<span class="sourceLineNo">310</span>        HFileContext hFileContext = contextBuilder.build();<a name="line.310"></a>
+<span class="sourceLineNo">311</span>                                    <a name="line.311"></a>
+<span class="sourceLineNo">312</span>        if (null == favoredNodes) {<a name="line.312"></a>
 <span class="sourceLineNo">313</span>          wl.writer =<a name="line.313"></a>
-<span class="sourceLineNo">314</span>              new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), new HFileSystem(fs))<a name="line.314"></a>
+<span class="sourceLineNo">314</span>              new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), fs)<a name="line.314"></a>
 <span class="sourceLineNo">315</span>                  .withOutputDir(familydir).withBloomType(bloomType)<a name="line.315"></a>
-<span class="sourceLineNo">316</span>                  .withComparator(CellComparator.COMPARATOR).withFileContext(hFileContext)<a name="line.316"></a>
-<span class="sourceLineNo">317</span>                  .withFavoredNodes(favoredNodes).build();<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>        this.writers.put(family, wl);<a name="line.320"></a>
-<span class="sourceLineNo">321</span>        return wl;<a name="line.321"></a>
-<span class="sourceLineNo">322</span>      }<a name="line.322"></a>
-<span class="sourceLineNo">323</span><a name="line.323"></a>
-<span class="sourceLineNo">324</span>      private void close(final StoreFileWriter w) throws IOException {<a name="line.324"></a>
-<span class="sourceLineNo">325</span>        if (w != null) {<a name="line.325"></a>
-<span class="sourceLineNo">326</span>          w.appendFileInfo(StoreFile.BULKLOAD_TIME_KEY,<a name="line.326"></a>
-<span class="sourceLineNo">327</span>              Bytes.toBytes(System.currentTimeMillis()));<a name="line.327"></a>
-<span class="sourceLineNo">328</span>          w.appendFileInfo(StoreFile.BULKLOAD_TASK_KEY,<a name="line.328"></a>
-<span class="sourceLineNo">329</span>              Bytes.toBytes(context.getTaskAttemptID().toString()));<a name="line.329"></a>
-<span class="sourceLineNo">330</span>          w.appendFileInfo(StoreFile.MAJOR_COMPACTION_KEY,<a name="line.330"></a>
-<span class="sourceLineNo">331</span>              Bytes.toBytes(true));<a name="line.331"></a>
-<span class="sourceLineNo">332</span>          w.appendFileInfo(StoreFile.EXCLUDE_FROM_MINOR_COMPACTION_KEY,<a name="line.332"></a>
-<span class="sourceLineNo">333</span>              Bytes.toBytes(compactionExclude));<a name="line.333"></a>
-<span class="sourceLineNo">334</span>          w.appendTrackedTimestampsToMetadata();<a name="line.334"></a>
-<span class="sourceLineNo">335</span>          w.close();<a name="line.335"></a>
-<span class="sourceLineNo">336</span>        }<a name="line.336"></a>
-<span class="sourceLineNo">337</span>      }<a name="line.337"></a>
-<span class="sourceLineNo">338</span><a name="line.338"></a>
-<span class="sourceLineNo">339</span>      @Override<a name="line.339"></a>
-<span class="sourceLineNo">340</span>      public void close(TaskAttemptContext c)<a name="line.340"></a>
-<span class="sourceLineNo">341</span>      throws IOException, InterruptedException {<a name="line.341"></a>
-<span class="sourceLineNo">342</span>        for (WriterLength wl: this.writers.values()) {<a name="line.342"></a>
-<span class="sourceLineNo">343</span>          close(wl.writer);<a name="line.343"></a>
-<span class="sourceLineNo">344</span>        }<a name="line.344"></a>
-<span class="sourceLineNo">345</span>      }<a name="line.345"></a>
-<span class="sourceLineNo">346</span>    };<a name="line.346"></a>
-<span class="sourceLineNo">347</span>  }<a name="line.347"></a>
-<span class="sourceLineNo">348</span><a name="line.348"></a>
-<span class="sourceLineNo">349</span>  /*<a name="line.349"></a>
-<span class="sourceLineNo">350</span>   * Data structure to hold a Writer and amount of data written on it.<a name="line.350"></a>
-<span class="sourceLineNo">351</span>   */<a name="line.351"></a>
-<span class="sourceLineNo">352</span>  static class WriterLength {<a name="line.352"></a>
-<span class="sourceLineNo">353</span>    long written = 0;<a name="line.353"></a>
-<span class="sourceLineNo">354</span>    StoreFileWriter writer = null;<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>   * Return the start keys of all of the regions in this table,<a name="line.358"></a>
-<span class="sourceLineNo">359</span>   * as a list of ImmutableBytesWritable.<a name="line.359"></a>
-<span class="sourceLineNo">360</span>   */<a name="line.360"></a>
-<span class="sourceLineNo">361</span>  private static List&lt;ImmutableBytesWritable&gt; getRegionStartKeys(RegionLocator table)<a name="line.361"></a>
-<span class="sourceLineNo">362</span>  throws IOException {<a name="line.362"></a>
-<span class="sourceLineNo">363</span>    byte[][] byteKeys = table.getStartKeys();<a name="line.363"></a>
-<span class="sourceLineNo">364</span>    ArrayList&lt;ImmutableBytesWritable&gt; ret =<a name="line.364"></a>
-<span class="sourceLineNo">365</span>      new ArrayList&lt;ImmutableBytesWritable&gt;(byteKeys.length);<a name="line.365"></a>
-<span class="sourceLineNo">366</span>    for (byte[] byteKey : byteKeys) {<a name="line.366"></a>
-<span class="sourceLineNo">367</span>      ret.add(new ImmutableBytesWritable(byteKey));<a name="line.367"></a>
-<span class="sourceLineNo">368</span>    }<a name="line.368"></a>
-<span class="sourceLineNo">369</span>    return ret;<a name="line.369"></a>
-<span class="sourceLineNo">370</span>  }<a name="line.370"></a>
-<span class="sourceLineNo">371</span><a name="line.371"></a>
-<span class="sourceLineNo">372</span>  /**<a name="line.372"></a>
-<span class="sourceLineNo">373</span>   * Write out a {@link SequenceFile} that can be read by<a name="line.373"></a>
-<span class="sourceLineNo">374</span>   * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.374"></a>
-<span class="sourceLineNo">375</span>   */<a name="line.375"></a>
-<span class="sourceLineNo">376</span>  @SuppressWarnings("deprecation")<a name="line.376"></a>
-<span class="sourceLineNo">377</span>  private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.377"></a>
-<span class="sourceLineNo">378</span>      List&lt;ImmutableBytesWritable&gt; startKeys) throws IOException {<a name="line.378"></a>
-<span class="sourceLineNo">379</span>    LOG.info("Writing partition information to " + partitionsPath);<a name="line.379"></a>
-<span class="sourceLineNo">380</span>    if (startKeys.isEmpty()) {<a name="line.380"></a>
-<span class="sourceLineNo">381</span>      throw new IllegalArgumentException("No regions passed");<a name="line.381"></a>
-<span class="sourceLineNo">382</span>    }<a name="line.382"></a>
-<span class="sourceLineNo">383</span><a name="line.383"></a>
-<span class="sourceLineNo">384</span>    // We're generating a list of split points, and we don't ever<a name="line.384"></a>
-<span class="sourceLineNo">385</span>    // have keys &lt; the first region (which has an empty start key)<a name="line.385"></a>
-<span class="sourceLineNo">386</span>    // so we need to remove it. Otherwise we would end up with an<a name="line.386"></a>
-<span class="sourceLineNo">387</span>    // empty reducer with index 0<a name="line.387"></a>
-<span class="sourceLineNo">388</span>    TreeSet&lt;ImmutableBytesWritable&gt; sorted =<a name="line.388"></a>
-<span class="sourceLineNo">389</span>      new TreeSet&lt;ImmutableBytesWritable&gt;(startKeys);<a name="line.389"></a>
-<span class="sourceLineNo">390</span><a name="line.390"></a>
-<span class="sourceLineNo">391</span>    ImmutableBytesWritable first = sorted.first();<a name="line.391"></a>
-<span class="sourceLineNo">392</span>    if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.392"></a>
-<span class="sourceLineNo">393</span>      throw new IllegalArgumentException(<a name="line.393"></a>
-<span class="sourceLineNo">394</span>          "First region of table should have empty start key. Instead has: "<a name="line.394"></a>
-<span class="sourceLineNo">395</span>          + Bytes.toStringBinary(first.get()));<a name="line.395"></a>
-<span class="sourceLineNo">396</span>    }<a name="line.396"></a>
-<span class="sourceLineNo">397</span>    sorted.remove(first);<a name="line.397"></a>
-<span class="sourceLineNo">398</span><a name="line.398"></a>
-<span class="sourceLineNo">399</span>    // Write the actual file<a name="line.399"></a>
-<span class="sourceLineNo">400</span>    FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.400"></a>
-<span class="sourceLineNo">401</span>    SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.401"></a>
-<span class="sourceLineNo">402</span>      fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.402"></a>
-<span class="sourceLineNo">403</span>      NullWritable.class);<a name="line.403"></a>
-<span class="sourceLineNo">404</span><a name="line.404"></a>
-<span class="sourceLineNo">405</span>    try {<a name="line.405"></a>
-<span class="sourceLineNo">406</span>      for (ImmutableBytesWritable startKey : sorted) {<a name="line.406"></a>
-<span class="sourceLineNo">407</span>        writer.append(startKey, NullWritable.get());<a name="line.407"></a>
-<span class="sourceLineNo">408</span>      }<a name="line.408"></a>
-<span class="sourceLineNo">409</span>    } finally {<a name="line.409"></a>
-<span class="sourceLineNo">410</span>      writer.close();<a name="line.410"></a>
-<span class="sourceLineNo">411</span>    }<a name="line.411"></a>
-<span class="sourceLineNo">412</span>  }<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>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.415"></a>
-<span class="sourceLineNo">416</span>   * table. This<a name="line.416"></a>
-<span class="sourceLineNo">417</span>   * &lt;ul&gt;<a name="line.417"></a>
-<span class="sourceLineNo">418</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.418"></a>
-<span class="sourceLineNo">419</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.419"></a>
-<span class="sourceLineNo">420</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.420"></a>
-<span class="sourceLineNo">421</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.421"></a>
-<span class="sourceLineNo">422</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.422"></a>
-<span class="sourceLineNo">423</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.423"></a>
-<span class="sourceLineNo">424</span>   * &lt;/ul&gt;<a name="line.424"></a>
-<span class="sourceLineNo">425</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.425"></a>
-<span class="sourceLineNo">426</span>   * running this function.<a name="line.426"></a>
-<span class="sourceLineNo">427</span>   */<a name="line.427"></a>
-<span class="sourceLineNo">428</span>  public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.428"></a>
-<span class="sourceLineNo">429</span>      throws IOException {<a name="line.429"></a>
-<span class="sourceLineNo">430</span>    configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.430"></a>
-<span class="sourceLineNo">431</span>  }<a name="line.431"></a>
-<span class="sourceLineNo">432</span><a name="line.432"></a>
-<span class="sourceLineNo">433</span>  /**<a name="line.433"></a>
-<span class="sourceLineNo">434</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.434"></a>
-<span class="sourceLineNo">435</span>   * table. This<a name="line.435"></a>
-<span class="sourceLineNo">436</span>   * &lt;ul&gt;<a name="line.436"></a>
-<span class="sourceLineNo">437</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.437"></a>
-<span class="sourceLineNo">438</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.438"></a>
-<span class="sourceLineNo">439</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.439"></a>
-<span class="sourceLineNo">440</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.440"></a>
-<span class="sourceLineNo">441</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.441"></a>
-<span class="sourceLineNo">442</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.442"></a>
-<span class="sourceLineNo">443</span>   * &lt;/ul&gt;<a name="line.443"></a>
-<span class="sourceLineNo">444</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.444"></a>
-<span class="sourceLineNo">445</span>   * running this function.<a name="line.445"></a>
-<span class="sourceLineNo">446</span>   */<a name="line.446"></a>
-<span class="sourceLineNo">447</span>  public static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.447"></a>
-<span class="sourceLineNo">448</span>      RegionLocator regionLocator) throws IOException {<a name="line.448"></a>
-<span class="sourceLineNo">449</span>    configureIncrementalLoad(job, tableDescriptor, regionLocator, HFileOutputFormat2.class);<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>  static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.452"></a>
-<span class="sourceLineNo">453</span>      RegionLocator regionLocator, Class&lt;? extends OutputFormat&lt;?, ?&gt;&gt; cls) throws IOException,<a name="line.453"></a>
-<span class="sourceLineNo">454</span>      UnsupportedEncodingException {<a name="line.454"></a>
-<span class="sourceLineNo">455</span>    Configuration conf = job.getConfiguration();<a name="line.455"></a>
-<span class="sourceLineNo">456</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.456"></a>
-<span class="sourceLineNo">457</span>    job.setOutputValueClass(KeyValue.class);<a name="line.457"></a>
-<span class="sourceLineNo">458</span>    job.setOutputFormatClass(cls);<a name="line.458"></a>
-<span class="sourceLineNo">459</span><a name="line.459"></a>
-<span class="sourceLineNo">460</span>    // Based on the configured map output class, set the correct reducer to properly<a name="line.460"></a>
-<span class="sourceLineNo">461</span>    // sort the incoming values.<a name="line.461"></a>
-<span class="sourceLineNo">462</span>    // TODO it would be nice to pick one or the other of these formats.<a name="line.462"></a>
-<span class="sourceLineNo">463</span>    if (KeyValue.class.equals(job.getMapOutputValueClass())) {<a name="line.463"></a>
-<span class="sourceLineNo">464</span>      job.setReducerClass(KeyValueSortReducer.class);<a name="line.464"></a>
-<span class="sourceLineNo">465</span>    } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.465"></a>
-<span class="sourceLineNo">466</span>      job.setReducerClass(PutSortReducer.class);<a name="line.466"></a>
-<span class="sourceLineNo">467</span>    } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.467"></a>
-<span class="sourceLineNo">468</span>      job.setReducerClass(TextSortReducer.class);<a name="line.468"></a>
-<span class="sourceLineNo">469</span>    } else {<a name="line.469"></a>
-<span class="sourceLineNo">470</span>      LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.470"></a>
-<span class="sourceLineNo">471</span>    }<a name="line.471"></a>
-<span class="sourceLineNo">472</span><a name="line.472"></a>
-<span class="sourceLineNo">473</span>    conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.473"></a>
-<span class="sourceLineNo">474</span>        MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.474"></a>
-<span class="sourceLineNo">475</span>        KeyValueSerialization.class.getName());<a name="line.475"></a>
-<span class="sourceLineNo">476</span><a name="line.476"></a>
-<span class="sourceLineNo">477</span>    if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.477"></a>
-<span class="sourceLineNo">478</span>      // record this table name for creating writer by favored nodes<a name="line.478"></a>
-<span class="sourceLineNo">479</span>      LOG.info("bulkload locality sensitive enabled");<a name="line.479"></a>
-<span class="sourceLineNo">480</span>      conf.set(OUTPUT_TABLE_NAME_CONF_KEY, regionLocator.getName().getNameAsString());<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>    // Use table's region boundaries for TOP split points.<a name="line.483"></a>
-<span class="sourceLineNo">484</span>    LOG.info("Looking up current regions for table " + regionLocator.getName());<a name="line.484"></a>
-<span class="sourceLineNo">485</span>    List&lt;ImmutableBytesWritable&gt; startKeys = getRegionStartKeys(regionLocator);<a name="line.485"></a>
-<span class="sourceLineNo">486</span>    LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.486"></a>
-<span class="sourceLineNo">487</span>        "to match current region count");<a name="line.487"></a>
-<span class="sourceLineNo">488</span>    job.setNumReduceTasks(startKeys.size());<a name="line.488"></a>
-<span class="sourceLineNo">489</span><a name="line.489"></a>
-<span class="sourceLineNo">490</span>    configurePartitioner(job, startKeys);<a name="line.490"></a>
-<span class="sourceLineNo">491</span>    // Set compression algorithms based on column families<a name="line.491"></a>
-<span class="sourceLineNo">492</span>    configureCompression(conf, tableDescriptor);<a name="line.492"></a>
-<span class="sourceLineNo">493</span>    configureBloomType(tableDescriptor, conf);<a name="line.493"></a>
-<span class="sourceLineNo">494</span>    configureBlockSize(tableDescriptor, conf);<a name="line.494"></a>
-<span class="sourceLineNo">495</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.495"></a>
-<span class="sourceLineNo">496</span><a name="line.496"></a>
-<span class="sourceLineNo">497</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.497"></a>
-<span class="sourceLineNo">498</span>    TableMapReduceUtil.initCredentials(job);<a name="line.498"></a>
-<span class="sourceLineNo">499</span>    LOG.info("Incremental table " + regionLocator.getName() + " output configured.");<a name="line.499"></a>
-<span class="sourceLineNo">500</span>  }<a name="line.500"></a>
-<span class="sourceLineNo">501</span>  <a name="line.501"></a>
-<span class="sourceLineNo">502</span>  public static void configureIncrementalLoadMap(Job job, HTableDescriptor tableDescriptor) throws<a name="line.502"></a>
-<span class="sourceLineNo">503</span>      IOException {<a name="line.503"></a>
-<span class="sourceLineNo">504</span>    Configuration conf = job.getConfiguration();<a name="line.504"></a>
-<span class="sourceLineNo">505</span><a name="line.505"></a>
-<span class="sourceLineNo">506</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.506"></a>
-<span class="sourceLineNo">507</span>    job.setOutputValueClass(KeyValue.class);<a name="line.507"></a>
-<span class="sourceLineNo">508</span>    job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.508"></a>
-<span class="sourceLineNo">509</span><a name="line.509"></a>
-<span class="sourceLineNo">510</span>    // Set compression algorithms based on column families<a name="line.510"></a>
-<span class="sourceLineNo">511</span>    configureCompression(conf, tableDescriptor);<a name="line.511"></a>
-<span class="sourceLineNo">512</span>    configureBloomType(tableDescriptor, conf);<a name="line.512"></a>
-<span class="sourceLineNo">513</span>    configureBlockSize(tableDescriptor, conf);<a name="line.513"></a>
-<span class="sourceLineNo">514</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.514"></a>
-<span class="sourceLineNo">515</span><a name="line.515"></a>
-<span class="sourceLineNo">516</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.516"></a>
-<span class="sourceLineNo">517</span>    TableMapReduceUtil.initCredentials(job);<a name="line.517"></a>
-<span class="sourceLineNo">518</span>    LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<a name="line.518"></a>
-<span class="sourceLineNo">519</span>  }<a name="line.519"></a>
+<span class="sourceLineNo">316</span>                  .withComparator(CellComparator.COMPARATOR).withFileContext(hFileContext).build();<a name="line.316"></a>
+<span class="sourceLineNo">317</span>        } else {<a name="line.317"></a>
+<span class="sourceLineNo">318</span>          wl.writer =<a name="line.318"></a>
+<span class="sourceLineNo">319</span>              new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), new HFileSystem(fs))<a name="line.319"></a>
+<span class="sourceLineNo">320</span>                  .withOutputDir(familydir).withBloomType(bloomType)<a name="line.320"></a>
+<span class="sourceLineNo">321</span>                  .withComparator(CellComparator.COMPARATOR).withFileContext(hFileContext)<a name="line.321"></a>
+<span class="sourceLineNo">322</span>                  .withFavoredNodes(favoredNodes).build();<a name="line.322"></a>
+<span class="sourceLineNo">323</span>        }<a name="line.323"></a>
+<span class="sourceLineNo">324</span><a name="line.324"></a>
+<span class="sourceLineNo">325</span>        this.writers.put(family, wl);<a name="line.325"></a>
+<span class="sourceLineNo">326</span>        return wl;<a name="line.326"></a>
+<span class="sourceLineNo">327</span>      }<a name="line.327"></a>
+<span class="sourceLineNo">328</span><a name="line.328"></a>
+<span class="sourceLineNo">329</span>      private void close(final StoreFileWriter w) throws IOException {<a name="line.329"></a>
+<span class="sourceLineNo">330</span>        if (w != null) {<a name="line.330"></a>
+<span class="sourceLineNo">331</span>          w.appendFileInfo(StoreFile.BULKLOAD_TIME_KEY,<a name="line.331"></a>
+<span class="sourceLineNo">332</span>              Bytes.toBytes(System.currentTimeMillis()));<a name="line.332"></a>
+<span class="sourceLineNo">333</span>          w.appendFileInfo(StoreFile.BULKLOAD_TASK_KEY,<a name="line.333"></a>
+<span class="sourceLineNo">334</span>              Bytes.toBytes(context.getTaskAttemptID().toString()));<a name="line.334"></a>
+<span class="sourceLineNo">335</span>          w.appendFileInfo(StoreFile.MAJOR_COMPACTION_KEY,<a name="line.335"></a>
+<span class="sourceLineNo">336</span>              Bytes.toBytes(true));<a name="line.336"></a>
+<span class="sourceLineNo">337</span>          w.appendFileInfo(StoreFile.EXCLUDE_FROM_MINOR_COMPACTION_KEY,<a name="line.337"></a>
+<span class="sourceLineNo">338</span>              Bytes.toBytes(compactionExclude));<a name="line.338"></a>
+<span class="sourceLineNo">339</span>          w.appendTrackedTimestampsToMetadata();<a name="line.339"></a>
+<span class="sourceLineNo">340</span>          w.close();<a name="line.340"></a>
+<span class="sourceLineNo">341</span>        }<a name="line.341"></a>
+<span class="sourceLineNo">342</span>      }<a name="line.342"></a>
+<span class="sourceLineNo">343</span><a name="line.343"></a>
+<span class="sourceLineNo">344</span>      @Override<a name="line.344"></a>
+<span class="sourceLineNo">345</span>      public void close(TaskAttemptContext c)<a name="line.345"></a>
+<span class="sourceLineNo">346</span>      throws IOException, InterruptedException {<a name="line.346"></a>
+<span class="sourceLineNo">347</span>        for (WriterLength wl: this.writers.values()) {<a name="line.347"></a>
+<span class="sourceLineNo">348</span>          close(wl.writer);<a name="line.348"></a>
+<span class="sourceLineNo">349</span>        }<a name="line.349"></a>
+<span class="sourceLineNo">350</span>      }<a name="line.350"></a>
+<span class="sourceLineNo">351</span>    };<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>  /*<a name="line.354"></a>
+<span class="sourceLineNo">355</span>   * Data structure to hold a Writer and amount of data written on it.<a name="line.355"></a>
+<span class="sourceLineNo">356</span>   */<a name="line.356"></a>
+<span class="sourceLineNo">357</span>  static class WriterLength {<a name="line.357"></a>
+<span class="sourceLineNo">358</span>    long written = 0;<a name="line.358"></a>
+<span class="sourceLineNo">359</span>    StoreFileWriter writer = null;<a name="line.359"></a>
+<span class="sourceLineNo">360</span>  }<a name="line.360"></a>
+<span class="sourceLineNo">361</span><a name="line.361"></a>
+<span class="sourceLineNo">362</span>  /**<a name="line.362"></a>
+<span class="sourceLineNo">363</span>   * Return the start keys of all of the regions in this table,<a name="line.363"></a>
+<span class="sourceLineNo">364</span>   * as a list of ImmutableBytesWritable.<a name="line.364"></a>
+<span class="sourceLineNo">365</span>   */<a name="line.365"></a>
+<span class="sourceLineNo">366</span>  private static List&lt;ImmutableBytesWritable&gt; getRegionStartKeys(RegionLocator table)<a name="line.366"></a>
+<span class="sourceLineNo">367</span>  throws IOException {<a name="line.367"></a>
+<span class="sourceLineNo">368</span>    byte[][] byteKeys = table.getStartKeys();<a name="line.368"></a>
+<span class="sourceLineNo">369</span>    ArrayList&lt;ImmutableBytesWritable&gt; ret =<a name="line.369"></a>
+<span class="sourceLineNo">370</span>      new ArrayList&lt;ImmutableBytesWritable&gt;(byteKeys.length);<a name="line.370"></a>
+<span class="sourceLineNo">371</span>    for (byte[] byteKey : byteKeys) {<a name="line.371"></a>
+<span class="sourceLineNo">372</span>      ret.add(new ImmutableBytesWritable(byteKey));<a name="line.372"></a>
+<span class="sourceLineNo">373</span>    }<a name="line.373"></a>
+<span class="sourceLineNo">374</span>    return ret;<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>  /**<a name="line.377"></a>
+<span class="sourceLineNo">378</span>   * Write out a {@link SequenceFile} that can be read by<a name="line.378"></a>
+<span class="sourceLineNo">379</span>   * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.379"></a>
+<span class="sourceLineNo">380</span>   */<a name="line.380"></a>
+<span class="sourceLineNo">381</span>  @SuppressWarnings("deprecation")<a name="line.381"></a>
+<span class="sourceLineNo">382</span>  private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.382"></a>
+<span class="sourceLineNo">383</span>      List&lt;ImmutableBytesWritable&gt; startKeys) throws IOException {<a name="line.383"></a>
+<span class="sourceLineNo">384</span>    LOG.info("Writing partition information to " + partitionsPath);<a name="line.384"></a>
+<span class="sourceLineNo">385</span>    if (startKeys.isEmpty()) {<a name="line.385"></a>
+<span class="sourceLineNo">386</span>      throw new IllegalArgumentException("No regions passed");<a name="line.386"></a>
+<span class="sourceLineNo">387</span>    }<a name="line.387"></a>
+<span class="sourceLineNo">388</span><a name="line.388"></a>
+<span class="sourceLineNo">389</span>    // We're generating a list of split points, and we don't ever<a name="line.389"></a>
+<span class="sourceLineNo">390</span>    // have keys &lt; the first region (which has an empty start key)<a name="line.390"></a>
+<span class="sourceLineNo">391</span>    // so we need to remove it. Otherwise we would end up with an<a name="line.391"></a>
+<span class="sourceLineNo">392</span>    // empty reducer with index 0<a name="line.392"></a>
+<span class="sourceLineNo">393</span>    TreeSet&lt;ImmutableBytesWritable&gt; sorted =<a name="line.393"></a>
+<span class="sourceLineNo">394</span>      new TreeSet&lt;ImmutableBytesWritable&gt;(startKeys);<a name="line.394"></a>
+<span class="sourceLineNo">395</span><a name="line.395"></a>
+<span class="sourceLineNo">396</span>    ImmutableBytesWritable first = sorted.first();<a name="line.396"></a>
+<span class="sourceLineNo">397</span>    if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.397"></a>
+<span class="sourceLineNo">398</span>      throw new IllegalArgumentException(<a name="line.398"></a>
+<span class="sourceLineNo">399</span>          "First region of table should have empty start key. Instead has: "<a name="line.399"></a>
+<span class="sourceLineNo">400</span>          + Bytes.toStringBinary(first.get()));<a name="line.400"></a>
+<span class="sourceLineNo">401</span>    }<a name="line.401"></a>
+<span class="sourceLineNo">402</span>    sorted.remove(first);<a name="line.402"></a>
+<span class="sourceLineNo">403</span><a name="line.403"></a>
+<span class="sourceLineNo">404</span>    // Write the actual file<a name="line.404"></a>
+<span class="sourceLineNo">405</span>    FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.405"></a>
+<span class="sourceLineNo">406</span>    SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.406"></a>
+<span class="sourceLineNo">407</span>      fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.407"></a>
+<span class="sourceLineNo">408</span>      NullWritable.class);<a name="line.408"></a>
+<span class="sourceLineNo">409</span><a name="line.409"></a>
+<span class="sourceLineNo">410</span>    try {<a name="line.410"></a>
+<span class="sourceLineNo">411</span>      for (ImmutableBytesWritable startKey : sorted) {<a name="line.411"></a>
+<span class="sourceLineNo">412</span>        writer.append(startKey, NullWritable.get());<a name="line.412"></a>
+<span class="sourceLineNo">413</span>      }<a name="line.413"></a>
+<span class="sourceLineNo">414</span>    } finally {<a name="line.414"></a>
+<span class="sourceLineNo">415</span>      writer.close();<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>  /**<a name="line.419"></a>
+<span class="sourceLineNo">420</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.420"></a>
+<span class="sourceLineNo">421</span>   * table. This<a name="line.421"></a>
+<span class="sourceLineNo">422</span>   * &lt;ul&gt;<a name="line.422"></a>
+<span class="sourceLineNo">423</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.423"></a>
+<span class="sourceLineNo">424</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.424"></a>
+<span class="sourceLineNo">425</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.425"></a>
+<span class="sourceLineNo">426</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.426"></a>
+<span class="sourceLineNo">427</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.427"></a>
+<span class="sourceLineNo">428</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.428"></a>
+<span class="sourceLineNo">429</span>   * &lt;/ul&gt;<a name="line.429"></a>
+<span class="sourceLineNo">430</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.430"></a>
+<span class="sourceLineNo">431</span>   * running this function.<a name="line.431"></a>
+<span class="sourceLineNo">432</span>   */<a name="line.432"></a>
+<span class="sourceLineNo">433</span>  public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.433"></a>
+<span class="sourceLineNo">434</span>      throws IOException {<a name="line.434"></a>
+<span class="sourceLineNo">435</span>    configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.435"></a>
+<span class="sourceLineNo">436</span>  }<a name="line.436"></a>
+<span class="sourceLineNo">437</span><a name="line.437"></a>
+<span class="sourceLineNo">438</span>  /**<a name="line.438"></a>
+<span class="sourceLineNo">439</span>   * Configure a MapReduce Job to perform an incremental load into the given<a name="line.439"></a>
+<span class="sourceLineNo">440</span>   * table. This<a name="line.440"></a>
+<span class="sourceLineNo">441</span>   * &lt;ul&gt;<a name="line.441"></a>
+<span class="sourceLineNo">442</span>   *   &lt;li&gt;Inspects the table to configure a total order partitioner&lt;/li&gt;<a name="line.442"></a>
+<span class="sourceLineNo">443</span>   *   &lt;li&gt;Uploads the partitions file to the cluster and adds it to the DistributedCache&lt;/li&gt;<a name="line.443"></a>
+<span class="sourceLineNo">444</span>   *   &lt;li&gt;Sets the number of reduce tasks to match the current number of regions&lt;/li&gt;<a name="line.444"></a>
+<span class="sourceLineNo">445</span>   *   &lt;li&gt;Sets the output key/value class to match HFileOutputFormat2's requirements&lt;/li&gt;<a name="line.445"></a>
+<span class="sourceLineNo">446</span>   *   &lt;li&gt;Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.446"></a>
+<span class="sourceLineNo">447</span>   *     PutSortReducer)&lt;/li&gt;<a name="line.447"></a>
+<span class="sourceLineNo">448</span>   * &lt;/ul&gt;<a name="line.448"></a>
+<span class="sourceLineNo">449</span>   * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.449"></a>
+<span class="sourceLineNo">450</span>   * running this function.<a name="line.450"></a>
+<span class="sourceLineNo">451</span>   */<a name="line.451"></a>
+<span class="sourceLineNo">452</span>  public static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.452"></a>
+<span class="sourceLineNo">453</span>      RegionLocator regionLocator) throws IOException {<a name="line.453"></a>
+<span class="sourceLineNo">454</span>    configureIncrementalLoad(job, tableDescriptor, regionLocator, HFileOutputFormat2.class);<a name="line.454"></a>
+<span class="sourceLineNo">455</span>  }<a name="line.455"></a>
+<span class="sourceLineNo">456</span><a name="line.456"></a>
+<span class="sourceLineNo">457</span>  static void configureIncrementalLoad(Job job, HTableDescriptor tableDescriptor,<a name="line.457"></a>
+<span class="sourceLineNo">458</span>      RegionLocator regionLocator, Class&lt;? extends OutputFormat&lt;?, ?&gt;&gt; cls) throws IOException,<a name="line.458"></a>
+<span class="sourceLineNo">459</span>      UnsupportedEncodingException {<a name="line.459"></a>
+<span class="sourceLineNo">460</span>    Configuration conf = job.getConfiguration();<a name="line.460"></a>
+<span class="sourceLineNo">461</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.461"></a>
+<span class="sourceLineNo">462</span>    job.setOutputValueClass(KeyValue.class);<a name="line.462"></a>
+<span class="sourceLineNo">463</span>    job.setOutputFormatClass(cls);<a name="line.463"></a>
+<span class="sourceLineNo">464</span><a name="line.464"></a>
+<span class="sourceLineNo">465</span>    // Based on the configured map output class, set the correct reducer to properly<a name="line.465"></a>
+<span class="sourceLineNo">466</span>    // sort the incoming values.<a name="line.466"></a>
+<span class="sourceLineNo">467</span>    // TODO it would be nice to pick one or the other of these formats.<a name="line.467"></a>
+<span class="sourceLineNo">468</span>    if (KeyValue.class.equals(job.getMapOutputValueClass())) {<a name="line.468"></a>
+<span class="sourceLineNo">469</span>      job.setReducerClass(KeyValueSortReducer.class);<a name="line.469"></a>
+<span class="sourceLineNo">470</span>    } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.470"></a>
+<span class="sourceLineNo">471</span>      job.setReducerClass(PutSortReducer.class);<a name="line.471"></a>
+<span class="sourceLineNo">472</span>    } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.472"></a>
+<span class="sourceLineNo">473</span>      job.setReducerClass(TextSortReducer.class);<a name="line.473"></a>
+<span class="sourceLineNo">474</span>    } else {<a name="line.474"></a>
+<span class="sourceLineNo">475</span>      LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.475"></a>
+<span class="sourceLineNo">476</span>    }<a name="line.476"></a>
+<span class="sourceLineNo">477</span><a name="line.477"></a>
+<span class="sourceLineNo">478</span>    conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.478"></a>
+<span class="sourceLineNo">479</span>        MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.479"></a>
+<span class="sourceLineNo">480</span>        KeyValueSerialization.class.getName());<a name="line.480"></a>
+<span class="sourceLineNo">481</span><a name="line.481"></a>
+<span class="sourceLineNo">482</span>    if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.482"></a>
+<span class="sourceLineNo">483</span>      // record this table name for creating writer by favored nodes<a name="line.483"></a>
+<span class="sourceLineNo">484</span>      LOG.info("bulkload locality sensitive enabled");<a name="line.484"></a>
+<span class="sourceLineNo">485</span>      conf.set(OUTPUT_TABLE_NAME_CONF_KEY, regionLocator.getName().getNameAsString());<a name="line.485"></a>
+<span class="sourceLineNo">486</span>    }<a name="line.486"></a>
+<span class="sourceLineNo">487</span><a name="line.487"></a>
+<span class="sourceLineNo">488</span>    // Use table's region boundaries for TOP split points.<a name="line.488"></a>
+<span class="sourceLineNo">489</span>    LOG.info("Looking up current regions for table " + regionLocator.getName());<a name="line.489"></a>
+<span class="sourceLineNo">490</span>    List&lt;ImmutableBytesWritable&gt; startKeys = getRegionStartKeys(regionLocator);<a name="line.490"></a>
+<span class="sourceLineNo">491</span>    LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.491"></a>
+<span class="sourceLineNo">492</span>        "to match current region count");<a name="line.492"></a>
+<span class="sourceLineNo">493</span>    job.setNumReduceTasks(startKeys.size());<a name="line.493"></a>
+<span class="sourceLineNo">494</span><a name="line.494"></a>
+<span class="sourceLineNo">495</span>    configurePartitioner(job, startKeys);<a name="line.495"></a>
+<span class="sourceLineNo">496</span>    // Set compression algorithms based on column families<a name="line.496"></a>
+<span class="sourceLineNo">497</span>    configureCompression(conf, tableDescriptor);<a name="line.497"></a>
+<span class="sourceLineNo">498</span>    configureBloomType(tableDescriptor, conf);<a name="line.498"></a>
+<span class="sourceLineNo">499</span>    configureBlockSize(tableDescriptor, conf);<a name="line.499"></a>
+<span class="sourceLineNo">500</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.500"></a>
+<span class="sourceLineNo">501</span><a name="line.501"></a>
+<span class="sourceLineNo">502</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.502"></a>
+<span class="sourceLineNo">503</span>    TableMapReduceUtil.initCredentials(job);<a name="line.503"></a>
+<span class="sourceLineNo">504</span>    LOG.info("Incremental table " + regionLocator.getName() + " output configured.");<a name="line.504"></a>
+<span class="sourceLineNo">505</span>  }<a name="line.505"></a>
+<span class="sourceLineNo">506</span>  <a name="line.506"></a>
+<span class="sourceLineNo">507</span>  public static void configureIncrementalLoadMap(Job job, HTableDescriptor tableDescriptor) throws<a name="line.507"></a>
+<span class="sourceLineNo">508</span>      IOException {<a name="line.508"></a>
+<span class="sourceLineNo">509</span>    Configuration conf = job.getConfiguration();<a name="line.509"></a>
+<span class="sourceLineNo">510</span><a name="line.510"></a>
+<span class="sourceLineNo">511</span>    job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.511"></a>
+<span class="sourceLineNo">512</span>    job.setOutputValueClass(KeyValue.class);<a name="line.512"></a>
+<span class="sourceLineNo">513</span>    job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.513"></a>
+<span class="sourceLineNo">514</span><a name="line.514"></a>
+<span class="sourceLineNo">515</span>    // Set compression algorithms based on column families<a name="line.515"></a>
+<span class="sourceLineNo">516</span>    configureCompression(conf, tableDescriptor);<a name="line.516"></a>
+<span class="sourceLineNo">517</span>    configureBloomType(tableDescriptor, conf);<a name="line.517"></a>
+<span class="sourceLineNo">518</span>    configureBlockSize(tableDescriptor, conf);<a name="line.518"></a>
+<span class="sourceLineNo">519</span>    configureDataBlockEncoding(tableDescriptor, conf);<a name="line.519"></a>
 <span class="sourceLineNo">520</span><a name="line.520"></a>
-<span class="sourceLineNo">521</span>  /**<a name="line.521"></a>
-<span class="sourceLineNo">522</span>   * Runs inside the task to deserialize column family to compression algorithm<a name="line.522"></a>
-<span class="sourceLineNo">523</span>   * map from the configuration.<a name="line.523"></a>
-<span class="sourceLineNo">524</span>   *<a name="line.524"></a>
-<span class="sourceLineNo">525</span>   * @param conf to read the serialized values from<a name="line.525"></a>
-<span class="sourceLineNo">526</span>   * @return a map from column family to the configured compression algorithm<a name="line.526"></a>
-<span class="sourceLineNo">527</span>   */<a name="line.527"></a>
-<span class="sourceLineNo">528</span>  @VisibleForTesting<a name="line.528"></a>
-<span class="sourceLineNo">529</span>  static Map&lt;byte[], Algorithm&gt; createFamilyCompressionMap(Configuration<a name="line.529"></a>
-<span class="sourceLineNo">530</span>      conf) {<a name="line.530"></a>
-<span class="sourceLineNo">531</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.531"></a>
-<span class="sourceLineNo">532</span>        COMPRESSION_FAMILIES_CONF_KEY);<a name="line.532"></a>
-<span class="sourceLineNo">533</span>    Map&lt;byte[], Algorithm&gt; compressionMap = new TreeMap&lt;byte[],<a name="line.533"></a>
-<span class="sourceLineNo">534</span>        Algorithm&gt;(Bytes.BYTES_COMPARATOR);<a name="line.534"></a>
-<span class="sourceLineNo">535</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.535"></a>
-<span class="sourceLineNo">536</span>      Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.536"></a>
-<span class="sourceLineNo">537</span>      compressionMap.put(e.getKey(), algorithm);<a name="line.537"></a>
-<span class="sourceLineNo">538</span>    }<a name="line.538"></a>
-<span class="sourceLineNo">539</span>    return compressionMap;<a name="line.539"></a>
-<span class="sourceLineNo">540</span>  }<a name="line.540"></a>
-<span class="sourceLineNo">541</span><a name="line.541"></a>
-<span class="sourceLineNo">542</span>  /**<a name="line.542"></a>
-<span class="sourceLineNo">543</span>   * Runs inside the task to deserialize column family to bloom filter type<a name="line.543"></a>
-<span class="sourceLineNo">544</span>   * map from the configuration.<a name="line.544"></a>
-<span class="sourceLineNo">545</span>   *<a name="line.545"></a>
-<span class="sourceLineNo">546</span>   * @param conf to read the serialized values from<a name="line.546"></a>
-<span class="sourceLineNo">547</span>   * @return a map from column family to the the configured bloom filter type<a name="line.547"></a>
-<span class="sourceLineNo">548</span>   */<a name="line.548"></a>
-<span class="sourceLineNo">549</span>  @VisibleForTesting<a name="line.549"></a>
-<span class="sourceLineNo">550</span>  static Map&lt;byte[], BloomType&gt; createFamilyBloomTypeMap(Configuration conf) {<a name="line.550"></a>
-<span class="sourceLineNo">551</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.551"></a>
-<span class="sourceLineNo">552</span>        BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.552"></a>
-<span class="sourceLineNo">553</span>    Map&lt;byte[], BloomType&gt; bloomTypeMap = new TreeMap&lt;byte[],<a name="line.553"></a>
-<span class="sourceLineNo">554</span>        BloomType&gt;(Bytes.BYTES_COMPARATOR);<a name="line.554"></a>
-<span class="sourceLineNo">555</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.555"></a>
-<span class="sourceLineNo">556</span>      BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.556"></a>
-<span class="sourceLineNo">557</span>      bloomTypeMap.put(e.getKey(), bloomType);<a name="line.557"></a>
-<span class="sourceLineNo">558</span>    }<a name="line.558"></a>
-<span class="sourceLineNo">559</span>    return bloomTypeMap;<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>  /**<a name="line.562"></a>
-<span class="sourceLineNo">563</span>   * Runs inside the task to deserialize column family to block size<a name="line.563"></a>
-<span class="sourceLineNo">564</span>   * map from the configuration.<a name="line.564"></a>
-<span class="sourceLineNo">565</span>   *<a name="line.565"></a>
-<span class="sourceLineNo">566</span>   * @param conf to read the serialized values from<a name="line.566"></a>
-<span class="sourceLineNo">567</span>   * @return a map from column family to the configured block size<a name="line.567"></a>
-<span class="sourceLineNo">568</span>   */<a name="line.568"></a>
-<span class="sourceLineNo">569</span>  @VisibleForTesting<a name="line.569"></a>
-<span class="sourceLineNo">570</span>  static Map&lt;byte[], Integer&gt; createFamilyBlockSizeMap(Configuration conf) {<a name="line.570"></a>
-<span class="sourceLineNo">571</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.571"></a>
-<span class="sourceLineNo">572</span>        BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.572"></a>
-<span class="sourceLineNo">573</span>    Map&lt;byte[], Integer&gt; blockSizeMap = new TreeMap&lt;byte[],<a name="line.573"></a>
-<span class="sourceLineNo">574</span>        Integer&gt;(Bytes.BYTES_COMPARATOR);<a name="line.574"></a>
-<span class="sourceLineNo">575</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.575"></a>
-<span class="sourceLineNo">576</span>      Integer blockSize = Integer.parseInt(e.getValue());<a name="line.576"></a>
-<span class="sourceLineNo">577</span>      blockSizeMap.put(e.getKey(), blockSize);<a name="line.577"></a>
-<span class="sourceLineNo">578</span>    }<a name="line.578"></a>
-<span class="sourceLineNo">579</span>    return blockSizeMap;<a name="line.579"></a>
-<span class="sourceLineNo">580</span>  }<a name="line.580"></a>
-<span class="sourceLineNo">581</span><a name="line.581"></a>
-<span class="sourceLineNo">582</span>  /**<a name="line.582"></a>
-<span class="sourceLineNo">583</span>   * Runs inside the task to deserialize column family to data block encoding<a name="line.583"></a>
-<span class="sourceLineNo">584</span>   * type map from the configuration.<a name="line.584"></a>
-<span class="sourceLineNo">585</span>   *<a name="line.585"></a>
-<span class="sourceLineNo">586</span>   * @param conf to read the serialized values from<a name="line.586"></a>
-<span class="sourceLineNo">587</span>   * @return a map from column family to HFileDataBlockEncoder for the<a name="line.587"></a>
-<span class="sourceLineNo">588</span>   *         configured data block type for the family<a name="line.588"></a>
-<span class="sourceLineNo">589</span>   */<a name="line.589"></a>
-<span class="sourceLineNo">590</span>  @VisibleForTesting<a name="line.590"></a>
-<span class="sourceLineNo">591</span>  static Map&lt;byte[], DataBlockEncoding&gt; createFamilyDataBlockEncodingMap(<a name="line.591"></a>
-<span class="sourceLineNo">592</span>      Configuration conf) {<a name="line.592"></a>
-<span class="sourceLineNo">593</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.593"></a>
-<span class="sourceLineNo">594</span>        DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.594"></a>
-<span class="sourceLineNo">595</span>    Map&lt;byte[], DataBlockEncoding&gt; encoderMap = new TreeMap&lt;byte[],<a name="line.595"></a>
-<span class="sourceLineNo">596</span>        DataBlockEncoding&gt;(Bytes.BYTES_COMPARATOR);<a name="line.596"></a>
-<span class="sourceLineNo">597</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.597"></a>
-<span class="sourceLineNo">598</span>      encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.598"></a>
-<span class="sourceLineNo">599</span>    }<a name="line.599"></a>
-<span class="sourceLineNo">600</span>    return encoderMap;<a name="line.600"></a>
-<span class="sourceLineNo">601</span>  }<a name="line.601"></a>
-<span class="sourceLineNo">602</span><a name="line.602"></a>
-<span class="sourceLineNo">603</span><a name="line.603"></a>
-<span class="sourceLineNo">604</span>  /**<a name="line.604"></a>
-<span class="sourceLineNo">605</span>   * Run inside the task to deserialize column family to given conf value map.<a name="line.605"></a>
-<span class="sourceLineNo">606</span>   *<a name="line.606"></a>
-<span class="sourceLineNo">607</span>   * @param conf to read the serialized values from<a name="line.607"></a>
-<span class="sourceLineNo">608</span>   * @param confName conf key to read from the configuration<a name="line.608"></a>
-<span class="sourceLineNo">609</span>   * @return a map of column family to the given configuration value<a name="line.609"></a>
-<span class="sourceLineNo">610</span>   */<a name="line.610"></a>
-<span class="sourceLineNo">611</span>  private static Map&lt;byte[], String&gt; createFamilyConfValueMap(<a name="line.611"></a>
-<span class="sourceLineNo">612</span>      Configuration conf, String confName) {<a name="line.612"></a>
-<span class="sourceLineNo">613</span>    Map&lt;byte[], String&gt; confValMap = new TreeMap&lt;byte[], String&gt;(Bytes.BYTES_COMPARATOR);<a name="line.613"></a>
-<span class="sourceLineNo">614</span>    String confVal = conf.get(confName, "");<a name="line.614"></a>
-<span class="sourceLineNo">615</span>    for (String familyConf : confVal.split("&amp;")) {<a name="line.615"></a>
-<span class="sourceLineNo">616</span>      String[] familySplit = familyConf.split("=");<a name="line.616"></a>
-<span class="sourceLineNo">617</span>      if (familySplit.length != 2) {<a name="line.617"></a>
-<span class="sourceLineNo">618</span>        continue;<a name="line.618"></a>
-<span class="sourceLineNo">619</span>      }<a name="line.619"></a>
-<span class="sourceLineNo">620</span>      try {<a name="line.620"></a>
-<span class="sourceLineNo">621</span>        confValMap.put(URLDecoder.decode(familySplit[0], "UTF-8").getBytes(),<a name="line.621"></a>
-<span class="sourceLineNo">622</span>            URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.622"></a>
-<span class="sourceLineNo">623</span>      } catch (UnsupportedEncodingException e) {<a name="line.623"></a>
-<span class="sourceLineNo">624</span>        // will not happen with UTF-8 encoding<a name="line.624"></a>
-<span class="sourceLineNo">625</span>        throw new AssertionError(e);<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>    return confValMap;<a name="line.628"></a>
-<span class="sourceLineNo">629</span>  }<a name="line.629"></a>
-<span class="sourceLineNo">630</span><a name="line.630"></a>
-<span class="sourceLineNo">631</span>  /**<a name="line.631"></a>
-<span class="sourceLineNo">632</span>   * Configure &lt;code&gt;job&lt;/code&gt; with a TotalOrderPartitioner, partitioning against<a name="line.632"></a>
-<span class="sourceLineNo">633</span>   * &lt;code&gt;splitPoints&lt;/code&gt;. Cleans up the partitions file after job exists.<a name="line.633"></a>
-<span class="sourceLineNo">634</span>   */<a name="line.634"></a>
-<span class="sourceLineNo">635</span>  static void configurePartitioner(Job job, List&lt;ImmutableBytesWritable&gt; splitPoints)<a name="line.635"></a>
-<span class="sourceLineNo">636</span>      throws IOException {<a name="line.636"></a>
-<span class="sourceLineNo">637</span>    Configuration conf = job.getConfiguration();<a name="line.637"></a>
-<span class="sourceLineNo">638</span>    // create the partitions file<a name="line.638"></a>
-<span class="sourceLineNo">639</span>    FileSystem fs = FileSystem.get(conf);<a name="line.639"></a>
-<span class="sourceLineNo">640</span>    String hbaseTmpFsDir =<a name="line.640"></a>
-<span class="sourceLineNo">641</span>        conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.641"></a>
-<span class="sourceLineNo">642</span>          HConstants.DEFAULT_TEMPORARY_HDFS_DIRECTORY);<a name="line.642"></a>
-<span class="sourceLineNo">643</span>    Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.643"></a>
-<span class="sourceLineNo">644</span>    fs.makeQualified(partitionsPath);<a name="line.644"></a>
-<span class="sourceLineNo">645</span>    writePartitions(conf, partitionsPath, splitPoints);<a name="line.645"></a>
-<span class="sourceLineNo">646</span>    fs.deleteOnExit(partitionsPath);<a name="line.646"></a>
-<span class="sourceLineNo">647</span><a name="line.647"></a>
-<span class="sourceLineNo">648</span>    // configure job to use it<a name="line.648"></a>
-<span class="sourceLineNo">649</span>    job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.649"></a>
-<span class="sourceLineNo">650</span>    TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.650"></a>
-<span class="sourceLineNo">651</span>  }<a name="line.651"></a>
+<span class="sourceLineNo">521</span>    TableMapReduceUtil.addDependencyJars(job);<a name="line.521"></a>
+<span class="sourceLineNo">522</span>    TableMapReduceUtil.initCredentials(job);<a name="line.522"></a>
+<span class="sourceLineNo">523</span>    LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<a name="line.523"></a>
+<span class="sourceLineNo">524</span>  }<a name="line.524"></a>
+<span class="sourceLineNo">525</span><a name="line.525"></a>
+<span class="sourceLineNo">526</span>  /**<a name="line.526"></a>
+<span class="sourceLineNo">527</span>   * Runs inside the task to deserialize column family to compression algorithm<a name="line.527"></a>
+<span class="sourceLineNo">528</span>   * map from the configuration.<a name="line.528"></a>
+<span class="sourceLineNo">529</span>   *<a name="line.529"></a>
+<span class="sourceLineNo">530</span>   * @param conf to read the serialized values from<a name="line.530"></a>
+<span class="sourceLineNo">531</span>   * @return a map from column family to the configured compression algorithm<a name="line.531"></a>
+<span class="sourceLineNo">532</span>   */<a name="line.532"></a>
+<span class="sourceLineNo">533</span>  @VisibleForTesting<a name="line.533"></a>
+<span class="sourceLineNo">534</span>  static Map&lt;byte[], Algorithm&gt; createFamilyCompressionMap(Configuration<a name="line.534"></a>
+<span class="sourceLineNo">535</span>      conf) {<a name="line.535"></a>
+<span class="sourceLineNo">536</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.536"></a>
+<span class="sourceLineNo">537</span>        COMPRESSION_FAMILIES_CONF_KEY);<a name="line.537"></a>
+<span class="sourceLineNo">538</span>    Map&lt;byte[], Algorithm&gt; compressionMap = new TreeMap&lt;byte[],<a name="line.538"></a>
+<span class="sourceLineNo">539</span>        Algorithm&gt;(Bytes.BYTES_COMPARATOR);<a name="line.539"></a>
+<span class="sourceLineNo">540</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.540"></a>
+<span class="sourceLineNo">541</span>      Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.541"></a>
+<span class="sourceLineNo">542</span>      compressionMap.put(e.getKey(), algorithm);<a name="line.542"></a>
+<span class="sourceLineNo">543</span>    }<a name="line.543"></a>
+<span class="sourceLineNo">544</span>    return compressionMap;<a name="line.544"></a>
+<span class="sourceLineNo">545</span>  }<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>   * Runs inside the task to deserialize column family to bloom filter type<a name="line.548"></a>
+<span class="sourceLineNo">549</span>   * map from the configuration.<a name="line.549"></a>
+<span class="sourceLineNo">550</span>   *<a name="line.550"></a>
+<span class="sourceLineNo">551</span>   * @param conf to read the serialized values from<a name="line.551"></a>
+<span class="sourceLineNo">552</span>   * @return a map from column family to the the configured bloom filter type<a name="line.552"></a>
+<span class="sourceLineNo">553</span>   */<a name="line.553"></a>
+<span class="sourceLineNo">554</span>  @VisibleForTesting<a name="line.554"></a>
+<span class="sourceLineNo">555</span>  static Map&lt;byte[], BloomType&gt; createFamilyBloomTypeMap(Configuration conf) {<a name="line.555"></a>
+<span class="sourceLineNo">556</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.556"></a>
+<span class="sourceLineNo">557</span>        BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.557"></a>
+<span class="sourceLineNo">558</span>    Map&lt;byte[], BloomType&gt; bloomTypeMap = new TreeMap&lt;byte[],<a name="line.558"></a>
+<span class="sourceLineNo">559</span>        BloomType&gt;(Bytes.BYTES_COMPARATOR);<a name="line.559"></a>
+<span class="sourceLineNo">560</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.560"></a>
+<span class="sourceLineNo">561</span>      BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.561"></a>
+<span class="sourceLineNo">562</span>      bloomTypeMap.put(e.getKey(), bloomType);<a name="line.562"></a>
+<span class="sourceLineNo">563</span>    }<a name="line.563"></a>
+<span class="sourceLineNo">564</span>    return bloomTypeMap;<a name="line.564"></a>
+<span class="sourceLineNo">565</span>  }<a name="line.565"></a>
+<span class="sourceLineNo">566</span><a name="line.566"></a>
+<span class="sourceLineNo">567</span>  /**<a name="line.567"></a>
+<span class="sourceLineNo">568</span>   * Runs inside the task to deserialize column family to block size<a name="line.568"></a>
+<span class="sourceLineNo">569</span>   * map from the configuration.<a name="line.569"></a>
+<span class="sourceLineNo">570</span>   *<a name="line.570"></a>
+<span class="sourceLineNo">571</span>   * @param conf to read the serialized values from<a name="line.571"></a>
+<span class="sourceLineNo">572</span>   * @return a map from column family to the configured block size<a name="line.572"></a>
+<span class="sourceLineNo">573</span>   */<a name="line.573"></a>
+<span class="sourceLineNo">574</span>  @VisibleForTesting<a name="line.574"></a>
+<span class="sourceLineNo">575</span>  static Map&lt;byte[], Integer&gt; createFamilyBlockSizeMap(Configuration conf) {<a name="line.575"></a>
+<span class="sourceLineNo">576</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.576"></a>
+<span class="sourceLineNo">577</span>        BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.577"></a>
+<span class="sourceLineNo">578</span>    Map&lt;byte[], Integer&gt; blockSizeMap = new TreeMap&lt;byte[],<a name="line.578"></a>
+<span class="sourceLineNo">579</span>        Integer&gt;(Bytes.BYTES_COMPARATOR);<a name="line.579"></a>
+<span class="sourceLineNo">580</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.580"></a>
+<span class="sourceLineNo">581</span>      Integer blockSize = Integer.parseInt(e.getValue());<a name="line.581"></a>
+<span class="sourceLineNo">582</span>      blockSizeMap.put(e.getKey(), blockSize);<a name="line.582"></a>
+<span class="sourceLineNo">583</span>    }<a name="line.583"></a>
+<span class="sourceLineNo">584</span>    return blockSizeMap;<a name="line.584"></a>
+<span class="sourceLineNo">585</span>  }<a name="line.585"></a>
+<span class="sourceLineNo">586</span><a name="line.586"></a>
+<span class="sourceLineNo">587</span>  /**<a name="line.587"></a>
+<span class="sourceLineNo">588</span>   * Runs inside the task to deserialize column family to data block encoding<a name="line.588"></a>
+<span class="sourceLineNo">589</span>   * type map from the configuration.<a name="line.589"></a>
+<span class="sourceLineNo">590</span>   *<a name="line.590"></a>
+<span class="sourceLineNo">591</span>   * @param conf to read the serialized values from<a name="line.591"></a>
+<span class="sourceLineNo">592</span>   * @return a map from column family to HFileDataBlockEncoder for the<a name="line.592"></a>
+<span class="sourceLineNo">593</span>   *         configured data block type for the family<a name="line.593"></a>
+<span class="sourceLineNo">594</span>   */<a name="line.594"></a>
+<span class="sourceLineNo">595</span>  @VisibleForTesting<a name="line.595"></a>
+<span class="sourceLineNo">596</span>  static Map&lt;byte[], DataBlockEncoding&gt; createFamilyDataBlockEncodingMap(<a name="line.596"></a>
+<span class="sourceLineNo">597</span>      Configuration conf) {<a name="line.597"></a>
+<span class="sourceLineNo">598</span>    Map&lt;byte[], String&gt; stringMap = createFamilyConfValueMap(conf,<a name="line.598"></a>
+<span class="sourceLineNo">599</span>        DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.599"></a>
+<span class="sourceLineNo">600</span>    Map&lt;byte[], DataBlockEncoding&gt; encoderMap = new TreeMap&lt;byte[],<a name="line.600"></a>
+<span class="sourceLineNo">601</span>        DataBlockEncoding&gt;(Bytes.BYTES_COMPARATOR);<a name="line.601"></a>
+<span class="sourceLineNo">602</span>    for (Map.Entry&lt;byte[], String&gt; e : stringMap.entrySet()) {<a name="line.602"></a>
+<span class="sourceLineNo">603</span>      encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.603"></a>
+<span class="sourceLineNo">604</span>    }<a name="line.604"></a>
+<span class="sourceLineNo">605</span>    return encoderMap;<a name="line.605"></a>
+<span class="sourceLineNo">606</span>  }<a name="line.606"></a>
+<span class="sourceLineNo">607</span><a name="line.607"></a>
+<span class="sourceLineNo">608</span><a name="line.608"></a>
+<span class="sourceLineNo">609</span>  /**<a name="line.609"></a>
+<span class="sourceLineNo">610</span>   * Run inside the task to deserialize column family to given conf value map.<a name="line.610"></a>
+<span class="sourceLineNo">611</span>   *<a name="line.611"></a>
+<span class="sourceLineNo">612</span>   * @param conf to read the serialized values from<a name="line.612"></a>
+<span class="sourceLineNo">613</span>   * @param confName conf key to read from the configuration<a name="line.613"></a>
+<span class="sourceLineNo">614</span>   * @return a map of column family to the given configuration value<a name="line.614"></a>
+<span class="sourceLineNo">615</span>   */<a name="line.615"></a>
+<span class="sourceLineNo">616</span>  private static Map&lt;byte[], String&gt; createFamilyConfValueMap(<a name="line.616"></a>
+<span class="sourceLineNo">617</span>      Configuration conf, String confName) {<a name="line.617"></a>
+<span class="sourceLineNo">618</span>    Map&lt;byte[], String&gt; confValMap = new TreeMap&lt;byte[], String&gt;(Bytes.BYTES_COMPARATOR);<a name="line.618"></a>
+<span class="sourceLineNo">619</span>    String confVal = conf.get(confName, "");<a name="line.619"></a>
+<span class="sourceLineNo">620</span>    for (String familyConf : confVal.split("&amp;")) {<a name="line.620"></a>
+<span class="sourceLineNo">621</span>      String[] familySplit = familyConf.split("=");<a name="line.621"></a>
+<span class="sourceLineNo">622</span>      if (familySplit.length != 2) {<a name="line.622"></a>
+<span class="sourceLineNo">623</span>        continue;<a name="line.623"></a>
+<span class="sourceLineNo">624</span>      }<a name="line.624"></a>
+<span class="sourceLineNo">625</span>      try {<a name="line.625"></a>
+<span class="sourceLineNo">626</span>        confValMap.put(URLDecoder.decode(familySplit[0], "UTF-8").getBytes(),<a name="line.626"></a>
+<span class="sourceLineNo">627</span>            URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.627"></a>
+<span class="sourceLineNo">628</span>      } catch (UnsupportedEncodingException e) {<a name="line.628"></a>
+<span class="sourceLineNo">629</span>        // will not happen with UTF-8 encoding<a name="line.629"></a>
+<span class="sourceLineNo">630</span>        throw new AssertionError(e);<a name="line.630"></a>
+<span class="sourceLineNo">631</span>      }<a name="line.631"></a>
+<span class="sourceLineNo">632</span>    }<a name="line.632"></a>
+<span class="sourceLineNo">633</span>    return confValMap;<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>  /**<a name="line.636"></a>
+<span class="sourceLineNo">637</span>   * Configure &lt;code&gt;job&lt;/code&gt; with a TotalOrderPartitioner, partitioning against<a name="line.637"></a>
+<span class="sourceLineNo">638</span>   * &lt;code&gt;splitPoints&lt;/code&gt;. Cleans up the partitions file after job exists.<a name="line.638"></a>
+<span class="sourceLineNo">639</span>   */<a name="line.639"></a>
+<span class="sourceLineNo">640</span>  static void configurePartitioner(Job job, List&lt;ImmutableBytesWritable&gt; splitPoints)<a name="line.640"></a>
+<span class="sourceLineNo">641</span>      throws IOException {<a name="line.641"></a>
+<span class="sourceLineNo">642</span>    Configuration conf = job.getConfiguration();<a name="line.642"></a>
+<span class="sourceLineNo">643</span>    // create the partitions file<a name="line.643"></a>
+<span class="sourceLineNo">644</span>    FileSystem fs = FileSystem.get(conf);<a name="line.644"></a>
+<span class="sourceLineNo">645</span>    String hbaseTmpFsDir =<a name="line.645"></a>
+<span class="sourceLineNo">646</span>        conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.646"></a>
+<span class="sourceLineNo">647</span>          HConstants.DEFAULT_TEMPORARY_HDFS_DIRECTORY);<a name="line.647"></a>
+<span class="sourceLineNo">648</span>    Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.648"></a>
+<span class="sourceLineNo">649</span>    fs.makeQualified(partitionsPath);<a name="line.649"></a>
+<span class="sourceLineNo">650</span>    writePartitions(conf, partitionsPath, splitPoints);<a name="line.650"></a>
+<span class="sourceLineNo">651</span>    fs.deleteOnExit(partitionsPath);<a name="line.651"></a>
 <span class="sourceLineNo">652</span><a name="line.652"></a>
-<span class="sourceLineNo">653</span>  /**<a name="line.653"></a>
-<span class="sourceLineNo">654</span>   * Serialize column family to compression algorithm map to configuration.<a name="line.654"></a>
-<span class="sourceLineNo">655</span>   * Invoked while configuring the MR job for incremental load.<a name="line.655"></a>
-<span class="sourceLineNo">656</span>   *<a name="line.656"></a>
-<span class="sourceLineNo">657</span>   * @param tableDescriptor to read the properties from<a name="line.657"></a>
-<span class="sourceLineNo">658</span>   * @param conf to persist serialized values into<a name="line.658"></a>
-<span class="sourceLineNo">659</span>   * @throws IOException<a name="line.659"></a>
-<span class="sourceLineNo">660</span>   *           on failure to read column family descriptors<a name="line.660"></a>
-<span class="sourceLineNo">661</span>   */<a name="line.661"></a>
-<span class="sourceLineNo">662</span>  @edu.umd.cs.findbugs.annotations.SuppressWarnings(<a name="line.662"></a>
-<span class="sourceLineNo">663</span>      value="RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.663"></a>
-<span class="sourceLineNo">664</span>  @VisibleForTesting<a name="line.664"></a>
-<span class="sourceLineNo">665</span>  static void configureCompression(Configuration conf, HTableDescriptor tableDescriptor)<a name="line.665"></a>
-<span class="sourceLineNo">666</span>      throws UnsupportedEncodingException {<a name="line.666"></a>
-<span class="sourceLineNo">667</span>    StringBuilder compressionConfigValue = new StringBuilder();<a name="line.667"></a>
-<span class="sourceLineNo">668</span>    if(tableDescriptor == null){<a name="line.668"></a>
-<span class="sourceLineNo">669</span>      // could happen with mock table instance<a name="line.669"></a>
-<span class="sourceLineNo">670</span>      return;<a name="line.670"></a>
-<span class="sourceLineNo">671</span>    }<a name="line.671"></a>
-<span class="sourceLineNo">672</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.672"></a>
-<span class="sourceLineNo">673</span>    int i = 0;<a name="line.673"></a>
-<span class="sourceLineNo">674</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.674"></a>
-<span class="sourceLineNo">675</span>      if (i++ &gt; 0) {<a name="line.675"></a>
-<span class="sourceLineNo">676</span>        compressionConfigValue.append('&amp;');<a name="line.676"></a>
-<span class="sourceLineNo">677</span>      }<a name="line.677"></a>
-<span class="sourceLineNo">678</span>      compressionConfigValue.append(URLEncoder.encode(<a name="line.678"></a>
-<span class="sourceLineNo">679</span>        familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.679"></a>
-<span class="sourceLineNo">680</span>      compressionConfigValue.append('=');<a name="line.680"></a>
-<span class="sourceLineNo">681</span>      compressionConfigValue.append(URLEncoder.encode(<a name="line.681"></a>
-<span class="sourceLineNo">682</span>        familyDescriptor.getCompressionType().getName(), "UTF-8"));<a name="line.682"></a>
-<span class="sourceLineNo">683</span>    }<a name="line.683"></a>
-<span class="sourceLineNo">684</span>    // Get rid of the last ampersand<a name="line.684"></a>
-<span class="sourceLineNo">685</span>    conf.set(COMPRESSION_FAMILIES_CONF_KEY, compressionConfigValue.toString());<a name="line.685"></a>
-<span class="sourceLineNo">686</span>  }<a name="line.686"></a>
-<span class="sourceLineNo">687</span><a name="line.687"></a>
-<span class="sourceLineNo">688</span>  /**<a name="line.688"></a>
-<span class="sourceLineNo">689</span>   * Serialize column family to block size map to configuration.<a name="line.689"></a>
-<span class="sourceLineNo">690</span>   * Invoked while configuring the MR job for incremental load.<a name="line.690"></a>
-<span class="sourceLineNo">691</span>   * @param tableDescriptor to read the properties from<a name="line.691"></a>
-<span class="sourceLineNo">692</span>   * @param conf to persist serialized values into<a name="line.692"></a>
-<span class="sourceLineNo">693</span>   *<a name="line.693"></a>
-<span class="sourceLineNo">694</span>   * @throws IOException<a name="line.694"></a>
-<span class="sourceLineNo">695</span>   *           on failure to read column family descriptors<a name="line.695"></a>
-<span class="sourceLineNo">696</span>   */<a name="line.696"></a>
-<span class="sourceLineNo">697</span>  @VisibleForTesting<a name="line.697"></a>
-<span class="sourceLineNo">698</span>  static void configureBlockSize(HTableDescriptor tableDescriptor, Configuration conf)<a name="line.698"></a>
-<span class="sourceLineNo">699</span>      throws UnsupportedEncodingException {<a name="line.699"></a>
-<span class="sourceLineNo">700</span>    StringBuilder blockSizeConfigValue = new StringBuilder();<a name="line.700"></a>
-<span class="sourceLineNo">701</span>    if (tableDescriptor == null) {<a name="line.701"></a>
-<span class="sourceLineNo">702</span>      // could happen with mock table instance<a name="line.702"></a>
-<span class="sourceLineNo">703</span>      return;<a name="line.703"></a>
-<span class="sourceLineNo">704</span>    }<a name="line.704"></a>
-<span class="sourceLineNo">705</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.705"></a>
-<span class="sourceLineNo">706</span>    int i = 0;<a name="line.706"></a>
-<span class="sourceLineNo">707</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.707"></a>
-<span class="sourceLineNo">708</span>      if (i++ &gt; 0) {<a name="line.708"></a>
-<span class="sourceLineNo">709</span>        blockSizeConfigValue.append('&amp;');<a name="line.709"></a>
-<span class="sourceLineNo">710</span>      }<a name="line.710"></a>
-<span class="sourceLineNo">711</span>      blockSizeConfigValue.append(URLEncoder.encode(<a name="line.711"></a>
-<span class="sourceLineNo">712</span>          familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.712"></a>
-<span class="sourceLineNo">713</span>      blockSizeConfigValue.append('=');<a name="line.713"></a>
-<span class="sourceLineNo">714</span>      blockSizeConfigValue.append(URLEncoder.encode(<a name="line.714"></a>
-<span class="sourceLineNo">715</span>          String.valueOf(familyDescriptor.getBlocksize()), "UTF-8"));<a name="line.715"></a>
-<span class="sourceLineNo">716</span>    }<a name="line.716"></a>
-<span class="sourceLineNo">717</span>    // Get rid of the last ampersand<a name="line.717"></a>
-<span class="sourceLineNo">718</span>    conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, blockSizeConfigValue.toString());<a name="line.718"></a>
-<span class="sourceLineNo">719</span>  }<a name="line.719"></a>
-<span class="sourceLineNo">720</span><a name="line.720"></a>
-<span class="sourceLineNo">721</span>  /**<a name="line.721"></a>
-<span class="sourceLineNo">722</span>   * Serialize column family to bloom type map to configuration.<a name="line.722"></a>
-<span class="sourceLineNo">723</span>   * Invoked while configuring the MR job for incremental load.<a name="line.723"></a>
-<span class="sourceLineNo">724</span>   * @param tableDescriptor to read the properties from<a name="line.724"></a>
-<span class="sourceLineNo">725</span>   * @param conf to persist serialized values into<a name="line.725"></a>
-<span class="sourceLineNo">726</span>   *<a name="line.726"></a>
-<span class="sourceLineNo">727</span>   * @throws IOException<a name="line.727"></a>
-<span class="sourceLineNo">728</span>   *           on failure to read column family descriptors<a name="line.728"></a>
-<span class="sourceLineNo">729</span>   */<a name="line.729"></a>
-<span class="sourceLineNo">730</span>  @VisibleForTesting<a name="line.730"></a>
-<span class="sourceLineNo">731</span>  static void configureBloomType(HTableDescriptor tableDescriptor, Configuration conf)<a name="line.731"></a>
-<span class="sourceLineNo">732</span>      throws UnsupportedEncodingException {<a name="line.732"></a>
-<span class="sourceLineNo">733</span>    if (tableDescriptor == null) {<a name="line.733"></a>
-<span class="sourceLineNo">734</span>      // could happen with mock table instance<a name="line.734"></a>
-<span class="sourceLineNo">735</span>      return;<a name="line.735"></a>
-<span class="sourceLineNo">736</span>    }<a name="line.736"></a>
-<span class="sourceLineNo">737</span>    StringBuilder bloomTypeConfigValue = new StringBuilder();<a name="line.737"></a>
-<span class="sourceLineNo">738</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.738"></a>
-<span class="sourceLineNo">739</span>    int i = 0;<a name="line.739"></a>
-<span class="sourceLineNo">740</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.740"></a>
-<span class="sourceLineNo">741</span>      if (i++ &gt; 0) {<a name="line.741"></a>
-<span class="sourceLineNo">742</span>        bloomTypeConfigValue.append('&amp;');<a name="line.742"></a>
-<span class="sourceLineNo">743</span>      }<a name="line.743"></a>
-<span class="sourceLineNo">744</span>      bloomTypeConfigValue.append(URLEncoder.encode(<a name="line.744"></a>
-<span class="sourceLineNo">745</span>        familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.745"></a>
-<span class="sourceLineNo">746</span>      bloomTypeConfigValue.append('=');<a name="line.746"></a>
-<span class="sourceLineNo">747</span>      String bloomType = familyDescriptor.getBloomFilterType().toString();<a name="line.747"></a>
-<span class="sourceLineNo">748</span>      if (bloomType == null) {<a name="line.748"></a>
-<span class="sourceLineNo">749</span>        bloomType = HColumnDescriptor.DEFAULT_BLOOMFILTER;<a name="line.749"></a>
-<span class="sourceLineNo">750</span>      }<a name="line.750"></a>
-<span class="sourceLineNo">751</span>      bloomTypeConfigValue.append(URLEncoder.encode(bloomType, "UTF-8"));<a name="line.751"></a>
-<span class="sourceLineNo">752</span>    }<a name="line.752"></a>
-<span class="sourceLineNo">753</span>    conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, bloomTypeConfigValue.toString());<a name="line.753"></a>
-<span class="sourceLineNo">754</span>  }<a name="line.754"></a>
-<span class="sourceLineNo">755</span><a name="line.755"></a>
-<span class="sourceLineNo">756</span>  /**<a name="line.756"></a>
-<span class="sourceLineNo">757</span>   * Serialize column family to data block encoding map to configuration.<a name="line.757"></a>
-<span class="sourceLineNo">758</span>   * Invoked while configuring the MR job for incremental load.<a name="line.758"></a>
-<span class="sourceLineNo">759</span>   *<a name="line.759"></a>
-<span class="sourceLineNo">760</span>   * @param tableDescriptor to read the properties from<a name="line.760"></a>
-<span class="sourceLineNo">761</span>   * @param conf to persist serialized values into<a name="line.761"></a>
-<span class="sourceLineNo">762</span>   * @throws IOException<a name="line.762"></a>
-<span class="sourceLineNo">763</span>   *           on failure to read column family descriptors<a name="line.763"></a>
-<span class="sourceLineNo">764</span>   */<a name="line.764"></a>
-<span class="sourceLineNo">765</span>  @VisibleForTesting<a name="line.765"></a>
-<span class="sourceLineNo">766</span>  static void configureDataBlockEncoding(HTableDescriptor tableDescriptor,<a name="line.766"></a>
-<span class="sourceLineNo">767</span>      Configuration conf) throws UnsupportedEncodingException {<a name="line.767"></a>
-<span class="sourceLineNo">768</span>    if (tableDescriptor == null) {<a name="line.768"></a>
-<span class="sourceLineNo">769</span>      // could happen with mock table instance<a name="line.769"></a>
-<span class="sourceLineNo">770</span>      return;<a name="line.770"></a>
-<span class="sourceLineNo">771</span>    }<a name="line.771"></a>
-<span class="sourceLineNo">772</span>    StringBuilder dataBlockEncodingConfigValue = new StringBuilder();<a name="line.772"></a>
-<span class="sourceLineNo">773</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.773"></a>
-<span class="sourceLineNo">774</span>    int i = 0;<a name="line.774"></a>
-<span class="sourceLineNo">775</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.775"></a>
-<span class="sourceLineNo">776</span>      if (i++ &gt; 0) {<a name="line.776"></a>
-<span class="sourceLineNo">777</span>        dataBlockEncodingConfigValue.append('&amp;');<a name="line.777"></a>
-<span class="sourceLineNo">778</span>      }<a name="line.778"></a>
-<span class="sourceLineNo">779</span>      dataBlockEncodingConfigValue.append(<a name="line.779"></a>
-<span class="sourceLineNo">780</span>          URLEncoder.encode(familyDescriptor.getNameAsString(), "UTF-8"));<a name="line.780"></a>
-<span class="sourceLineNo">781</span>      dataBlockEncodingConfigValue.append('=');<a name="line.781"></a>
-<span class="sourceLineNo">782</span>      DataBlockEncoding encoding = familyDescriptor.getDataBlockEncoding();<a name="line.782"></a>
-<span class="sourceLineNo">783</span>      if (encoding == null) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span>        encoding = DataBlockEncoding.NONE;<a name="line.784"></a>
-<span class="sourceLineNo">785</span>      }<a name="line.785"></a>
-<span class="sourceLineNo">786</span>      dataBlockEncodingConfigValue.append(URLEncoder.encode(encoding.toString(),<a name="line.786"></a>
-<span class="sourceLineNo">787</span>          "UTF-8"));<a name="line.787"></a>
-<span class="sourceLineNo">788</span>    }<a name="line.788"></a>
-<span class="sourceLineNo">789</span>    conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.789"></a>
-<span class="sourceLineNo">790</span>        dataBlockEncodingConfigValue.toString());<a name="line.790"></a>
-<span class="sourceLineNo">791</span>  }<a name="line.791"></a>
-<span class="sourceLineNo">792</span>}<a name="line.792"></a>
+<span class="sourceLineNo">653</span>    // configure job to use it<a name="line.653"></a>
+<span class="sourceLineNo">654</span>    job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.654"></a>
+<span class="sourceLineNo">655</span>    TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.655"></a>
+<span class="sourceLineNo">656</span>  }<a name="line.656"></a>
+<span class="sourceLineNo">657</span><a name="line.657"></a>
+<span class="sourceLineNo">658</span>  /**<a name="line.658"></a>
+<span class="sourceLineNo">659</span>   * Serialize column family to compression algorithm map to configuration.<a name="line.659"></a>
+<span class="sourceLineNo">660</span>   * Invoked while configuring the MR job for incremental load.<a name="line.660"></a>
+<span class="sourceLineNo">661</span>   *<a name="line.661"></a>
+<span class="sourceLineNo">662</span>   * @param tableDescriptor to read the properties from<a name="line.662"></a>
+<span class="sourceLineNo">663</span>   * @param conf to persist serialized values into<a name="line.663"></a>
+<span class="sourceLineNo">664</span>   * @throws IOException<a name="line.664"></a>
+<span class="sourceLineNo">665</span>   *           on failure to read column family descriptors<a name="line.665"></a>
+<span class="sourceLineNo">666</span>   */<a name="line.666"></a>
+<span class="sourceLineNo">667</span>  @edu.umd.cs.findbugs.annotations.SuppressWarnings(<a name="line.667"></a>
+<span class="sourceLineNo">668</span>      value="RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.668"></a>
+<span class="sourceLineNo">669</span>  @VisibleForTesting<a name="line.669"></a>
+<span class="sourceLineNo">670</span>  static void configureCompression(Configuration conf, HTableDescriptor tableDescriptor)<a name="line.670"></a>
+<span class="sourceLineNo">671</span>      throws UnsupportedEncodingException {<a name="line.671"></a>
+<span class="sourceLineNo">672</span>    StringBuilder compressionConfigValue = new StringBuilder();<a name="line.672"></a>
+<span class="sourceLineNo">673</span>    if(tableDescriptor == null){<a name="line.673"></a>
+<span class="sourceLineNo">674</span>      // could happen with mock table instance<a name="line.674"></a>
+<span class="sourceLineNo">675</span>      return;<a name="line.675"></a>
+<span class="sourceLineNo">676</span>    }<a name="line.676"></a>
+<span class="sourceLineNo">677</span>    Collection&lt;HColumnDescriptor&gt; families = tableDescriptor.getFamilies();<a name="line.677"></a>
+<span class="sourceLineNo">678</span>    int i = 0;<a name="line.678"></a>
+<span class="sourceLineNo">679</span>    for (HColumnDescriptor familyDescriptor : families) {<a name="line.679"></a>
+<span class="sourceLineNo">680</span>      if (i++ &gt; 0) {<a name="line.680"></a>
+<span class="sourceLineNo">681</span>        compressionConfigValue.append('&amp;');<a name="line.681"></a>
+<span class="sourceLineNo">682</span>      }<a name="line.682"></a>
+<span class="sourceLineNo">683</span>      compressionConfigValue.ap

<TRUNCATED>

Mime
View raw message