You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by gi...@apache.org on 2018/12/12 14:52:40 UTC
[24/41] hbase-site git commit: Published site at
67d6d5084cf8fc094cda4bd3f091d8a0a9cb1d3e.
http://git-wip-us.apache.org/repos/asf/hbase-site/blob/d702fb71/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.WriterLength.html
----------------------------------------------------------------------
diff --git a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.WriterLength.html b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.WriterLength.html
index 5ee521a..68dec5a 100644
--- a/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.WriterLength.html
+++ b/devapidocs/src-html/org/apache/hadoop/hbase/mapreduce/HFileOutputFormat2.WriterLength.html
@@ -422,572 +422,570 @@
<span class="sourceLineNo">414</span> DataBlockEncoding encoding = overriddenEncoding;<a name="line.414"></a>
<span class="sourceLineNo">415</span> encoding = encoding == null ? datablockEncodingMap.get(tableAndFamily) : encoding;<a name="line.415"></a>
<span class="sourceLineNo">416</span> encoding = encoding == null ? DataBlockEncoding.NONE : encoding;<a name="line.416"></a>
-<span class="sourceLineNo">417</span> Configuration tempConf = new Configuration(conf);<a name="line.417"></a>
-<span class="sourceLineNo">418</span> tempConf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0.0f);<a name="line.418"></a>
-<span class="sourceLineNo">419</span> HFileContextBuilder contextBuilder = new HFileContextBuilder()<a name="line.419"></a>
-<span class="sourceLineNo">420</span> .withCompression(compression)<a name="line.420"></a>
-<span class="sourceLineNo">421</span> .withChecksumType(HStore.getChecksumType(conf))<a name="line.421"></a>
-<span class="sourceLineNo">422</span> .withBytesPerCheckSum(HStore.getBytesPerChecksum(conf))<a name="line.422"></a>
-<span class="sourceLineNo">423</span> .withBlockSize(blockSize);<a name="line.423"></a>
-<span class="sourceLineNo">424</span><a name="line.424"></a>
-<span class="sourceLineNo">425</span> if (HFile.getFormatVersion(conf) >= HFile.MIN_FORMAT_VERSION_WITH_TAGS) {<a name="line.425"></a>
-<span class="sourceLineNo">426</span> contextBuilder.withIncludesTags(true);<a name="line.426"></a>
-<span class="sourceLineNo">427</span> }<a name="line.427"></a>
-<span class="sourceLineNo">428</span><a name="line.428"></a>
-<span class="sourceLineNo">429</span> contextBuilder.withDataBlockEncoding(encoding);<a name="line.429"></a>
-<span class="sourceLineNo">430</span> HFileContext hFileContext = contextBuilder.build();<a name="line.430"></a>
-<span class="sourceLineNo">431</span> if (null == favoredNodes) {<a name="line.431"></a>
-<span class="sourceLineNo">432</span> wl.writer =<a name="line.432"></a>
-<span class="sourceLineNo">433</span> new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), fs)<a name="line.433"></a>
-<span class="sourceLineNo">434</span> .withOutputDir(familydir).withBloomType(bloomType)<a name="line.434"></a>
-<span class="sourceLineNo">435</span> .withComparator(CellComparator.getInstance()).withFileContext(hFileContext).build();<a name="line.435"></a>
-<span class="sourceLineNo">436</span> } else {<a name="line.436"></a>
-<span class="sourceLineNo">437</span> wl.writer =<a name="line.437"></a>
-<span class="sourceLineNo">438</span> new StoreFileWriter.Builder(conf, new CacheConfig(tempConf), new HFileSystem(fs))<a name="line.438"></a>
-<span class="sourceLineNo">439</span> .withOutputDir(familydir).withBloomType(bloomType)<a name="line.439"></a>
-<span class="sourceLineNo">440</span> .withComparator(CellComparator.getInstance()).withFileContext(hFileContext)<a name="line.440"></a>
-<span class="sourceLineNo">441</span> .withFavoredNodes(favoredNodes).build();<a name="line.441"></a>
-<span class="sourceLineNo">442</span> }<a name="line.442"></a>
-<span class="sourceLineNo">443</span><a name="line.443"></a>
-<span class="sourceLineNo">444</span> this.writers.put(tableAndFamily, wl);<a name="line.444"></a>
-<span class="sourceLineNo">445</span> return wl;<a name="line.445"></a>
-<span class="sourceLineNo">446</span> }<a name="line.446"></a>
-<span class="sourceLineNo">447</span><a name="line.447"></a>
-<span class="sourceLineNo">448</span> private void close(final StoreFileWriter w) throws IOException {<a name="line.448"></a>
-<span class="sourceLineNo">449</span> if (w != null) {<a name="line.449"></a>
-<span class="sourceLineNo">450</span> w.appendFileInfo(BULKLOAD_TIME_KEY,<a name="line.450"></a>
-<span class="sourceLineNo">451</span> Bytes.toBytes(System.currentTimeMillis()));<a name="line.451"></a>
-<span class="sourceLineNo">452</span> w.appendFileInfo(BULKLOAD_TASK_KEY,<a name="line.452"></a>
-<span class="sourceLineNo">453</span> Bytes.toBytes(context.getTaskAttemptID().toString()));<a name="line.453"></a>
-<span class="sourceLineNo">454</span> w.appendFileInfo(MAJOR_COMPACTION_KEY,<a name="line.454"></a>
-<span class="sourceLineNo">455</span> Bytes.toBytes(true));<a name="line.455"></a>
-<span class="sourceLineNo">456</span> w.appendFileInfo(EXCLUDE_FROM_MINOR_COMPACTION_KEY,<a name="line.456"></a>
-<span class="sourceLineNo">457</span> Bytes.toBytes(compactionExclude));<a name="line.457"></a>
-<span class="sourceLineNo">458</span> w.appendTrackedTimestampsToMetadata();<a name="line.458"></a>
-<span class="sourceLineNo">459</span> w.close();<a name="line.459"></a>
-<span class="sourceLineNo">460</span> }<a name="line.460"></a>
-<span class="sourceLineNo">461</span> }<a name="line.461"></a>
-<span class="sourceLineNo">462</span><a name="line.462"></a>
-<span class="sourceLineNo">463</span> @Override<a name="line.463"></a>
-<span class="sourceLineNo">464</span> public void close(TaskAttemptContext c)<a name="line.464"></a>
-<span class="sourceLineNo">465</span> throws IOException, InterruptedException {<a name="line.465"></a>
-<span class="sourceLineNo">466</span> for (WriterLength wl: this.writers.values()) {<a name="line.466"></a>
-<span class="sourceLineNo">467</span> close(wl.writer);<a name="line.467"></a>
-<span class="sourceLineNo">468</span> }<a name="line.468"></a>
-<span class="sourceLineNo">469</span> }<a name="line.469"></a>
-<span class="sourceLineNo">470</span> };<a name="line.470"></a>
-<span class="sourceLineNo">471</span> }<a name="line.471"></a>
-<span class="sourceLineNo">472</span><a name="line.472"></a>
-<span class="sourceLineNo">473</span> /**<a name="line.473"></a>
-<span class="sourceLineNo">474</span> * Configure block storage policy for CF after the directory is created.<a name="line.474"></a>
-<span class="sourceLineNo">475</span> */<a name="line.475"></a>
-<span class="sourceLineNo">476</span> static void configureStoragePolicy(final Configuration conf, final FileSystem fs,<a name="line.476"></a>
-<span class="sourceLineNo">477</span> byte[] tableAndFamily, Path cfPath) {<a name="line.477"></a>
-<span class="sourceLineNo">478</span> if (null == conf || null == fs || null == tableAndFamily || null == cfPath) {<a name="line.478"></a>
-<span class="sourceLineNo">479</span> return;<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> String policy =<a name="line.482"></a>
-<span class="sourceLineNo">483</span> conf.get(STORAGE_POLICY_PROPERTY_CF_PREFIX + Bytes.toString(tableAndFamily),<a name="line.483"></a>
-<span class="sourceLineNo">484</span> conf.get(STORAGE_POLICY_PROPERTY));<a name="line.484"></a>
-<span class="sourceLineNo">485</span> FSUtils.setStoragePolicy(fs, cfPath, policy);<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> /*<a name="line.488"></a>
-<span class="sourceLineNo">489</span> * Data structure to hold a Writer and amount of data written on it.<a name="line.489"></a>
-<span class="sourceLineNo">490</span> */<a name="line.490"></a>
-<span class="sourceLineNo">491</span> static class WriterLength {<a name="line.491"></a>
-<span class="sourceLineNo">492</span> long written = 0;<a name="line.492"></a>
-<span class="sourceLineNo">493</span> StoreFileWriter writer = null;<a name="line.493"></a>
-<span class="sourceLineNo">494</span> }<a name="line.494"></a>
-<span class="sourceLineNo">495</span><a name="line.495"></a>
-<span class="sourceLineNo">496</span> /**<a name="line.496"></a>
-<span class="sourceLineNo">497</span> * Return the start keys of all of the regions in this table,<a name="line.497"></a>
-<span class="sourceLineNo">498</span> * as a list of ImmutableBytesWritable.<a name="line.498"></a>
-<span class="sourceLineNo">499</span> */<a name="line.499"></a>
-<span class="sourceLineNo">500</span> private static List<ImmutableBytesWritable> getRegionStartKeys(List<RegionLocator> regionLocators,<a name="line.500"></a>
-<span class="sourceLineNo">501</span> boolean writeMultipleTables)<a name="line.501"></a>
-<span class="sourceLineNo">502</span> throws IOException {<a name="line.502"></a>
-<span class="sourceLineNo">503</span><a name="line.503"></a>
-<span class="sourceLineNo">504</span> ArrayList<ImmutableBytesWritable> ret = new ArrayList<>();<a name="line.504"></a>
-<span class="sourceLineNo">505</span> for(RegionLocator regionLocator : regionLocators)<a name="line.505"></a>
-<span class="sourceLineNo">506</span> {<a name="line.506"></a>
-<span class="sourceLineNo">507</span> TableName tableName = regionLocator.getName();<a name="line.507"></a>
-<span class="sourceLineNo">508</span> LOG.info("Looking up current regions for table " + tableName);<a name="line.508"></a>
-<span class="sourceLineNo">509</span> byte[][] byteKeys = regionLocator.getStartKeys();<a name="line.509"></a>
-<span class="sourceLineNo">510</span> for (byte[] byteKey : byteKeys) {<a name="line.510"></a>
-<span class="sourceLineNo">511</span> byte[] fullKey = byteKey; //HFileOutputFormat2 use case<a name="line.511"></a>
-<span class="sourceLineNo">512</span> if (writeMultipleTables)<a name="line.512"></a>
-<span class="sourceLineNo">513</span> {<a name="line.513"></a>
-<span class="sourceLineNo">514</span> //MultiTableHFileOutputFormat use case<a name="line.514"></a>
-<span class="sourceLineNo">515</span> fullKey = combineTableNameSuffix(tableName.getName(), byteKey);<a name="line.515"></a>
-<span class="sourceLineNo">516</span> }<a name="line.516"></a>
-<span class="sourceLineNo">517</span> if (LOG.isDebugEnabled()) {<a name="line.517"></a>
-<span class="sourceLineNo">518</span> LOG.debug("SplitPoint startkey for table [" + tableName + "]: [" + Bytes.toStringBinary<a name="line.518"></a>
-<span class="sourceLineNo">519</span> (fullKey) + "]");<a name="line.519"></a>
-<span class="sourceLineNo">520</span> }<a name="line.520"></a>
-<span class="sourceLineNo">521</span> ret.add(new ImmutableBytesWritable(fullKey));<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> return ret;<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> /**<a name="line.527"></a>
-<span class="sourceLineNo">528</span> * Write out a {@link SequenceFile} that can be read by<a name="line.528"></a>
-<span class="sourceLineNo">529</span> * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.529"></a>
-<span class="sourceLineNo">530</span> */<a name="line.530"></a>
-<span class="sourceLineNo">531</span> @SuppressWarnings("deprecation")<a name="line.531"></a>
-<span class="sourceLineNo">532</span> private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.532"></a>
-<span class="sourceLineNo">533</span> List<ImmutableBytesWritable> startKeys, boolean writeMultipleTables) throws IOException {<a name="line.533"></a>
-<span class="sourceLineNo">534</span> LOG.info("Writing partition information to " + partitionsPath);<a name="line.534"></a>
-<span class="sourceLineNo">535</span> if (startKeys.isEmpty()) {<a name="line.535"></a>
-<span class="sourceLineNo">536</span> throw new IllegalArgumentException("No regions passed");<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> // We're generating a list of split points, and we don't ever<a name="line.539"></a>
-<span class="sourceLineNo">540</span> // have keys < the first region (which has an empty start key)<a name="line.540"></a>
-<span class="sourceLineNo">541</span> // so we need to remove it. Otherwise we would end up with an<a name="line.541"></a>
-<span class="sourceLineNo">542</span> // empty reducer with index 0<a name="line.542"></a>
-<span class="sourceLineNo">543</span> TreeSet<ImmutableBytesWritable> sorted = new TreeSet<>(startKeys);<a name="line.543"></a>
-<span class="sourceLineNo">544</span> ImmutableBytesWritable first = sorted.first();<a name="line.544"></a>
-<span class="sourceLineNo">545</span> if (writeMultipleTables) {<a name="line.545"></a>
-<span class="sourceLineNo">546</span> first = new ImmutableBytesWritable(MultiTableHFileOutputFormat.getSuffix(sorted.first<a name="line.546"></a>
-<span class="sourceLineNo">547</span> ().get()));<a name="line.547"></a>
-<span class="sourceLineNo">548</span> }<a name="line.548"></a>
-<span class="sourceLineNo">549</span> if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.549"></a>
-<span class="sourceLineNo">550</span> throw new IllegalArgumentException(<a name="line.550"></a>
-<span class="sourceLineNo">551</span> "First region of table should have empty start key. Instead has: "<a name="line.551"></a>
-<span class="sourceLineNo">552</span> + Bytes.toStringBinary(first.get()));<a name="line.552"></a>
-<span class="sourceLineNo">553</span> }<a name="line.553"></a>
-<span class="sourceLineNo">554</span> sorted.remove(sorted.first());<a name="line.554"></a>
-<span class="sourceLineNo">555</span><a name="line.555"></a>
-<span class="sourceLineNo">556</span> // Write the actual file<a name="line.556"></a>
-<span class="sourceLineNo">557</span> FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.557"></a>
-<span class="sourceLineNo">558</span> SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.558"></a>
-<span class="sourceLineNo">559</span> fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.559"></a>
-<span class="sourceLineNo">560</span> NullWritable.class);<a name="line.560"></a>
-<span class="sourceLineNo">561</span><a name="line.561"></a>
-<span class="sourceLineNo">562</span> try {<a name="line.562"></a>
-<span class="sourceLineNo">563</span> for (ImmutableBytesWritable startKey : sorted) {<a name="line.563"></a>
-<span class="sourceLineNo">564</span> writer.append(startKey, NullWritable.get());<a name="line.564"></a>
-<span class="sourceLineNo">565</span> }<a name="line.565"></a>
-<span class="sourceLineNo">566</span> } finally {<a name="line.566"></a>
-<span class="sourceLineNo">567</span> writer.close();<a name="line.567"></a>
-<span class="sourceLineNo">568</span> }<a name="line.568"></a>
-<span class="sourceLineNo">569</span> }<a name="line.569"></a>
-<span class="sourceLineNo">570</span><a name="line.570"></a>
-<span class="sourceLineNo">571</span> /**<a name="line.571"></a>
-<span class="sourceLineNo">572</span> * Configure a MapReduce Job to perform an incremental load into the given<a name="line.572"></a>
-<span class="sourceLineNo">573</span> * table. This<a name="line.573"></a>
-<span class="sourceLineNo">574</span> * <ul><a name="line.574"></a>
-<span class="sourceLineNo">575</span> * <li>Inspects the table to configure a total order partitioner</li><a name="line.575"></a>
-<span class="sourceLineNo">576</span> * <li>Uploads the partitions file to the cluster and adds it to the DistributedCache</li><a name="line.576"></a>
-<span class="sourceLineNo">577</span> * <li>Sets the number of reduce tasks to match the current number of regions</li><a name="line.577"></a>
-<span class="sourceLineNo">578</span> * <li>Sets the output key/value class to match HFileOutputFormat2's requirements</li><a name="line.578"></a>
-<span class="sourceLineNo">579</span> * <li>Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.579"></a>
-<span class="sourceLineNo">580</span> * PutSortReducer)</li><a name="line.580"></a>
-<span class="sourceLineNo">581</span> * </ul><a name="line.581"></a>
-<span class="sourceLineNo">582</span> * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.582"></a>
-<span class="sourceLineNo">583</span> * running this function.<a name="line.583"></a>
-<span class="sourceLineNo">584</span> */<a name="line.584"></a>
-<span class="sourceLineNo">585</span> public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.585"></a>
-<span class="sourceLineNo">586</span> throws IOException {<a name="line.586"></a>
-<span class="sourceLineNo">587</span> configureIncrementalLoad(job, table.getDescriptor(), regionLocator);<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> /**<a name="line.590"></a>
-<span class="sourceLineNo">591</span> * Configure a MapReduce Job to perform an incremental load into the given<a name="line.591"></a>
-<span class="sourceLineNo">592</span> * table. This<a name="line.592"></a>
-<span class="sourceLineNo">593</span> * <ul><a name="line.593"></a>
-<span class="sourceLineNo">594</span> * <li>Inspects the table to configure a total order partitioner</li><a name="line.594"></a>
-<span class="sourceLineNo">595</span> * <li>Uploads the partitions file to the cluster and adds it to the DistributedCache</li><a name="line.595"></a>
-<span class="sourceLineNo">596</span> * <li>Sets the number of reduce tasks to match the current number of regions</li><a name="line.596"></a>
-<span class="sourceLineNo">597</span> * <li>Sets the output key/value class to match HFileOutputFormat2's requirements</li><a name="line.597"></a>
-<span class="sourceLineNo">598</span> * <li>Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.598"></a>
-<span class="sourceLineNo">599</span> * PutSortReducer)</li><a name="line.599"></a>
-<span class="sourceLineNo">600</span> * </ul><a name="line.600"></a>
-<span class="sourceLineNo">601</span> * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.601"></a>
-<span class="sourceLineNo">602</span> * running this function.<a name="line.602"></a>
-<span class="sourceLineNo">603</span> */<a name="line.603"></a>
-<span class="sourceLineNo">604</span> public static void configureIncrementalLoad(Job job, TableDescriptor tableDescriptor,<a name="line.604"></a>
-<span class="sourceLineNo">605</span> RegionLocator regionLocator) throws IOException {<a name="line.605"></a>
-<span class="sourceLineNo">606</span> ArrayList<TableInfo> singleTableInfo = new ArrayList<>();<a name="line.606"></a>
-<span class="sourceLineNo">607</span> singleTableInfo.add(new TableInfo(tableDescriptor, regionLocator));<a name="line.607"></a>
-<span class="sourceLineNo">608</span> configureIncrementalLoad(job, singleTableInfo, HFileOutputFormat2.class);<a name="line.608"></a>
-<span class="sourceLineNo">609</span> }<a name="line.609"></a>
-<span class="sourceLineNo">610</span><a name="line.610"></a>
-<span class="sourceLineNo">611</span> static void configureIncrementalLoad(Job job, List<TableInfo> multiTableInfo,<a name="line.611"></a>
-<span class="sourceLineNo">612</span> Class<? extends OutputFormat<?, ?>> cls) throws IOException {<a name="line.612"></a>
-<span class="sourceLineNo">613</span> Configuration conf = job.getConfiguration();<a name="line.613"></a>
-<span class="sourceLineNo">614</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.614"></a>
-<span class="sourceLineNo">615</span> job.setOutputValueClass(MapReduceExtendedCell.class);<a name="line.615"></a>
-<span class="sourceLineNo">616</span> job.setOutputFormatClass(cls);<a name="line.616"></a>
-<span class="sourceLineNo">617</span><a name="line.617"></a>
-<span class="sourceLineNo">618</span> if (multiTableInfo.stream().distinct().count() != multiTableInfo.size()) {<a name="line.618"></a>
-<span class="sourceLineNo">619</span> throw new IllegalArgumentException("Duplicate entries found in TableInfo argument");<a name="line.619"></a>
-<span class="sourceLineNo">620</span> }<a name="line.620"></a>
-<span class="sourceLineNo">621</span> boolean writeMultipleTables = false;<a name="line.621"></a>
-<span class="sourceLineNo">622</span> if (MultiTableHFileOutputFormat.class.equals(cls)) {<a name="line.622"></a>
-<span class="sourceLineNo">623</span> writeMultipleTables = true;<a name="line.623"></a>
-<span class="sourceLineNo">624</span> conf.setBoolean(MULTI_TABLE_HFILEOUTPUTFORMAT_CONF_KEY, true);<a name="line.624"></a>
-<span class="sourceLineNo">625</span> }<a name="line.625"></a>
-<span class="sourceLineNo">626</span> // Based on the configured map output class, set the correct reducer to properly<a name="line.626"></a>
-<span class="sourceLineNo">627</span> // sort the incoming values.<a name="line.627"></a>
-<span class="sourceLineNo">628</span> // TODO it would be nice to pick one or the other of these formats.<a name="line.628"></a>
-<span class="sourceLineNo">629</span> if (KeyValue.class.equals(job.getMapOutputValueClass())<a name="line.629"></a>
-<span class="sourceLineNo">630</span> || MapReduceExtendedCell.class.equals(job.getMapOutputValueClass())) {<a name="line.630"></a>
-<span class="sourceLineNo">631</span> job.setReducerClass(CellSortReducer.class);<a name="line.631"></a>
-<span class="sourceLineNo">632</span> } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.632"></a>
-<span class="sourceLineNo">633</span> job.setReducerClass(PutSortReducer.class);<a name="line.633"></a>
-<span class="sourceLineNo">634</span> } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.634"></a>
-<span class="sourceLineNo">635</span> job.setReducerClass(TextSortReducer.class);<a name="line.635"></a>
-<span class="sourceLineNo">636</span> } else {<a name="line.636"></a>
-<span class="sourceLineNo">637</span> LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.637"></a>
-<span class="sourceLineNo">638</span> }<a name="line.638"></a>
-<span class="sourceLineNo">639</span><a name="line.639"></a>
-<span class="sourceLineNo">640</span> conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.640"></a>
-<span class="sourceLineNo">641</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.641"></a>
-<span class="sourceLineNo">642</span> CellSerialization.class.getName());<a name="line.642"></a>
-<span class="sourceLineNo">643</span><a name="line.643"></a>
-<span class="sourceLineNo">644</span> if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.644"></a>
-<span class="sourceLineNo">645</span> LOG.info("bulkload locality sensitive enabled");<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> /* Now get the region start keys for every table required */<a name="line.648"></a>
-<span class="sourceLineNo">649</span> List<String> allTableNames = new ArrayList<>(multiTableInfo.size());<a name="line.649"></a>
-<span class="sourceLineNo">650</span> List<RegionLocator> regionLocators = new ArrayList<>( multiTableInfo.size());<a name="line.650"></a>
-<span class="sourceLineNo">651</span> List<TableDescriptor> tableDescriptors = new ArrayList<>( multiTableInfo.size());<a name="line.651"></a>
-<span class="sourceLineNo">652</span><a name="line.652"></a>
-<span class="sourceLineNo">653</span> for( TableInfo tableInfo : multiTableInfo )<a name="line.653"></a>
-<span class="sourceLineNo">654</span> {<a name="line.654"></a>
-<span class="sourceLineNo">655</span> regionLocators.add(tableInfo.getRegionLocator());<a name="line.655"></a>
-<span class="sourceLineNo">656</span> String tn = writeMultipleTables?<a name="line.656"></a>
-<span class="sourceLineNo">657</span> tableInfo.getRegionLocator().getName().getNameWithNamespaceInclAsString():<a name="line.657"></a>
-<span class="sourceLineNo">658</span> tableInfo.getRegionLocator().getName().getNameAsString();<a name="line.658"></a>
-<span class="sourceLineNo">659</span> allTableNames.add(tn);<a name="line.659"></a>
-<span class="sourceLineNo">660</span> tableDescriptors.add(tableInfo.getTableDescriptor());<a name="line.660"></a>
-<span class="sourceLineNo">661</span> }<a name="line.661"></a>
-<span class="sourceLineNo">662</span> // Record tablenames for creating writer by favored nodes, and decoding compression, block size and other attributes of columnfamily per table<a name="line.662"></a>
-<span class="sourceLineNo">663</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, StringUtils.join(allTableNames, Bytes<a name="line.663"></a>
-<span class="sourceLineNo">664</span> .toString(tableSeparator)));<a name="line.664"></a>
-<span class="sourceLineNo">665</span> List<ImmutableBytesWritable> startKeys = getRegionStartKeys(regionLocators, writeMultipleTables);<a name="line.665"></a>
-<span class="sourceLineNo">666</span> // Use table's region boundaries for TOP split points.<a name="line.666"></a>
-<span class="sourceLineNo">667</span> LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.667"></a>
-<span class="sourceLineNo">668</span> "to match current region count for all tables");<a name="line.668"></a>
-<span class="sourceLineNo">669</span> job.setNumReduceTasks(startKeys.size());<a name="line.669"></a>
-<span class="sourceLineNo">670</span><a name="line.670"></a>
-<span class="sourceLineNo">671</span> configurePartitioner(job, startKeys, writeMultipleTables);<a name="line.671"></a>
-<span class="sourceLineNo">672</span> // Set compression algorithms based on column families<a name="line.672"></a>
-<span class="sourceLineNo">673</span><a name="line.673"></a>
-<span class="sourceLineNo">674</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(compressionDetails,<a name="line.674"></a>
+<span class="sourceLineNo">417</span> HFileContextBuilder contextBuilder = new HFileContextBuilder()<a name="line.417"></a>
+<span class="sourceLineNo">418</span> .withCompression(compression)<a name="line.418"></a>
+<span class="sourceLineNo">419</span> .withChecksumType(HStore.getChecksumType(conf))<a name="line.419"></a>
+<span class="sourceLineNo">420</span> .withBytesPerCheckSum(HStore.getBytesPerChecksum(conf))<a name="line.420"></a>
+<span class="sourceLineNo">421</span> .withBlockSize(blockSize);<a name="line.421"></a>
+<span class="sourceLineNo">422</span><a name="line.422"></a>
+<span class="sourceLineNo">423</span> if (HFile.getFormatVersion(conf) >= HFile.MIN_FORMAT_VERSION_WITH_TAGS) {<a name="line.423"></a>
+<span class="sourceLineNo">424</span> contextBuilder.withIncludesTags(true);<a name="line.424"></a>
+<span class="sourceLineNo">425</span> }<a name="line.425"></a>
+<span class="sourceLineNo">426</span><a name="line.426"></a>
+<span class="sourceLineNo">427</span> contextBuilder.withDataBlockEncoding(encoding);<a name="line.427"></a>
+<span class="sourceLineNo">428</span> HFileContext hFileContext = contextBuilder.build();<a name="line.428"></a>
+<span class="sourceLineNo">429</span> if (null == favoredNodes) {<a name="line.429"></a>
+<span class="sourceLineNo">430</span> wl.writer =<a name="line.430"></a>
+<span class="sourceLineNo">431</span> new StoreFileWriter.Builder(conf, CacheConfig.DISABLED, fs)<a name="line.431"></a>
+<span class="sourceLineNo">432</span> .withOutputDir(familydir).withBloomType(bloomType)<a name="line.432"></a>
+<span class="sourceLineNo">433</span> .withComparator(CellComparator.getInstance()).withFileContext(hFileContext).build();<a name="line.433"></a>
+<span class="sourceLineNo">434</span> } else {<a name="line.434"></a>
+<span class="sourceLineNo">435</span> wl.writer =<a name="line.435"></a>
+<span class="sourceLineNo">436</span> new StoreFileWriter.Builder(conf, CacheConfig.DISABLED, new HFileSystem(fs))<a name="line.436"></a>
+<span class="sourceLineNo">437</span> .withOutputDir(familydir).withBloomType(bloomType)<a name="line.437"></a>
+<span class="sourceLineNo">438</span> .withComparator(CellComparator.getInstance()).withFileContext(hFileContext)<a name="line.438"></a>
+<span class="sourceLineNo">439</span> .withFavoredNodes(favoredNodes).build();<a name="line.439"></a>
+<span class="sourceLineNo">440</span> }<a name="line.440"></a>
+<span class="sourceLineNo">441</span><a name="line.441"></a>
+<span class="sourceLineNo">442</span> this.writers.put(tableAndFamily, wl);<a name="line.442"></a>
+<span class="sourceLineNo">443</span> return wl;<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> private void close(final StoreFileWriter w) throws IOException {<a name="line.446"></a>
+<span class="sourceLineNo">447</span> if (w != null) {<a name="line.447"></a>
+<span class="sourceLineNo">448</span> w.appendFileInfo(BULKLOAD_TIME_KEY,<a name="line.448"></a>
+<span class="sourceLineNo">449</span> Bytes.toBytes(System.currentTimeMillis()));<a name="line.449"></a>
+<span class="sourceLineNo">450</span> w.appendFileInfo(BULKLOAD_TASK_KEY,<a name="line.450"></a>
+<span class="sourceLineNo">451</span> Bytes.toBytes(context.getTaskAttemptID().toString()));<a name="line.451"></a>
+<span class="sourceLineNo">452</span> w.appendFileInfo(MAJOR_COMPACTION_KEY,<a name="line.452"></a>
+<span class="sourceLineNo">453</span> Bytes.toBytes(true));<a name="line.453"></a>
+<span class="sourceLineNo">454</span> w.appendFileInfo(EXCLUDE_FROM_MINOR_COMPACTION_KEY,<a name="line.454"></a>
+<span class="sourceLineNo">455</span> Bytes.toBytes(compactionExclude));<a name="line.455"></a>
+<span class="sourceLineNo">456</span> w.appendTrackedTimestampsToMetadata();<a name="line.456"></a>
+<span class="sourceLineNo">457</span> w.close();<a name="line.457"></a>
+<span class="sourceLineNo">458</span> }<a name="line.458"></a>
+<span class="sourceLineNo">459</span> }<a name="line.459"></a>
+<span class="sourceLineNo">460</span><a name="line.460"></a>
+<span class="sourceLineNo">461</span> @Override<a name="line.461"></a>
+<span class="sourceLineNo">462</span> public void close(TaskAttemptContext c)<a name="line.462"></a>
+<span class="sourceLineNo">463</span> throws IOException, InterruptedException {<a name="line.463"></a>
+<span class="sourceLineNo">464</span> for (WriterLength wl: this.writers.values()) {<a name="line.464"></a>
+<span class="sourceLineNo">465</span> close(wl.writer);<a name="line.465"></a>
+<span class="sourceLineNo">466</span> }<a name="line.466"></a>
+<span class="sourceLineNo">467</span> }<a name="line.467"></a>
+<span class="sourceLineNo">468</span> };<a name="line.468"></a>
+<span class="sourceLineNo">469</span> }<a name="line.469"></a>
+<span class="sourceLineNo">470</span><a name="line.470"></a>
+<span class="sourceLineNo">471</span> /**<a name="line.471"></a>
+<span class="sourceLineNo">472</span> * Configure block storage policy for CF after the directory is created.<a name="line.472"></a>
+<span class="sourceLineNo">473</span> */<a name="line.473"></a>
+<span class="sourceLineNo">474</span> static void configureStoragePolicy(final Configuration conf, final FileSystem fs,<a name="line.474"></a>
+<span class="sourceLineNo">475</span> byte[] tableAndFamily, Path cfPath) {<a name="line.475"></a>
+<span class="sourceLineNo">476</span> if (null == conf || null == fs || null == tableAndFamily || null == cfPath) {<a name="line.476"></a>
+<span class="sourceLineNo">477</span> return;<a name="line.477"></a>
+<span class="sourceLineNo">478</span> }<a name="line.478"></a>
+<span class="sourceLineNo">479</span><a name="line.479"></a>
+<span class="sourceLineNo">480</span> String policy =<a name="line.480"></a>
+<span class="sourceLineNo">481</span> conf.get(STORAGE_POLICY_PROPERTY_CF_PREFIX + Bytes.toString(tableAndFamily),<a name="line.481"></a>
+<span class="sourceLineNo">482</span> conf.get(STORAGE_POLICY_PROPERTY));<a name="line.482"></a>
+<span class="sourceLineNo">483</span> FSUtils.setStoragePolicy(fs, cfPath, policy);<a name="line.483"></a>
+<span class="sourceLineNo">484</span> }<a name="line.484"></a>
+<span class="sourceLineNo">485</span><a name="line.485"></a>
+<span class="sourceLineNo">486</span> /*<a name="line.486"></a>
+<span class="sourceLineNo">487</span> * Data structure to hold a Writer and amount of data written on it.<a name="line.487"></a>
+<span class="sourceLineNo">488</span> */<a name="line.488"></a>
+<span class="sourceLineNo">489</span> static class WriterLength {<a name="line.489"></a>
+<span class="sourceLineNo">490</span> long written = 0;<a name="line.490"></a>
+<span class="sourceLineNo">491</span> StoreFileWriter writer = null;<a name="line.491"></a>
+<span class="sourceLineNo">492</span> }<a name="line.492"></a>
+<span class="sourceLineNo">493</span><a name="line.493"></a>
+<span class="sourceLineNo">494</span> /**<a name="line.494"></a>
+<span class="sourceLineNo">495</span> * Return the start keys of all of the regions in this table,<a name="line.495"></a>
+<span class="sourceLineNo">496</span> * as a list of ImmutableBytesWritable.<a name="line.496"></a>
+<span class="sourceLineNo">497</span> */<a name="line.497"></a>
+<span class="sourceLineNo">498</span> private static List<ImmutableBytesWritable> getRegionStartKeys(List<RegionLocator> regionLocators,<a name="line.498"></a>
+<span class="sourceLineNo">499</span> boolean writeMultipleTables)<a name="line.499"></a>
+<span class="sourceLineNo">500</span> throws IOException {<a name="line.500"></a>
+<span class="sourceLineNo">501</span><a name="line.501"></a>
+<span class="sourceLineNo">502</span> ArrayList<ImmutableBytesWritable> ret = new ArrayList<>();<a name="line.502"></a>
+<span class="sourceLineNo">503</span> for(RegionLocator regionLocator : regionLocators)<a name="line.503"></a>
+<span class="sourceLineNo">504</span> {<a name="line.504"></a>
+<span class="sourceLineNo">505</span> TableName tableName = regionLocator.getName();<a name="line.505"></a>
+<span class="sourceLineNo">506</span> LOG.info("Looking up current regions for table " + tableName);<a name="line.506"></a>
+<span class="sourceLineNo">507</span> byte[][] byteKeys = regionLocator.getStartKeys();<a name="line.507"></a>
+<span class="sourceLineNo">508</span> for (byte[] byteKey : byteKeys) {<a name="line.508"></a>
+<span class="sourceLineNo">509</span> byte[] fullKey = byteKey; //HFileOutputFormat2 use case<a name="line.509"></a>
+<span class="sourceLineNo">510</span> if (writeMultipleTables)<a name="line.510"></a>
+<span class="sourceLineNo">511</span> {<a name="line.511"></a>
+<span class="sourceLineNo">512</span> //MultiTableHFileOutputFormat use case<a name="line.512"></a>
+<span class="sourceLineNo">513</span> fullKey = combineTableNameSuffix(tableName.getName(), byteKey);<a name="line.513"></a>
+<span class="sourceLineNo">514</span> }<a name="line.514"></a>
+<span class="sourceLineNo">515</span> if (LOG.isDebugEnabled()) {<a name="line.515"></a>
+<span class="sourceLineNo">516</span> LOG.debug("SplitPoint startkey for table [" + tableName + "]: [" + Bytes.toStringBinary<a name="line.516"></a>
+<span class="sourceLineNo">517</span> (fullKey) + "]");<a name="line.517"></a>
+<span class="sourceLineNo">518</span> }<a name="line.518"></a>
+<span class="sourceLineNo">519</span> ret.add(new ImmutableBytesWritable(fullKey));<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> return ret;<a name="line.522"></a>
+<span class="sourceLineNo">523</span> }<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> * Write out a {@link SequenceFile} that can be read by<a name="line.526"></a>
+<span class="sourceLineNo">527</span> * {@link TotalOrderPartitioner} that contains the split points in startKeys.<a name="line.527"></a>
+<span class="sourceLineNo">528</span> */<a name="line.528"></a>
+<span class="sourceLineNo">529</span> @SuppressWarnings("deprecation")<a name="line.529"></a>
+<span class="sourceLineNo">530</span> private static void writePartitions(Configuration conf, Path partitionsPath,<a name="line.530"></a>
+<span class="sourceLineNo">531</span> List<ImmutableBytesWritable> startKeys, boolean writeMultipleTables) throws IOException {<a name="line.531"></a>
+<span class="sourceLineNo">532</span> LOG.info("Writing partition information to " + partitionsPath);<a name="line.532"></a>
+<span class="sourceLineNo">533</span> if (startKeys.isEmpty()) {<a name="line.533"></a>
+<span class="sourceLineNo">534</span> throw new IllegalArgumentException("No regions passed");<a name="line.534"></a>
+<span class="sourceLineNo">535</span> }<a name="line.535"></a>
+<span class="sourceLineNo">536</span><a name="line.536"></a>
+<span class="sourceLineNo">537</span> // We're generating a list of split points, and we don't ever<a name="line.537"></a>
+<span class="sourceLineNo">538</span> // have keys < the first region (which has an empty start key)<a name="line.538"></a>
+<span class="sourceLineNo">539</span> // so we need to remove it. Otherwise we would end up with an<a name="line.539"></a>
+<span class="sourceLineNo">540</span> // empty reducer with index 0<a name="line.540"></a>
+<span class="sourceLineNo">541</span> TreeSet<ImmutableBytesWritable> sorted = new TreeSet<>(startKeys);<a name="line.541"></a>
+<span class="sourceLineNo">542</span> ImmutableBytesWritable first = sorted.first();<a name="line.542"></a>
+<span class="sourceLineNo">543</span> if (writeMultipleTables) {<a name="line.543"></a>
+<span class="sourceLineNo">544</span> first = new ImmutableBytesWritable(MultiTableHFileOutputFormat.getSuffix(sorted.first<a name="line.544"></a>
+<span class="sourceLineNo">545</span> ().get()));<a name="line.545"></a>
+<span class="sourceLineNo">546</span> }<a name="line.546"></a>
+<span class="sourceLineNo">547</span> if (!first.equals(HConstants.EMPTY_BYTE_ARRAY)) {<a name="line.547"></a>
+<span class="sourceLineNo">548</span> throw new IllegalArgumentException(<a name="line.548"></a>
+<span class="sourceLineNo">549</span> "First region of table should have empty start key. Instead has: "<a name="line.549"></a>
+<span class="sourceLineNo">550</span> + Bytes.toStringBinary(first.get()));<a name="line.550"></a>
+<span class="sourceLineNo">551</span> }<a name="line.551"></a>
+<span class="sourceLineNo">552</span> sorted.remove(sorted.first());<a name="line.552"></a>
+<span class="sourceLineNo">553</span><a name="line.553"></a>
+<span class="sourceLineNo">554</span> // Write the actual file<a name="line.554"></a>
+<span class="sourceLineNo">555</span> FileSystem fs = partitionsPath.getFileSystem(conf);<a name="line.555"></a>
+<span class="sourceLineNo">556</span> SequenceFile.Writer writer = SequenceFile.createWriter(<a name="line.556"></a>
+<span class="sourceLineNo">557</span> fs, conf, partitionsPath, ImmutableBytesWritable.class,<a name="line.557"></a>
+<span class="sourceLineNo">558</span> NullWritable.class);<a name="line.558"></a>
+<span class="sourceLineNo">559</span><a name="line.559"></a>
+<span class="sourceLineNo">560</span> try {<a name="line.560"></a>
+<span class="sourceLineNo">561</span> for (ImmutableBytesWritable startKey : sorted) {<a name="line.561"></a>
+<span class="sourceLineNo">562</span> writer.append(startKey, NullWritable.get());<a name="line.562"></a>
+<span class="sourceLineNo">563</span> }<a name="line.563"></a>
+<span class="sourceLineNo">564</span> } finally {<a name="line.564"></a>
+<span class="sourceLineNo">565</span> writer.close();<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><a name="line.568"></a>
+<span class="sourceLineNo">569</span> /**<a name="line.569"></a>
+<span class="sourceLineNo">570</span> * Configure a MapReduce Job to perform an incremental load into the given<a name="line.570"></a>
+<span class="sourceLineNo">571</span> * table. This<a name="line.571"></a>
+<span class="sourceLineNo">572</span> * <ul><a name="line.572"></a>
+<span class="sourceLineNo">573</span> * <li>Inspects the table to configure a total order partitioner</li><a name="line.573"></a>
+<span class="sourceLineNo">574</span> * <li>Uploads the partitions file to the cluster and adds it to the DistributedCache</li><a name="line.574"></a>
+<span class="sourceLineNo">575</span> * <li>Sets the number of reduce tasks to match the current number of regions</li><a name="line.575"></a>
+<span class="sourceLineNo">576</span> * <li>Sets the output key/value class to match HFileOutputFormat2's requirements</li><a name="line.576"></a>
+<span class="sourceLineNo">577</span> * <li>Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.577"></a>
+<span class="sourceLineNo">578</span> * PutSortReducer)</li><a name="line.578"></a>
+<span class="sourceLineNo">579</span> * </ul><a name="line.579"></a>
+<span class="sourceLineNo">580</span> * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.580"></a>
+<span class="sourceLineNo">581</span> * running this function.<a name="line.581"></a>
+<span class="sourceLineNo">582</span> */<a name="line.582"></a>
+<span class="sourceLineNo">583</span> public static void configureIncrementalLoad(Job job, Table table, RegionLocator regionLocator)<a name="line.583"></a>
+<span class="sourceLineNo">584</span> throws IOException {<a name="line.584"></a>
+<span class="sourceLineNo">585</span> configureIncrementalLoad(job, table.getDescriptor(), regionLocator);<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> /**<a name="line.588"></a>
+<span class="sourceLineNo">589</span> * Configure a MapReduce Job to perform an incremental load into the given<a name="line.589"></a>
+<span class="sourceLineNo">590</span> * table. This<a name="line.590"></a>
+<span class="sourceLineNo">591</span> * <ul><a name="line.591"></a>
+<span class="sourceLineNo">592</span> * <li>Inspects the table to configure a total order partitioner</li><a name="line.592"></a>
+<span class="sourceLineNo">593</span> * <li>Uploads the partitions file to the cluster and adds it to the DistributedCache</li><a name="line.593"></a>
+<span class="sourceLineNo">594</span> * <li>Sets the number of reduce tasks to match the current number of regions</li><a name="line.594"></a>
+<span class="sourceLineNo">595</span> * <li>Sets the output key/value class to match HFileOutputFormat2's requirements</li><a name="line.595"></a>
+<span class="sourceLineNo">596</span> * <li>Sets the reducer up to perform the appropriate sorting (either KeyValueSortReducer or<a name="line.596"></a>
+<span class="sourceLineNo">597</span> * PutSortReducer)</li><a name="line.597"></a>
+<span class="sourceLineNo">598</span> * </ul><a name="line.598"></a>
+<span class="sourceLineNo">599</span> * The user should be sure to set the map output value class to either KeyValue or Put before<a name="line.599"></a>
+<span class="sourceLineNo">600</span> * running this function.<a name="line.600"></a>
+<span class="sourceLineNo">601</span> */<a name="line.601"></a>
+<span class="sourceLineNo">602</span> public static void configureIncrementalLoad(Job job, TableDescriptor tableDescriptor,<a name="line.602"></a>
+<span class="sourceLineNo">603</span> RegionLocator regionLocator) throws IOException {<a name="line.603"></a>
+<span class="sourceLineNo">604</span> ArrayList<TableInfo> singleTableInfo = new ArrayList<>();<a name="line.604"></a>
+<span class="sourceLineNo">605</span> singleTableInfo.add(new TableInfo(tableDescriptor, regionLocator));<a name="line.605"></a>
+<span class="sourceLineNo">606</span> configureIncrementalLoad(job, singleTableInfo, HFileOutputFormat2.class);<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> static void configureIncrementalLoad(Job job, List<TableInfo> multiTableInfo,<a name="line.609"></a>
+<span class="sourceLineNo">610</span> Class<? extends OutputFormat<?, ?>> cls) throws IOException {<a name="line.610"></a>
+<span class="sourceLineNo">611</span> Configuration conf = job.getConfiguration();<a name="line.611"></a>
+<span class="sourceLineNo">612</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.612"></a>
+<span class="sourceLineNo">613</span> job.setOutputValueClass(MapReduceExtendedCell.class);<a name="line.613"></a>
+<span class="sourceLineNo">614</span> job.setOutputFormatClass(cls);<a name="line.614"></a>
+<span class="sourceLineNo">615</span><a name="line.615"></a>
+<span class="sourceLineNo">616</span> if (multiTableInfo.stream().distinct().count() != multiTableInfo.size()) {<a name="line.616"></a>
+<span class="sourceLineNo">617</span> throw new IllegalArgumentException("Duplicate entries found in TableInfo argument");<a name="line.617"></a>
+<span class="sourceLineNo">618</span> }<a name="line.618"></a>
+<span class="sourceLineNo">619</span> boolean writeMultipleTables = false;<a name="line.619"></a>
+<span class="sourceLineNo">620</span> if (MultiTableHFileOutputFormat.class.equals(cls)) {<a name="line.620"></a>
+<span class="sourceLineNo">621</span> writeMultipleTables = true;<a name="line.621"></a>
+<span class="sourceLineNo">622</span> conf.setBoolean(MULTI_TABLE_HFILEOUTPUTFORMAT_CONF_KEY, true);<a name="line.622"></a>
+<span class="sourceLineNo">623</span> }<a name="line.623"></a>
+<span class="sourceLineNo">624</span> // Based on the configured map output class, set the correct reducer to properly<a name="line.624"></a>
+<span class="sourceLineNo">625</span> // sort the incoming values.<a name="line.625"></a>
+<span class="sourceLineNo">626</span> // TODO it would be nice to pick one or the other of these formats.<a name="line.626"></a>
+<span class="sourceLineNo">627</span> if (KeyValue.class.equals(job.getMapOutputValueClass())<a name="line.627"></a>
+<span class="sourceLineNo">628</span> || MapReduceExtendedCell.class.equals(job.getMapOutputValueClass())) {<a name="line.628"></a>
+<span class="sourceLineNo">629</span> job.setReducerClass(CellSortReducer.class);<a name="line.629"></a>
+<span class="sourceLineNo">630</span> } else if (Put.class.equals(job.getMapOutputValueClass())) {<a name="line.630"></a>
+<span class="sourceLineNo">631</span> job.setReducerClass(PutSortReducer.class);<a name="line.631"></a>
+<span class="sourceLineNo">632</span> } else if (Text.class.equals(job.getMapOutputValueClass())) {<a name="line.632"></a>
+<span class="sourceLineNo">633</span> job.setReducerClass(TextSortReducer.class);<a name="line.633"></a>
+<span class="sourceLineNo">634</span> } else {<a name="line.634"></a>
+<span class="sourceLineNo">635</span> LOG.warn("Unknown map output value type:" + job.getMapOutputValueClass());<a name="line.635"></a>
+<span class="sourceLineNo">636</span> }<a name="line.636"></a>
+<span class="sourceLineNo">637</span><a name="line.637"></a>
+<span class="sourceLineNo">638</span> conf.setStrings("io.serializations", conf.get("io.serializations"),<a name="line.638"></a>
+<span class="sourceLineNo">639</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.639"></a>
+<span class="sourceLineNo">640</span> CellSerialization.class.getName());<a name="line.640"></a>
+<span class="sourceLineNo">641</span><a name="line.641"></a>
+<span class="sourceLineNo">642</span> if (conf.getBoolean(LOCALITY_SENSITIVE_CONF_KEY, DEFAULT_LOCALITY_SENSITIVE)) {<a name="line.642"></a>
+<span class="sourceLineNo">643</span> LOG.info("bulkload locality sensitive enabled");<a name="line.643"></a>
+<span class="sourceLineNo">644</span> }<a name="line.644"></a>
+<span class="sourceLineNo">645</span><a name="line.645"></a>
+<span class="sourceLineNo">646</span> /* Now get the region start keys for every table required */<a name="line.646"></a>
+<span class="sourceLineNo">647</span> List<String> allTableNames = new ArrayList<>(multiTableInfo.size());<a name="line.647"></a>
+<span class="sourceLineNo">648</span> List<RegionLocator> regionLocators = new ArrayList<>( multiTableInfo.size());<a name="line.648"></a>
+<span class="sourceLineNo">649</span> List<TableDescriptor> tableDescriptors = new ArrayList<>( multiTableInfo.size());<a name="line.649"></a>
+<span class="sourceLineNo">650</span><a name="line.650"></a>
+<span class="sourceLineNo">651</span> for( TableInfo tableInfo : multiTableInfo )<a name="line.651"></a>
+<span class="sourceLineNo">652</span> {<a name="line.652"></a>
+<span class="sourceLineNo">653</span> regionLocators.add(tableInfo.getRegionLocator());<a name="line.653"></a>
+<span class="sourceLineNo">654</span> String tn = writeMultipleTables?<a name="line.654"></a>
+<span class="sourceLineNo">655</span> tableInfo.getRegionLocator().getName().getNameWithNamespaceInclAsString():<a name="line.655"></a>
+<span class="sourceLineNo">656</span> tableInfo.getRegionLocator().getName().getNameAsString();<a name="line.656"></a>
+<span class="sourceLineNo">657</span> allTableNames.add(tn);<a name="line.657"></a>
+<span class="sourceLineNo">658</span> tableDescriptors.add(tableInfo.getTableDescriptor());<a name="line.658"></a>
+<span class="sourceLineNo">659</span> }<a name="line.659"></a>
+<span class="sourceLineNo">660</span> // Record tablenames for creating writer by favored nodes, and decoding compression, block size and other attributes of columnfamily per table<a name="line.660"></a>
+<span class="sourceLineNo">661</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, StringUtils.join(allTableNames, Bytes<a name="line.661"></a>
+<span class="sourceLineNo">662</span> .toString(tableSeparator)));<a name="line.662"></a>
+<span class="sourceLineNo">663</span> List<ImmutableBytesWritable> startKeys = getRegionStartKeys(regionLocators, writeMultipleTables);<a name="line.663"></a>
+<span class="sourceLineNo">664</span> // Use table's region boundaries for TOP split points.<a name="line.664"></a>
+<span class="sourceLineNo">665</span> LOG.info("Configuring " + startKeys.size() + " reduce partitions " +<a name="line.665"></a>
+<span class="sourceLineNo">666</span> "to match current region count for all tables");<a name="line.666"></a>
+<span class="sourceLineNo">667</span> job.setNumReduceTasks(startKeys.size());<a name="line.667"></a>
+<span class="sourceLineNo">668</span><a name="line.668"></a>
+<span class="sourceLineNo">669</span> configurePartitioner(job, startKeys, writeMultipleTables);<a name="line.669"></a>
+<span class="sourceLineNo">670</span> // Set compression algorithms based on column families<a name="line.670"></a>
+<span class="sourceLineNo">671</span><a name="line.671"></a>
+<span class="sourceLineNo">672</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(compressionDetails,<a name="line.672"></a>
+<span class="sourceLineNo">673</span> tableDescriptors));<a name="line.673"></a>
+<span class="sourceLineNo">674</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(blockSizeDetails,<a name="line.674"></a>
<span class="sourceLineNo">675</span> tableDescriptors));<a name="line.675"></a>
-<span class="sourceLineNo">676</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(blockSizeDetails,<a name="line.676"></a>
+<span class="sourceLineNo">676</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(bloomTypeDetails,<a name="line.676"></a>
<span class="sourceLineNo">677</span> tableDescriptors));<a name="line.677"></a>
-<span class="sourceLineNo">678</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(bloomTypeDetails,<a name="line.678"></a>
-<span class="sourceLineNo">679</span> tableDescriptors));<a name="line.679"></a>
-<span class="sourceLineNo">680</span> conf.set(BLOOM_PARAM_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(bloomParamDetails,<a name="line.680"></a>
-<span class="sourceLineNo">681</span> tableDescriptors));<a name="line.681"></a>
-<span class="sourceLineNo">682</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.682"></a>
-<span class="sourceLineNo">683</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, tableDescriptors));<a name="line.683"></a>
-<span class="sourceLineNo">684</span><a name="line.684"></a>
-<span class="sourceLineNo">685</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.685"></a>
-<span class="sourceLineNo">686</span> TableMapReduceUtil.initCredentials(job);<a name="line.686"></a>
-<span class="sourceLineNo">687</span> LOG.info("Incremental output configured for tables: " + StringUtils.join(allTableNames, ","));<a name="line.687"></a>
-<span class="sourceLineNo">688</span> }<a name="line.688"></a>
-<span class="sourceLineNo">689</span><a name="line.689"></a>
-<span class="sourceLineNo">690</span> public static void configureIncrementalLoadMap(Job job, TableDescriptor tableDescriptor) throws<a name="line.690"></a>
-<span class="sourceLineNo">691</span> IOException {<a name="line.691"></a>
-<span class="sourceLineNo">692</span> Configuration conf = job.getConfiguration();<a name="line.692"></a>
-<span class="sourceLineNo">693</span><a name="line.693"></a>
-<span class="sourceLineNo">694</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.694"></a>
-<span class="sourceLineNo">695</span> job.setOutputValueClass(MapReduceExtendedCell.class);<a name="line.695"></a>
-<span class="sourceLineNo">696</span> job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.696"></a>
-<span class="sourceLineNo">697</span><a name="line.697"></a>
-<span class="sourceLineNo">698</span> ArrayList<TableDescriptor> singleTableDescriptor = new ArrayList<>(1);<a name="line.698"></a>
-<span class="sourceLineNo">699</span> singleTableDescriptor.add(tableDescriptor);<a name="line.699"></a>
-<span class="sourceLineNo">700</span><a name="line.700"></a>
-<span class="sourceLineNo">701</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, tableDescriptor.getTableName().getNameAsString());<a name="line.701"></a>
-<span class="sourceLineNo">702</span> // Set compression algorithms based on column families<a name="line.702"></a>
-<span class="sourceLineNo">703</span> conf.set(COMPRESSION_FAMILIES_CONF_KEY,<a name="line.703"></a>
-<span class="sourceLineNo">704</span> serializeColumnFamilyAttribute(compressionDetails, singleTableDescriptor));<a name="line.704"></a>
-<span class="sourceLineNo">705</span> conf.set(BLOCK_SIZE_FAMILIES_CONF_KEY,<a name="line.705"></a>
-<span class="sourceLineNo">706</span> serializeColumnFamilyAttribute(blockSizeDetails, singleTableDescriptor));<a name="line.706"></a>
-<span class="sourceLineNo">707</span> conf.set(BLOOM_TYPE_FAMILIES_CONF_KEY,<a name="line.707"></a>
-<span class="sourceLineNo">708</span> serializeColumnFamilyAttribute(bloomTypeDetails, singleTableDescriptor));<a name="line.708"></a>
-<span class="sourceLineNo">709</span> conf.set(BLOOM_PARAM_FAMILIES_CONF_KEY,<a name="line.709"></a>
-<span class="sourceLineNo">710</span> serializeColumnFamilyAttribute(bloomParamDetails, singleTableDescriptor));<a name="line.710"></a>
-<span class="sourceLineNo">711</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.711"></a>
-<span class="sourceLineNo">712</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, singleTableDescriptor));<a name="line.712"></a>
-<span class="sourceLineNo">713</span><a name="line.713"></a>
-<span class="sourceLineNo">714</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.714"></a>
-<span class="sourceLineNo">715</span> TableMapReduceUtil.initCredentials(job);<a name="line.715"></a>
-<span class="sourceLineNo">716</span> LOG.info("Incremental table " + tableDescriptor.getTableName() + " output configured.");<a name="line.716"></a>
-<span class="sourceLineNo">717</span> }<a name="line.717"></a>
-<span class="sourceLineNo">718</span><a name="line.718"></a>
-<span class="sourceLineNo">719</span> /**<a name="line.719"></a>
-<span class="sourceLineNo">720</span> * Runs inside the task to deserialize column family to compression algorithm<a name="line.720"></a>
-<span class="sourceLineNo">721</span> * map from the configuration.<a name="line.721"></a>
-<span class="sourceLineNo">722</span> *<a name="line.722"></a>
-<span class="sourceLineNo">723</span> * @param conf to read the serialized values from<a name="line.723"></a>
-<span class="sourceLineNo">724</span> * @return a map from column family to the configured compression algorithm<a name="line.724"></a>
-<span class="sourceLineNo">725</span> */<a name="line.725"></a>
-<span class="sourceLineNo">726</span> @VisibleForTesting<a name="line.726"></a>
-<span class="sourceLineNo">727</span> static Map<byte[], Algorithm> createFamilyCompressionMap(Configuration<a name="line.727"></a>
-<span class="sourceLineNo">728</span> conf) {<a name="line.728"></a>
-<span class="sourceLineNo">729</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.729"></a>
-<span class="sourceLineNo">730</span> COMPRESSION_FAMILIES_CONF_KEY);<a name="line.730"></a>
-<span class="sourceLineNo">731</span> Map<byte[], Algorithm> compressionMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.731"></a>
-<span class="sourceLineNo">732</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.732"></a>
-<span class="sourceLineNo">733</span> Algorithm algorithm = HFileWriterImpl.compressionByName(e.getValue());<a name="line.733"></a>
-<span class="sourceLineNo">734</span> compressionMap.put(e.getKey(), algorithm);<a name="line.734"></a>
-<span class="sourceLineNo">735</span> }<a name="line.735"></a>
-<span class="sourceLineNo">736</span> return compressionMap;<a name="line.736"></a>
-<span class="sourceLineNo">737</span> }<a name="line.737"></a>
-<span class="sourceLineNo">738</span><a name="line.738"></a>
-<span class="sourceLineNo">739</span> /**<a name="line.739"></a>
-<span class="sourceLineNo">740</span> * Runs inside the task to deserialize column family to bloom filter type<a name="line.740"></a>
-<span class="sourceLineNo">741</span> * map from the configuration.<a name="line.741"></a>
-<span class="sourceLineNo">742</span> *<a name="line.742"></a>
-<span class="sourceLineNo">743</span> * @param conf to read the serialized values from<a name="line.743"></a>
-<span class="sourceLineNo">744</span> * @return a map from column family to the the configured bloom filter type<a name="line.744"></a>
-<span class="sourceLineNo">745</span> */<a name="line.745"></a>
-<span class="sourceLineNo">746</span> @VisibleForTesting<a name="line.746"></a>
-<span class="sourceLineNo">747</span> static Map<byte[], BloomType> createFamilyBloomTypeMap(Configuration conf) {<a name="line.747"></a>
-<span class="sourceLineNo">748</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.748"></a>
-<span class="sourceLineNo">749</span> BLOOM_TYPE_FAMILIES_CONF_KEY);<a name="line.749"></a>
-<span class="sourceLineNo">750</span> Map<byte[], BloomType> bloomTypeMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.750"></a>
-<span class="sourceLineNo">751</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.751"></a>
-<span class="sourceLineNo">752</span> BloomType bloomType = BloomType.valueOf(e.getValue());<a name="line.752"></a>
-<span class="sourceLineNo">753</span> bloomTypeMap.put(e.getKey(), bloomType);<a name="line.753"></a>
-<span class="sourceLineNo">754</span> }<a name="line.754"></a>
-<span class="sourceLineNo">755</span> return bloomTypeMap;<a name="line.755"></a>
-<span class="sourceLineNo">756</span> }<a name="line.756"></a>
-<span class="sourceLineNo">757</span><a name="line.757"></a>
-<span class="sourceLineNo">758</span> /**<a name="line.758"></a>
-<span class="sourceLineNo">759</span> * Runs inside the task to deserialize column family to bloom filter param<a name="line.759"></a>
-<span class="sourceLineNo">760</span> * map from the configuration.<a name="line.760"></a>
-<span class="sourceLineNo">761</span> *<a name="line.761"></a>
-<span class="sourceLineNo">762</span> * @param conf to read the serialized values from<a name="line.762"></a>
-<span class="sourceLineNo">763</span> * @return a map from column family to the the configured bloom filter param<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 Map<byte[], String> createFamilyBloomParamMap(Configuration conf) {<a name="line.766"></a>
-<span class="sourceLineNo">767</span> return createFamilyConfValueMap(conf, BLOOM_PARAM_FAMILIES_CONF_KEY);<a name="line.767"></a>
-<span class="sourceLineNo">768</span> }<a name="line.768"></a>
-<span class="sourceLineNo">769</span><a name="line.769"></a>
-<span class="sourceLineNo">770</span><a name="line.770"></a>
-<span class="sourceLineNo">771</span> /**<a name="line.771"></a>
-<span class="sourceLineNo">772</span> * Runs inside the task to deserialize column family to block size<a name="line.772"></a>
-<span class="sourceLineNo">773</span> * map from the configuration.<a name="line.773"></a>
-<span class="sourceLineNo">774</span> *<a name="line.774"></a>
-<span class="sourceLineNo">775</span> * @param conf to read the serialized values from<a name="line.775"></a>
-<span class="sourceLineNo">776</span> * @return a map from column family to the configured block size<a name="line.776"></a>
-<span class="sourceLineNo">777</span> */<a name="line.777"></a>
-<span class="sourceLineNo">778</span> @VisibleForTesting<a name="line.778"></a>
-<span class="sourceLineNo">779</span> static Map<byte[], Integer> createFamilyBlockSizeMap(Configuration conf) {<a name="line.779"></a>
-<span class="sourceLineNo">780</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.780"></a>
-<span class="sourceLineNo">781</span> BLOCK_SIZE_FAMILIES_CONF_KEY);<a name="line.781"></a>
-<span class="sourceLineNo">782</span> Map<byte[], Integer> blockSizeMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.782"></a>
-<span class="sourceLineNo">783</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.783"></a>
-<span class="sourceLineNo">784</span> Integer blockSize = Integer.parseInt(e.getValue());<a name="line.784"></a>
-<span class="sourceLineNo">785</span> blockSizeMap.put(e.getKey(), blockSize);<a name="line.785"></a>
-<span class="sourceLineNo">786</span> }<a name="line.786"></a>
-<span class="sourceLineNo">787</span> return blockSizeMap;<a name="line.787"></a>
-<span class="sourceLineNo">788</span> }<a name="line.788"></a>
-<span class="sourceLineNo">789</span><a name="line.789"></a>
-<span class="sourceLineNo">790</span> /**<a name="line.790"></a>
-<span class="sourceLineNo">791</span> * Runs inside the task to deserialize column family to data block encoding<a name="line.791"></a>
-<span class="sourceLineNo">792</span> * type map from the configuration.<a name="line.792"></a>
-<span class="sourceLineNo">793</span> *<a name="line.793"></a>
-<span class="sourceLineNo">794</span> * @param conf to read the serialized values from<a name="line.794"></a>
-<span class="sourceLineNo">795</span> * @return a map from column family to HFileDataBlockEncoder for the<a name="line.795"></a>
-<span class="sourceLineNo">796</span> * configured data block type for the family<a name="line.796"></a>
-<span class="sourceLineNo">797</span> */<a name="line.797"></a>
-<span class="sourceLineNo">798</span> @VisibleForTesting<a name="line.798"></a>
-<span class="sourceLineNo">799</span> static Map<byte[], DataBlockEncoding> createFamilyDataBlockEncodingMap(<a name="line.799"></a>
-<span class="sourceLineNo">800</span> Configuration conf) {<a name="line.800"></a>
-<span class="sourceLineNo">801</span> Map<byte[], String> stringMap = createFamilyConfValueMap(conf,<a name="line.801"></a>
-<span class="sourceLineNo">802</span> DATABLOCK_ENCODING_FAMILIES_CONF_KEY);<a name="line.802"></a>
-<span class="sourceLineNo">803</span> Map<byte[], DataBlockEncoding> encoderMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.803"></a>
-<span class="sourceLineNo">804</span> for (Map.Entry<byte[], String> e : stringMap.entrySet()) {<a name="line.804"></a>
-<span class="sourceLineNo">805</span> encoderMap.put(e.getKey(), DataBlockEncoding.valueOf((e.getValue())));<a name="line.805"></a>
-<span class="sourceLineNo">806</span> }<a name="line.806"></a>
-<span class="sourceLineNo">807</span> return encoderMap;<a name="line.807"></a>
-<span class="sourceLineNo">808</span> }<a name="line.808"></a>
-<span class="sourceLineNo">809</span><a name="line.809"></a>
-<span class="sourceLineNo">810</span><a name="line.810"></a>
-<span class="sourceLineNo">811</span> /**<a name="line.811"></a>
-<span class="sourceLineNo">812</span> * Run inside the task to deserialize column family to given conf value map.<a name="line.812"></a>
-<span class="sourceLineNo">813</span> *<a name="line.813"></a>
-<span class="sourceLineNo">814</span> * @param conf to read the serialized values from<a name="line.814"></a>
-<span class="sourceLineNo">815</span> * @param confName conf key to read from the configuration<a name="line.815"></a>
-<span class="sourceLineNo">816</span> * @return a map of column family to the given configuration value<a name="line.816"></a>
-<span class="sourceLineNo">817</span> */<a name="line.817"></a>
-<span class="sourceLineNo">818</span> private static Map<byte[], String> createFamilyConfValueMap(<a name="line.818"></a>
-<span class="sourceLineNo">819</span> Configuration conf, String confName) {<a name="line.819"></a>
-<span class="sourceLineNo">820</span> Map<byte[], String> confValMap = new TreeMap<>(Bytes.BYTES_COMPARATOR);<a name="line.820"></a>
-<span class="sourceLineNo">821</span> String confVal = conf.get(confName, "");<a name="line.821"></a>
-<span class="sourceLineNo">822</span> for (String familyConf : confVal.split("&")) {<a name="line.822"></a>
-<span class="sourceLineNo">823</span> String[] familySplit = familyConf.split("=");<a name="line.823"></a>
-<span class="sourceLineNo">824</span> if (familySplit.length != 2) {<a name="line.824"></a>
-<span class="sourceLineNo">825</span> continue;<a name="line.825"></a>
-<span class="sourceLineNo">826</span> }<a name="line.826"></a>
-<span class="sourceLineNo">827</span> try {<a name="line.827"></a>
-<span class="sourceLineNo">828</span> confValMap.put(Bytes.toBytes(URLDecoder.decode(familySplit[0], "UTF-8")),<a name="line.828"></a>
-<span class="sourceLineNo">829</span> URLDecoder.decode(familySplit[1], "UTF-8"));<a name="line.829"></a>
-<span class="sourceLineNo">830</span> } catch (UnsupportedEncodingException e) {<a name="line.830"></a>
-<span class="sourceLineNo">831</span> // will not happen with UTF-8 encoding<a name="line.831"></a>
-<span class="sourceLineNo">832</span> throw new AssertionError(e);<a name="line.832"></a>
-<span class="sourceLineNo">833</span> }<a name="line.833"></a>
-<span class="sourceLineNo">834</span> }<a name="line.834"></a>
-<span class="sourceLineNo">835</span> return confValMap;<a name="line.835"></a>
-<span class="sourceLineNo">836</span> }<a name="line.836"></a>
-<span class="sourceLineNo">837</span><a name="line.837"></a>
-<span class="sourceLineNo">838</span> /**<a name="line.838"></a>
-<span class="sourceLineNo">839</span> * Configure <code>job</code> with a TotalOrderPartitioner, partitioning against<a name="line.839"></a>
-<span class="sourceLineNo">840</span> * <code>splitPoints</code>. Cleans up the partitions file after job exists.<a name="line.840"></a>
-<span class="sourceLineNo">841</span> */<a name="line.841"></a>
-<span class="sourceLineNo">842</span> static void configurePartitioner(Job job, List<ImmutableBytesWritable> splitPoints, boolean<a name="line.842"></a>
-<span class="sourceLineNo">843</span> writeMultipleTables)<a name="line.843"></a>
-<span class="sourceLineNo">844</span> throws IOException {<a name="line.844"></a>
-<span class="sourceLineNo">845</span> Configuration conf = job.getConfiguration();<a name="line.845"></a>
-<span class="sourceLineNo">846</span> // create the partitions file<a name="line.846"></a>
-<span class="sourceLineNo">847</span> FileSystem fs = FileSystem.get(conf);<a name="line.847"></a>
-<span class="sourceLineNo">848</span> String hbaseTmpFsDir =<a name="line.848"></a>
-<span class="sourceLineNo">849</span> conf.get(HConstants.TEMPORARY_FS_DIRECTORY_KEY,<a name="line.849"></a>
-<span class="sourceLineNo">850</span> fs.getHomeDirectory() + "/hbase-staging");<a name="line.850"></a>
-<span class="sourceLineNo">851</span> Path partitionsPath = new Path(hbaseTmpFsDir, "partitions_" + UUID.randomUUID());<a name="line.851"></a>
-<span class="sourceLineNo">852</span> fs.makeQualified(partitionsPath);<a name="line.852"></a>
-<span class="sourceLineNo">853</span> writePartitions(conf, partitionsPath, splitPoints, writeMultipleTables);<a name="line.853"></a>
-<span class="sourceLineNo">854</span> fs.deleteOnExit(partitionsPath);<a name="line.854"></a>
-<span class="sourceLineNo">855</span><a name="line.855"></a>
-<span class="sourceLineNo">856</span> // configure job to use it<a name="line.856"></a>
-<span class="sourceLineNo">857</span> job.setPartitionerClass(TotalOrderPartitioner.class);<a name="line.857"></a>
-<span class="sourceLineNo">858</span> TotalOrderPartitioner.setPartitionFile(conf, partitionsPath);<a name="line.858"></a>
-<span class="sourceLineNo">859</span> }<a name="line.859"></a>
-<span class="sourceLineNo">860</span><a name="line.860"></a>
-<span class="sourceLineNo">861</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "RCN_REDUNDANT_NULLCHECK_OF_NONNULL_VALUE")<a name="line.861"></a>
-<span class="sourceLineNo">862</span> @VisibleForTesting<a name="line.862"></a>
-<span class="sourceLineNo">863</span> static String serializeColumnFamilyAttribute(Function<ColumnFamilyDescriptor, String> fn, List<TableDescriptor> allTables)<a name="line.863"></a>
-<span class="sourceLineNo">864</span> throws UnsupportedEncodingException {<a name="line.864"></a>
-<span class="sourceLineNo">865</span> StringBuilder attributeValue = new StringBuilder();<a name="line.865"></a>
-<span class="sourceLineNo">866</span> int i = 0;<a name="line.866"></a>
-<span class="sourceLineNo">867</span> for (TableDescriptor tableDescriptor : allTables) {<a name="line.867"></a>
-<span class="sourceLineNo">868</span> if (tableDescriptor == null) {<a name="line.868"></a>
-<span class="sourceLineNo">869</span> // could happen with mock table instance<a name="line.869"></a>
-<span class="sourceLineNo">870</span> // CODEREVIEW: Can I set an empty string in conf if mock table instance?<a name="line.870"></a>
-<span class="sourceLineNo">871</span> return "";<a name="line.871"></a>
-<span class="sourceLineNo">872</span> }<a name="line.872"></a>
-<span class="sourceLineNo">873</span> for (ColumnFamilyDescriptor familyDescriptor : tableDescriptor.getColumnFamilies()) {<a name="line.873"></a>
-<span class="sourceLineNo">874</span> if (i++ > 0) {<a name="line.874"></a>
-<span class="sourceLineNo">875</span> attributeValue.append('&');<a name="line.875"></a>
-<span class="sourceLineNo">876</span> }<a name="line.876"></a>
-<span class="sourceLineNo">877</span> attributeValue.append(URLEncoder.encode(<a name="line.877"></a>
-<span class="sourceLineNo">878</span> Bytes.toString(combineTableNameSuffix(tableDescriptor.getTableName().getName(), familyDescriptor.getName())),<a name="line.878"></a>
-<span class="sourceLineNo">879</span> "UTF-8"));<a name="line.879"></a>
-<span class="sourceLineNo">880</span> attributeValue.append('=');<a name="line.880"></a>
-<span class="sourceLineNo">881</span> attributeValue.append(URLEncoder.encode(fn.apply(familyDescriptor), "UTF-8"));<a name="line.881"></a>
-<span class="sourceLineNo">882</span> }<a name="line.882"></a>
-<span class="sourceLineNo">883</span> }<a name="line.883"></a>
-<span class="sourceLineNo">884</span> // Get rid of the last ampersand<a name="line.884"></a>
-<span class="sourceLineNo">885</span> return attributeValue.toString();<a name="line.885"></a>
-<span class="sourceLineNo">886</span> }<a name="line.886"></a>
-<span class="sourceLineNo">887</span><a name="line.887"></a>
-<span class="sourceLineNo">888</span> /**<a name="line.888"></a>
-<span class="sourceLineNo">889</span> * Serialize column family to compression algorithm map to configuration.<a name="line.889"></a>
-<span class="sourceLineNo">890</span> * Invoked while configuring the MR job for incremental load.<a name="line.890"></a>
-<span class="sourceLineNo">891</span> *<a name="line.891"></a>
-<span class="sourceLineNo">892</span> * @param tableDescriptor to read the properties from<a name="line.892"></a>
-<span class="sourceLineNo">893</span> * @param conf to persist serialized values into<a name="line.893"></a>
-<span class="sourceLineNo">894</span> * @throws IOException<a name="line.894"></a>
-<span class="sourceLineNo">895</span> * on failure to read column family descriptors<a name="line.895"></a>
-<span class="sourceLineNo">896</span> */<a name="line.896"></a>
-<span class="sourceLineNo">897</span> @VisibleForTesting<a name="line.897"></a>
-<span class="sourceLineNo">898</span> static Function<ColumnFamilyDescriptor, String> compressionDetails = familyDescriptor -><a name="line.898"></a>
-<span class="sourceLineNo">899</span> familyDescriptor.getCompressionType().getName();<a name="line.899"></a>
-<span class="sourceLineNo">900</span><a name="line.900"></a>
-<span class="sourceLineNo">901</span> /**<a name="line.901"></a>
-<span class="sourceLineNo">902</span> * Serialize column family to block size map to configuration. Invoked while<a name="line.902"></a>
-<span class="sourceLineNo">903</span> * configuring the MR job for incremental load.<a name="line.903"></a>
-<span class="sourceLineNo">904</span> *<a name="line.904"></a>
-<span class="sourceLineNo">905</span> * @param tableDescriptor<a name="line.905"></a>
-<span class="sourceLineNo">906</span> * to read the properties from<a name="line.906"></a>
-<span class="sourceLineNo">907</span> * @param conf<a name="line.907"></a>
-<span class="sourceLineNo">908</span> * to persist serialized values into<a name="line.908"></a>
-<span class="sourceLineNo">909</span> *<a name="line.909"></a>
-<span class="sourceLineNo">910</span> * @throws IOException<a name="line.910"></a>
-<span class="sourceLineNo">911</span> * on failure to read column family descriptors<a name="line.911"></a>
-<span class="sourceLineNo">912</span> */<a name="line.912"></a>
-<span class="sourceLineNo">913</span> @VisibleForTesting<a name="line.913"></a>
-<span class="sourceLineNo">914</span> static Function<ColumnFamilyDescriptor, String> blockSizeDetails = familyDescriptor -> String<a name="line.914"></a>
-<span class="sourceLineNo">915</span> .valueOf(familyDescriptor.getBlocksize());<a name="line.915"></a>
-<span class="sourceLineNo">916</span><a name="line.916"></a>
-<span class="sourceLineNo">917</span> /**<a name="line.917"></a>
-<span class="sourceLineNo">918</span> * Serialize column family to bloom type map to configuration. Invoked while<a name="line.918"></a>
-<span class="sourceLineNo">919</span> * configuring the MR job for incremental load.<a name="line.919"></a>
-<span class="sourceLineNo">920</span> *<a name="line.920"></a>
-<span class="sourceLineNo">921</span> * @param tableDescriptor<a name="line.921"></a>
-<span class="sourceLineNo">922</span> * to read the properties from<a name="line.922"></a>
-<span class="sourceLineNo">923</span> * @param conf<a name="line.923"></a>
-<span class="sourceLineNo">924</span> * to persist serialized values into<a name="line.924"></a>
-<span class="sourceLineNo">925</span> *<a name="line.925"></a>
-<span class="sourceLineNo">926</span> * @throws IOException<a name="line.926"></a>
-<span class="sourceLineNo">927</span> * on failure to read column family descriptors<a name="line.927"></a>
-<span class="sourceLineNo">928</span> */<a name="line.928"></a>
-<span class="sourceLineNo">929</span> @VisibleForTesting<a name="line.929"></a>
-<span class="sourceLineNo">930</span> static Function<ColumnFamilyDescriptor, String> bloomTypeDetails = familyDescriptor -> {<a name="line.930"></a>
-<span class="sourceLineNo">931</span> String bloomType = familyDescriptor.getBloomFilterType().toString();<a name="line.931"></a>
-<span class="sourceLineNo">932</span> if (bloomType == null) {<a name="line.932"></a>
-<span class="sourceLineNo">933</span> bloomType = ColumnFamilyDescriptorBuilder.DEFAULT_BLOOMFILTER.name();<a name="line.933"></a>
-<span class="sourceLineNo">934</span> }<a name="line.934"></a>
-<span class="sourceLineNo">935</span> return bloomType;<a name="line.935"></a>
-<span class="sourceLineNo">936</span> };<a name="line.936"></a>
-<span class="sourceLineNo">937</span><a name="line.937"></a>
-<span class="sourceLineNo">938</span> /**<a name="line.938"></a>
-<span class="sourceLineNo">939</span> * Serialize column family to bloom param map to configuration. Invoked while<a name="line.939"></a>
-<span class="sourceLineNo">940</span> * configuring the MR job for incremental load.<a name="line.940"></a>
-<span class="sourceLineNo">941</span> *<a name="line.941"></a>
-<span class="sourceLineNo">942</span> * @param tableDescriptor<a name="line.942"></a>
-<span class="sourceLineNo">943</span> * to read the properties from<a name="line.943"></a>
-<span class="sourceLineNo">944</span> * @param conf<a name="line.944"></a>
-<span class="sourceLineNo">945</span> * to persist serialized values into<a name="line.945"></a>
-<span class="sourceLineNo">946</span> *<a name="line.946"></a>
-<span class="sourceLineNo">947</span> * @throws IOException<a name="line.947"></a>
-<span class="sourceLineNo">948</span> * on failure to read column family descriptors<a name="line.948"></a>
-<span class="sourceLineNo">949</span> */<a name="line.949"></a>
-<span class="sourceLineNo">950</span> @VisibleForTesting<a name="line.950"></a>
-<span class="sourceLineNo">951</span> static Function<ColumnFamilyDescriptor, String> bloomParamDetails = familyDescriptor -> {<a name="line.951"></a>
-<span class="sourceLineNo">952</span> BloomType bloomType = familyDescriptor.getBloomFilterType();<a name="line.952"></a>
-<span class="sourceLineNo">953</span> String bloomParam = "";<a name="line.953"></a>
-<span class="sourceLineNo">954</span> if (bloomType == BloomType.ROWPREFIX_FIXED_LENGTH) {<a name="line.954"></a>
-<span class="sourceLineNo">955</span> bloomParam = familyDescriptor.getConfigurationValue(BloomFilterUtil.PREFIX_LENGTH_KEY);<a name="line.955"></a>
-<span class="sourceLineNo">956</span> } else if (bloomType == BloomType.ROWPREFIX_DELIMITED) {<a name="line.956"></a>
-<span class="sourceLineNo">957</span> bloomParam = familyDescriptor.getConfigurationValue(BloomFilterUtil.DELIMITER_KEY);<a name="line.957"></a>
-<span class="sourceLineNo">958</span> }<a name="line.958"></a>
-<span class="sourceLineNo">959</span> return bloomParam;<a name="line.959"></a>
-<span class="sourceLineNo">960</span> };<a name="line.960"></a>
-<span class="sourceLineNo">961</span><a name="line.961"></a>
-<span class="sourceLineNo">962</span> /**<a name="line.962"></a>
-<span class="sourceLineNo">963</span> * Serialize column family to data block encoding map to configuration.<a name="line.963"></a>
-<span class="sourceLineNo">964</span> * Invoked while configuring the MR job for incremental load.<a name="line.964"></a>
-<span class="sourceLineNo">965</span> *<a name="line.965"></a>
-<span class="sourceLineNo">966</span> * @param tableDescriptor<a name="line.966"></a>
-<span class="sourceLineNo">967</span> * to read the properties from<a name="line.967"></a>
-<span class="sourceLineNo">968</span> * @param conf<a name="line.968"></a>
-<span class="sourceLineNo">969</span> * to persist serialized values into<a name="line.969"></a>
-<span class="sourceLineNo">970</span> * @throws IOException<a name="line.970"></a>
-<span class="sourceLineNo">971</span> * on failure to read column family descriptors<a name="line.971"></a>
-<span class="sourceLineNo">972</span> */<a name="line.972"></a>
-<span class="sourceLineNo">973</span> @VisibleForTesting<a name="line.973"></a>
-<span class="sourceLineNo">974</span> static Function<ColumnFamilyDescriptor, String> dataBlockEncodingDetails = familyDescriptor -> {<a name="line.974"></a>
-<span class="sourceLineNo">975</span> DataBlockEncoding encoding = familyDescriptor.getDataBlockEncoding();<a name="line.975"></a>
-<span class="sourceLineNo">976</span> if (encoding == null) {<a name="line.976"></a>
-<span class="sourceLineNo">977</span> encoding = DataBlockEncoding.NONE;<a name="line.977"></a>
-<span class="sourceLineNo">978</span> }<a name="line.978"></a>
-<span class="sourceLineNo">979</span> return encoding.toString();<a name="line.979"></a>
-<span class="sourceLineNo">980</span> };<a name="line.980"></a>
-<span class="sourceLineNo">981</span><a name="line.981"></a>
-<span class="sourceLineNo">982</span>}<a name="line.982"></a>
+<span class="sourceLineNo">678</span> conf.set(BLOOM_PARAM_FAMILIES_CONF_KEY, serializeColumnFamilyAttribute(bloomParamDetails,<a name="line.678"></a>
+<span class="sourceLineNo">679</span> tableDescriptors));<a name="line.679"></a>
+<span class="sourceLineNo">680</span> conf.set(DATABLOCK_ENCODING_FAMILIES_CONF_KEY,<a name="line.680"></a>
+<span class="sourceLineNo">681</span> serializeColumnFamilyAttribute(dataBlockEncodingDetails, tableDescriptors));<a name="line.681"></a>
+<span class="sourceLineNo">682</span><a name="line.682"></a>
+<span class="sourceLineNo">683</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.683"></a>
+<span class="sourceLineNo">684</span> TableMapReduceUtil.initCredentials(job);<a name="line.684"></a>
+<span class="sourceLineNo">685</span> LOG.info("Incremental output configured for tables: " + StringUtils.join(allTableNames, ","));<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> public static void configureIncrementalLoadMap(Job job, TableDescriptor tableDescriptor) throws<a name="line.688"></a>
+<span class="sourceLineNo">689</span> IOException {<a name="line.689"></a>
+<span class="sourceLineNo">690</span> Configuration conf = job.getConfiguration();<a name="line.690"></a>
+<span class="sourceLineNo">691</span><a name="line.691"></a>
+<span class="sourceLineNo">692</span> job.setOutputKeyClass(ImmutableBytesWritable.class);<a name="line.692"></a>
+<span class="sourceLineNo">693</span> job.setOutputValueClass(MapReduceExtendedCell.class);<a name="line.693"></a>
+<span class="sourceLineNo">694</span> job.setOutputFormatClass(HFileOutputFormat2.class);<a name="line.694"></a>
+<span class="sourceLineNo">695</span><a name="line.695"></a>
+<span class="sourceLineNo">696</span> ArrayList<TableDescriptor> singleTableDescriptor = new ArrayList<>(1);<a name="line.696"></a>
+<span class="sourceLineNo">697</span> singleTableDescriptor.add(tableDescriptor);<a name="line.697"></a>
+<span class="sourceLineNo">698</span><a name="line.698"></a>
+<span class="sourceLineNo">699</span> conf.set(OUTPUT_TABLE_NAME_CONF_KEY, tableDescriptor.getTableName().getNameAsString());<a name="line.699"></a>
+<span class="sourceLineNo">700</span> // Set compression algorithms based on column familie
<TRUNCATED>