http://git-wip-us.apache.org/repos/asf/hbase-site/blob/6a13df3e/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestHFileOutputFormat2.html ---------------------------------------------------------------------- diff --git a/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestHFileOutputFormat2.html b/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestHFileOutputFormat2.html index f045a97..5692397 100644 --- a/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestHFileOutputFormat2.html +++ b/testdevapidocs/src-html/org/apache/hadoop/hbase/mapreduce/TestHFileOutputFormat2.html @@ -357,817 +357,818 @@ <span class="sourceLineNo">349</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.349"></a> <span class="sourceLineNo">350</span> public void testJobConfiguration() throws Exception {<a name="line.350"></a> <span class="sourceLineNo">351</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.351"></a> -<span class="sourceLineNo">352</span> conf.set("hbase.fs.tmp.dir", util.getDataTestDir("testJobConfiguration").toString());<a name="line.352"></a> -<span class="sourceLineNo">353</span> Job job = new Job(conf);<a name="line.353"></a> -<span class="sourceLineNo">354</span> job.setWorkingDirectory(util.getDataTestDir("testJobConfiguration"));<a name="line.354"></a> -<span class="sourceLineNo">355</span> Table table = Mockito.mock(Table.class);<a name="line.355"></a> -<span class="sourceLineNo">356</span> RegionLocator regionLocator = Mockito.mock(RegionLocator.class);<a name="line.356"></a> -<span class="sourceLineNo">357</span> setupMockStartKeys(regionLocator);<a name="line.357"></a> -<span class="sourceLineNo">358</span> setupMockTableName(regionLocator);<a name="line.358"></a> -<span class="sourceLineNo">359</span> HFileOutputFormat2.configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.359"></a> -<span class="sourceLineNo">360</span> assertEquals(job.getNumReduceTasks(), 4);<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> private byte [][] generateRandomStartKeys(int numKeys) {<a name="line.363"></a> -<span class="sourceLineNo">364</span> Random random = new Random();<a name="line.364"></a> -<span class="sourceLineNo">365</span> byte[][] ret = new byte[numKeys][];<a name="line.365"></a> -<span class="sourceLineNo">366</span> // first region start key is always empty<a name="line.366"></a> -<span class="sourceLineNo">367</span> ret[0] = HConstants.EMPTY_BYTE_ARRAY;<a name="line.367"></a> -<span class="sourceLineNo">368</span> for (int i = 1; i < numKeys; i++) {<a name="line.368"></a> -<span class="sourceLineNo">369</span> ret[i] =<a name="line.369"></a> -<span class="sourceLineNo">370</span> PerformanceEvaluation.generateData(random, PerformanceEvaluation.DEFAULT_VALUE_LENGTH);<a name="line.370"></a> -<span class="sourceLineNo">371</span> }<a name="line.371"></a> -<span class="sourceLineNo">372</span> return ret;<a name="line.372"></a> -<span class="sourceLineNo">373</span> }<a name="line.373"></a> -<span class="sourceLineNo">374</span><a name="line.374"></a> -<span class="sourceLineNo">375</span> private byte[][] generateRandomSplitKeys(int numKeys) {<a name="line.375"></a> -<span class="sourceLineNo">376</span> Random random = new Random();<a name="line.376"></a> -<span class="sourceLineNo">377</span> byte[][] ret = new byte[numKeys][];<a name="line.377"></a> -<span class="sourceLineNo">378</span> for (int i = 0; i < numKeys; i++) {<a name="line.378"></a> -<span class="sourceLineNo">379</span> ret[i] =<a name="line.379"></a> -<span class="sourceLineNo">380</span> PerformanceEvaluation.generateData(random, PerformanceEvaluation.DEFAULT_VALUE_LENGTH);<a name="line.380"></a> -<span class="sourceLineNo">381</span> }<a name="line.381"></a> -<span class="sourceLineNo">382</span> return ret;<a name="line.382"></a> -<span class="sourceLineNo">383</span> }<a name="line.383"></a> -<span class="sourceLineNo">384</span><a name="line.384"></a> -<span class="sourceLineNo">385</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.385"></a> -<span class="sourceLineNo">386</span> public void testMRIncrementalLoad() throws Exception {<a name="line.386"></a> -<span class="sourceLineNo">387</span> LOG.info("\nStarting test testMRIncrementalLoad\n");<a name="line.387"></a> -<span class="sourceLineNo">388</span> doIncrementalLoadTest(false, false);<a name="line.388"></a> -<span class="sourceLineNo">389</span> }<a name="line.389"></a> -<span class="sourceLineNo">390</span><a name="line.390"></a> -<span class="sourceLineNo">391</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.391"></a> -<span class="sourceLineNo">392</span> public void testMRIncrementalLoadWithSplit() throws Exception {<a name="line.392"></a> -<span class="sourceLineNo">393</span> LOG.info("\nStarting test testMRIncrementalLoadWithSplit\n");<a name="line.393"></a> -<span class="sourceLineNo">394</span> doIncrementalLoadTest(true, false);<a name="line.394"></a> -<span class="sourceLineNo">395</span> }<a name="line.395"></a> -<span class="sourceLineNo">396</span><a name="line.396"></a> -<span class="sourceLineNo">397</span> /**<a name="line.397"></a> -<span class="sourceLineNo">398</span> * Test for HFileOutputFormat2.LOCALITY_SENSITIVE_CONF_KEY = true<a name="line.398"></a> -<span class="sourceLineNo">399</span> * This test could only check the correctness of original logic if LOCALITY_SENSITIVE_CONF_KEY<a name="line.399"></a> -<span class="sourceLineNo">400</span> * is set to true. Because MiniHBaseCluster always run with single hostname (and different ports),<a name="line.400"></a> -<span class="sourceLineNo">401</span> * it's not possible to check the region locality by comparing region locations and DN hostnames.<a name="line.401"></a> -<span class="sourceLineNo">402</span> * When MiniHBaseCluster supports explicit hostnames parameter (just like MiniDFSCluster does),<a name="line.402"></a> -<span class="sourceLineNo">403</span> * we could test region locality features more easily.<a name="line.403"></a> -<span class="sourceLineNo">404</span> */<a name="line.404"></a> -<span class="sourceLineNo">405</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.405"></a> -<span class="sourceLineNo">406</span> public void testMRIncrementalLoadWithLocality() throws Exception {<a name="line.406"></a> -<span class="sourceLineNo">407</span> LOG.info("\nStarting test testMRIncrementalLoadWithLocality\n");<a name="line.407"></a> -<span class="sourceLineNo">408</span> doIncrementalLoadTest(false, true);<a name="line.408"></a> -<span class="sourceLineNo">409</span> doIncrementalLoadTest(true, true);<a name="line.409"></a> -<span class="sourceLineNo">410</span> }<a name="line.410"></a> -<span class="sourceLineNo">411</span><a name="line.411"></a> -<span class="sourceLineNo">412</span> private void doIncrementalLoadTest(boolean shouldChangeRegions, boolean shouldKeepLocality)<a name="line.412"></a> -<span class="sourceLineNo">413</span> throws Exception {<a name="line.413"></a> -<span class="sourceLineNo">414</span> util = new HBaseTestingUtility();<a name="line.414"></a> -<span class="sourceLineNo">415</span> Configuration conf = util.getConfiguration();<a name="line.415"></a> -<span class="sourceLineNo">416</span> conf.setBoolean(HFileOutputFormat2.LOCALITY_SENSITIVE_CONF_KEY, shouldKeepLocality);<a name="line.416"></a> -<span class="sourceLineNo">417</span> int hostCount = 1;<a name="line.417"></a> -<span class="sourceLineNo">418</span> int regionNum = 5;<a name="line.418"></a> -<span class="sourceLineNo">419</span> if(shouldKeepLocality) {<a name="line.419"></a> -<span class="sourceLineNo">420</span> // We should change host count higher than hdfs replica count when MiniHBaseCluster supports<a name="line.420"></a> -<span class="sourceLineNo">421</span> // explicit hostnames parameter just like MiniDFSCluster does.<a name="line.421"></a> -<span class="sourceLineNo">422</span> hostCount = 3;<a name="line.422"></a> -<span class="sourceLineNo">423</span> regionNum = 20;<a name="line.423"></a> -<span class="sourceLineNo">424</span> }<a name="line.424"></a> -<span class="sourceLineNo">425</span><a name="line.425"></a> -<span class="sourceLineNo">426</span> byte[][] splitKeys = generateRandomSplitKeys(regionNum - 1);<a name="line.426"></a> -<span class="sourceLineNo">427</span> String[] hostnames = new String[hostCount];<a name="line.427"></a> -<span class="sourceLineNo">428</span> for(int i = 0; i < hostCount; ++i) {<a name="line.428"></a> -<span class="sourceLineNo">429</span> hostnames[i] = "datanode_" + i;<a name="line.429"></a> -<span class="sourceLineNo">430</span> }<a name="line.430"></a> -<span class="sourceLineNo">431</span> util.startMiniCluster(1, hostCount, hostnames);<a name="line.431"></a> -<span class="sourceLineNo">432</span><a name="line.432"></a> -<span class="sourceLineNo">433</span> Table table = util.createTable(TABLE_NAME, FAMILIES, splitKeys);<a name="line.433"></a> -<span class="sourceLineNo">434</span> Path testDir = util.getDataTestDirOnTestFS("testLocalMRIncrementalLoad");<a name="line.434"></a> -<span class="sourceLineNo">435</span> try (RegionLocator r = util.getConnection().getRegionLocator(TABLE_NAME); Admin admin =<a name="line.435"></a> -<span class="sourceLineNo">436</span> util.getConnection().getAdmin();) {<a name="line.436"></a> -<span class="sourceLineNo">437</span> assertEquals("Should start with empty table", 0, util.countRows(table));<a name="line.437"></a> -<span class="sourceLineNo">438</span> int numRegions = r.getStartKeys().length;<a name="line.438"></a> -<span class="sourceLineNo">439</span> assertEquals("Should make " + regionNum + " regions", numRegions, regionNum);<a name="line.439"></a> -<span class="sourceLineNo">440</span><a name="line.440"></a> -<span class="sourceLineNo">441</span> // Generate the bulk load files<a name="line.441"></a> -<span class="sourceLineNo">442</span> runIncrementalPELoad(conf, table.getTableDescriptor(), r, testDir);<a name="line.442"></a> -<span class="sourceLineNo">443</span> // This doesn't write into the table, just makes files<a name="line.443"></a> -<span class="sourceLineNo">444</span> assertEquals("HFOF should not touch actual table", 0, util.countRows(table));<a name="line.444"></a> -<span class="sourceLineNo">445</span><a name="line.445"></a> -<span class="sourceLineNo">446</span> // Make sure that a directory was created for every CF<a name="line.446"></a> -<span class="sourceLineNo">447</span> int dir = 0;<a name="line.447"></a> -<span class="sourceLineNo">448</span> for (FileStatus f : testDir.getFileSystem(conf).listStatus(testDir)) {<a name="line.448"></a> -<span class="sourceLineNo">449</span> for (byte[] family : FAMILIES) {<a name="line.449"></a> -<span class="sourceLineNo">450</span> if (Bytes.toString(family).equals(f.getPath().getName())) {<a name="line.450"></a> -<span class="sourceLineNo">451</span> ++dir;<a name="line.451"></a> -<span class="sourceLineNo">452</span> }<a name="line.452"></a> -<span class="sourceLineNo">453</span> }<a name="line.453"></a> -<span class="sourceLineNo">454</span> }<a name="line.454"></a> -<span class="sourceLineNo">455</span> assertEquals("Column family not found in FS.", FAMILIES.length, dir);<a name="line.455"></a> -<span class="sourceLineNo">456</span><a name="line.456"></a> -<span class="sourceLineNo">457</span> // handle the split case<a name="line.457"></a> -<span class="sourceLineNo">458</span> if (shouldChangeRegions) {<a name="line.458"></a> -<span class="sourceLineNo">459</span> LOG.info("Changing regions in table");<a name="line.459"></a> -<span class="sourceLineNo">460</span> admin.disableTable(table.getName());<a name="line.460"></a> -<span class="sourceLineNo">461</span> while(util.getMiniHBaseCluster().getMaster().getAssignmentManager().<a name="line.461"></a> -<span class="sourceLineNo">462</span> getRegionStates().isRegionsInTransition()) {<a name="line.462"></a> -<span class="sourceLineNo">463</span> Threads.sleep(200);<a name="line.463"></a> -<span class="sourceLineNo">464</span> LOG.info("Waiting on table to finish disabling");<a name="line.464"></a> -<span class="sourceLineNo">465</span> }<a name="line.465"></a> -<span class="sourceLineNo">466</span> util.deleteTable(table.getName());<a name="line.466"></a> -<span class="sourceLineNo">467</span> byte[][] newSplitKeys = generateRandomSplitKeys(14);<a name="line.467"></a> -<span class="sourceLineNo">468</span> table = util.createTable(TABLE_NAME, FAMILIES, newSplitKeys);<a name="line.468"></a> -<span class="sourceLineNo">469</span><a name="line.469"></a> -<span class="sourceLineNo">470</span> while (util.getConnection().getRegionLocator(TABLE_NAME)<a name="line.470"></a> -<span class="sourceLineNo">471</span> .getAllRegionLocations().size() != 15 ||<a name="line.471"></a> -<span class="sourceLineNo">472</span> !admin.isTableAvailable(table.getName())) {<a name="line.472"></a> -<span class="sourceLineNo">473</span> Thread.sleep(200);<a name="line.473"></a> -<span class="sourceLineNo">474</span> LOG.info("Waiting for new region assignment to happen");<a name="line.474"></a> -<span class="sourceLineNo">475</span> }<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> // Perform the actual load<a name="line.478"></a> -<span class="sourceLineNo">479</span> new LoadIncrementalHFiles(conf).doBulkLoad(testDir, admin, table, r);<a name="line.479"></a> -<span class="sourceLineNo">480</span><a name="line.480"></a> -<span class="sourceLineNo">481</span> // Ensure data shows up<a name="line.481"></a> -<span class="sourceLineNo">482</span> int expectedRows = NMapInputFormat.getNumMapTasks(conf) * ROWSPERSPLIT;<a name="line.482"></a> -<span class="sourceLineNo">483</span> assertEquals("LoadIncrementalHFiles should put expected data in table",<a name="line.483"></a> -<span class="sourceLineNo">484</span> expectedRows, util.countRows(table));<a name="line.484"></a> -<span class="sourceLineNo">485</span> Scan scan = new Scan();<a name="line.485"></a> -<span class="sourceLineNo">486</span> ResultScanner results = table.getScanner(scan);<a name="line.486"></a> -<span class="sourceLineNo">487</span> for (Result res : results) {<a name="line.487"></a> -<span class="sourceLineNo">488</span> assertEquals(FAMILIES.length, res.rawCells().length);<a name="line.488"></a> -<span class="sourceLineNo">489</span> Cell first = res.rawCells()[0];<a name="line.489"></a> -<span class="sourceLineNo">490</span> for (Cell kv : res.rawCells()) {<a name="line.490"></a> -<span class="sourceLineNo">491</span> assertTrue(CellUtil.matchingRow(first, kv));<a name="line.491"></a> -<span class="sourceLineNo">492</span> assertTrue(Bytes.equals(CellUtil.cloneValue(first), CellUtil.cloneValue(kv)));<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> results.close();<a name="line.495"></a> -<span class="sourceLineNo">496</span> String tableDigestBefore = util.checksumRows(table);<a name="line.496"></a> -<span class="sourceLineNo">497</span><a name="line.497"></a> -<span class="sourceLineNo">498</span> // Check region locality<a name="line.498"></a> -<span class="sourceLineNo">499</span> HDFSBlocksDistribution hbd = new HDFSBlocksDistribution();<a name="line.499"></a> -<span class="sourceLineNo">500</span> for (HRegion region : util.getHBaseCluster().getRegions(TABLE_NAME)) {<a name="line.500"></a> -<span class="sourceLineNo">501</span> hbd.add(region.getHDFSBlocksDistribution());<a name="line.501"></a> -<span class="sourceLineNo">502</span> }<a name="line.502"></a> -<span class="sourceLineNo">503</span> for (String hostname : hostnames) {<a name="line.503"></a> -<span class="sourceLineNo">504</span> float locality = hbd.getBlockLocalityIndex(hostname);<a name="line.504"></a> -<span class="sourceLineNo">505</span> LOG.info("locality of [" + hostname + "]: " + locality);<a name="line.505"></a> -<span class="sourceLineNo">506</span> assertEquals(100, (int) (locality * 100));<a name="line.506"></a> -<span class="sourceLineNo">507</span> }<a name="line.507"></a> -<span class="sourceLineNo">508</span><a name="line.508"></a> -<span class="sourceLineNo">509</span> // Cause regions to reopen<a name="line.509"></a> -<span class="sourceLineNo">510</span> admin.disableTable(TABLE_NAME);<a name="line.510"></a> -<span class="sourceLineNo">511</span> while (!admin.isTableDisabled(TABLE_NAME)) {<a name="line.511"></a> -<span class="sourceLineNo">512</span> Thread.sleep(200);<a name="line.512"></a> -<span class="sourceLineNo">513</span> LOG.info("Waiting for table to disable");<a name="line.513"></a> -<span class="sourceLineNo">514</span> }<a name="line.514"></a> -<span class="sourceLineNo">515</span> admin.enableTable(TABLE_NAME);<a name="line.515"></a> -<span class="sourceLineNo">516</span> util.waitTableAvailable(TABLE_NAME);<a name="line.516"></a> -<span class="sourceLineNo">517</span> assertEquals("Data should remain after reopening of regions",<a name="line.517"></a> -<span class="sourceLineNo">518</span> tableDigestBefore, util.checksumRows(table));<a name="line.518"></a> -<span class="sourceLineNo">519</span> } finally {<a name="line.519"></a> -<span class="sourceLineNo">520</span> testDir.getFileSystem(conf).delete(testDir, true);<a name="line.520"></a> -<span class="sourceLineNo">521</span> util.deleteTable(TABLE_NAME);<a name="line.521"></a> -<span class="sourceLineNo">522</span> util.shutdownMiniCluster();<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> private void runIncrementalPELoad(Configuration conf, HTableDescriptor tableDescriptor,<a name="line.526"></a> -<span class="sourceLineNo">527</span> RegionLocator regionLocator, Path outDir) throws IOException, UnsupportedEncodingException,<a name="line.527"></a> -<span class="sourceLineNo">528</span> InterruptedException, ClassNotFoundException {<a name="line.528"></a> -<span class="sourceLineNo">529</span> Job job = new Job(conf, "testLocalMRIncrementalLoad");<a name="line.529"></a> -<span class="sourceLineNo">530</span> job.setWorkingDirectory(util.getDataTestDirOnTestFS("runIncrementalPELoad"));<a name="line.530"></a> -<span class="sourceLineNo">531</span> job.getConfiguration().setStrings("io.serializations", conf.get("io.serializations"),<a name="line.531"></a> -<span class="sourceLineNo">532</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.532"></a> -<span class="sourceLineNo">533</span> KeyValueSerialization.class.getName());<a name="line.533"></a> -<span class="sourceLineNo">534</span> setupRandomGeneratorMapper(job);<a name="line.534"></a> -<span class="sourceLineNo">535</span> HFileOutputFormat2.configureIncrementalLoad(job, tableDescriptor, regionLocator);<a name="line.535"></a> -<span class="sourceLineNo">536</span> FileOutputFormat.setOutputPath(job, outDir);<a name="line.536"></a> -<span class="sourceLineNo">537</span><a name="line.537"></a> -<span class="sourceLineNo">538</span> assertFalse(util.getTestFileSystem().exists(outDir)) ;<a name="line.538"></a> -<span class="sourceLineNo">539</span><a name="line.539"></a> -<span class="sourceLineNo">540</span> assertEquals(regionLocator.getAllRegionLocations().size(), job.getNumReduceTasks());<a name="line.540"></a> -<span class="sourceLineNo">541</span><a name="line.541"></a> -<span class="sourceLineNo">542</span> assertTrue(job.waitForCompletion(true));<a name="line.542"></a> -<span class="sourceLineNo">543</span> }<a name="line.543"></a> -<span class="sourceLineNo">544</span><a name="line.544"></a> -<span class="sourceLineNo">545</span> /**<a name="line.545"></a> -<span class="sourceLineNo">546</span> * Test for {@link HFileOutputFormat2#configureCompression(org.apache.hadoop.hbase.client.Table,<a name="line.546"></a> -<span class="sourceLineNo">547</span> * Configuration)} and {@link HFileOutputFormat2#createFamilyCompressionMap<a name="line.547"></a> -<span class="sourceLineNo">548</span> * (Configuration)}.<a name="line.548"></a> -<span class="sourceLineNo">549</span> * Tests that the compression map is correctly serialized into<a name="line.549"></a> -<span class="sourceLineNo">550</span> * and deserialized from configuration<a name="line.550"></a> -<span class="sourceLineNo">551</span> *<a name="line.551"></a> -<span class="sourceLineNo">552</span> * @throws IOException<a name="line.552"></a> -<span class="sourceLineNo">553</span> */<a name="line.553"></a> -<span class="sourceLineNo">554</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.554"></a> -<span class="sourceLineNo">555</span> public void testSerializeDeserializeFamilyCompressionMap() throws IOException {<a name="line.555"></a> -<span class="sourceLineNo">556</span> for (int numCfs = 0; numCfs <= 3; numCfs++) {<a name="line.556"></a> -<span class="sourceLineNo">557</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.557"></a> -<span class="sourceLineNo">558</span> Map<String, Compression.Algorithm> familyToCompression =<a name="line.558"></a> -<span class="sourceLineNo">559</span> getMockColumnFamiliesForCompression(numCfs);<a name="line.559"></a> -<span class="sourceLineNo">560</span> Table table = Mockito.mock(HTable.class);<a name="line.560"></a> -<span class="sourceLineNo">561</span> setupMockColumnFamiliesForCompression(table, familyToCompression);<a name="line.561"></a> -<span class="sourceLineNo">562</span> HFileOutputFormat2.configureCompression(conf, table.getTableDescriptor());<a name="line.562"></a> -<span class="sourceLineNo">563</span><a name="line.563"></a> -<span class="sourceLineNo">564</span> // read back family specific compression setting from the configuration<a name="line.564"></a> -<span class="sourceLineNo">565</span> Map<byte[], Algorithm> retrievedFamilyToCompressionMap = HFileOutputFormat2<a name="line.565"></a> -<span class="sourceLineNo">566</span> .createFamilyCompressionMap(conf);<a name="line.566"></a> -<span class="sourceLineNo">567</span><a name="line.567"></a> -<span class="sourceLineNo">568</span> // test that we have a value for all column families that matches with the<a name="line.568"></a> -<span class="sourceLineNo">569</span> // used mock values<a name="line.569"></a> -<span class="sourceLineNo">570</span> for (Entry<String, Algorithm> entry : familyToCompression.entrySet()) {<a name="line.570"></a> -<span class="sourceLineNo">571</span> assertEquals("Compression configuration incorrect for column family:"<a name="line.571"></a> -<span class="sourceLineNo">572</span> + entry.getKey(), entry.getValue(),<a name="line.572"></a> -<span class="sourceLineNo">573</span> retrievedFamilyToCompressionMap.get(entry.getKey().getBytes()));<a name="line.573"></a> -<span class="sourceLineNo">574</span> }<a name="line.574"></a> -<span class="sourceLineNo">575</span> }<a name="line.575"></a> -<span class="sourceLineNo">576</span> }<a name="line.576"></a> -<span class="sourceLineNo">577</span><a name="line.577"></a> -<span class="sourceLineNo">578</span> private void setupMockColumnFamiliesForCompression(Table table,<a name="line.578"></a> -<span class="sourceLineNo">579</span> Map<String, Compression.Algorithm> familyToCompression) throws IOException {<a name="line.579"></a> -<span class="sourceLineNo">580</span> HTableDescriptor mockTableDescriptor = new HTableDescriptor(TABLE_NAME);<a name="line.580"></a> -<span class="sourceLineNo">581</span> for (Entry<String, Compression.Algorithm> entry : familyToCompression.entrySet()) {<a name="line.581"></a> -<span class="sourceLineNo">582</span> mockTableDescriptor.addFamily(new HColumnDescriptor(entry.getKey())<a name="line.582"></a> -<span class="sourceLineNo">583</span> .setMaxVersions(1)<a name="line.583"></a> -<span class="sourceLineNo">584</span> .setCompressionType(entry.getValue())<a name="line.584"></a> -<span class="sourceLineNo">585</span> .setBlockCacheEnabled(false)<a name="line.585"></a> -<span class="sourceLineNo">586</span> .setTimeToLive(0));<a name="line.586"></a> -<span class="sourceLineNo">587</span> }<a name="line.587"></a> -<span class="sourceLineNo">588</span> Mockito.doReturn(mockTableDescriptor).when(table).getTableDescriptor();<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> /**<a name="line.591"></a> -<span class="sourceLineNo">592</span> * @return a map from column family names to compression algorithms for<a name="line.592"></a> -<span class="sourceLineNo">593</span> * testing column family compression. Column family names have special characters<a name="line.593"></a> -<span class="sourceLineNo">594</span> */<a name="line.594"></a> -<span class="sourceLineNo">595</span> private Map<String, Compression.Algorithm><a name="line.595"></a> -<span class="sourceLineNo">596</span> getMockColumnFamiliesForCompression (int numCfs) {<a name="line.596"></a> -<span class="sourceLineNo">597</span> Map<String, Compression.Algorithm> familyToCompression<a name="line.597"></a> -<span class="sourceLineNo">598</span> = new HashMap<String, Compression.Algorithm>();<a name="line.598"></a> -<span class="sourceLineNo">599</span> // use column family names having special characters<a name="line.599"></a> -<span class="sourceLineNo">600</span> if (numCfs-- > 0) {<a name="line.600"></a> -<span class="sourceLineNo">601</span> familyToCompression.put("Family1!@#!@#&", Compression.Algorithm.LZO);<a name="line.601"></a> -<span class="sourceLineNo">602</span> }<a name="line.602"></a> -<span class="sourceLineNo">603</span> if (numCfs-- > 0) {<a name="line.603"></a> -<span class="sourceLineNo">604</span> familyToCompression.put("Family2=asdads&!AASD", Compression.Algorithm.SNAPPY);<a name="line.604"></a> -<span class="sourceLineNo">605</span> }<a name="line.605"></a> -<span class="sourceLineNo">606</span> if (numCfs-- > 0) {<a name="line.606"></a> -<span class="sourceLineNo">607</span> familyToCompression.put("Family2=asdads&!AASD", Compression.Algorithm.GZ);<a name="line.607"></a> -<span class="sourceLineNo">608</span> }<a name="line.608"></a> -<span class="sourceLineNo">609</span> if (numCfs-- > 0) {<a name="line.609"></a> -<span class="sourceLineNo">610</span> familyToCompression.put("Family3", Compression.Algorithm.NONE);<a name="line.610"></a> -<span class="sourceLineNo">611</span> }<a name="line.611"></a> -<span class="sourceLineNo">612</span> return familyToCompression;<a name="line.612"></a> -<span class="sourceLineNo">613</span> }<a name="line.613"></a> -<span class="sourceLineNo">614</span><a name="line.614"></a> +<span class="sourceLineNo">352</span> conf.set(HConstants.TEMPORARY_FS_DIRECTORY_KEY, util.getDataTestDir("testJobConfiguration")<a name="line.352"></a> +<span class="sourceLineNo">353</span> .toString());<a name="line.353"></a> +<span class="sourceLineNo">354</span> Job job = new Job(conf);<a name="line.354"></a> +<span class="sourceLineNo">355</span> job.setWorkingDirectory(util.getDataTestDir("testJobConfiguration"));<a name="line.355"></a> +<span class="sourceLineNo">356</span> Table table = Mockito.mock(Table.class);<a name="line.356"></a> +<span class="sourceLineNo">357</span> RegionLocator regionLocator = Mockito.mock(RegionLocator.class);<a name="line.357"></a> +<span class="sourceLineNo">358</span> setupMockStartKeys(regionLocator);<a name="line.358"></a> +<span class="sourceLineNo">359</span> setupMockTableName(regionLocator);<a name="line.359"></a> +<span class="sourceLineNo">360</span> HFileOutputFormat2.configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.360"></a> +<span class="sourceLineNo">361</span> assertEquals(job.getNumReduceTasks(), 4);<a name="line.361"></a> +<span class="sourceLineNo">362</span> }<a name="line.362"></a> +<span class="sourceLineNo">363</span><a name="line.363"></a> +<span class="sourceLineNo">364</span> private byte [][] generateRandomStartKeys(int numKeys) {<a name="line.364"></a> +<span class="sourceLineNo">365</span> Random random = new Random();<a name="line.365"></a> +<span class="sourceLineNo">366</span> byte[][] ret = new byte[numKeys][];<a name="line.366"></a> +<span class="sourceLineNo">367</span> // first region start key is always empty<a name="line.367"></a> +<span class="sourceLineNo">368</span> ret[0] = HConstants.EMPTY_BYTE_ARRAY;<a name="line.368"></a> +<span class="sourceLineNo">369</span> for (int i = 1; i < numKeys; i++) {<a name="line.369"></a> +<span class="sourceLineNo">370</span> ret[i] =<a name="line.370"></a> +<span class="sourceLineNo">371</span> PerformanceEvaluation.generateData(random, PerformanceEvaluation.DEFAULT_VALUE_LENGTH);<a name="line.371"></a> +<span class="sourceLineNo">372</span> }<a name="line.372"></a> +<span class="sourceLineNo">373</span> return ret;<a name="line.373"></a> +<span class="sourceLineNo">374</span> }<a name="line.374"></a> +<span class="sourceLineNo">375</span><a name="line.375"></a> +<span class="sourceLineNo">376</span> private byte[][] generateRandomSplitKeys(int numKeys) {<a name="line.376"></a> +<span class="sourceLineNo">377</span> Random random = new Random();<a name="line.377"></a> +<span class="sourceLineNo">378</span> byte[][] ret = new byte[numKeys][];<a name="line.378"></a> +<span class="sourceLineNo">379</span> for (int i = 0; i < numKeys; i++) {<a name="line.379"></a> +<span class="sourceLineNo">380</span> ret[i] =<a name="line.380"></a> +<span class="sourceLineNo">381</span> PerformanceEvaluation.generateData(random, PerformanceEvaluation.DEFAULT_VALUE_LENGTH);<a name="line.381"></a> +<span class="sourceLineNo">382</span> }<a name="line.382"></a> +<span class="sourceLineNo">383</span> return ret;<a name="line.383"></a> +<span class="sourceLineNo">384</span> }<a name="line.384"></a> +<span class="sourceLineNo">385</span><a name="line.385"></a> +<span class="sourceLineNo">386</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.386"></a> +<span class="sourceLineNo">387</span> public void testMRIncrementalLoad() throws Exception {<a name="line.387"></a> +<span class="sourceLineNo">388</span> LOG.info("\nStarting test testMRIncrementalLoad\n");<a name="line.388"></a> +<span class="sourceLineNo">389</span> doIncrementalLoadTest(false, false);<a name="line.389"></a> +<span class="sourceLineNo">390</span> }<a name="line.390"></a> +<span class="sourceLineNo">391</span><a name="line.391"></a> +<span class="sourceLineNo">392</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.392"></a> +<span class="sourceLineNo">393</span> public void testMRIncrementalLoadWithSplit() throws Exception {<a name="line.393"></a> +<span class="sourceLineNo">394</span> LOG.info("\nStarting test testMRIncrementalLoadWithSplit\n");<a name="line.394"></a> +<span class="sourceLineNo">395</span> doIncrementalLoadTest(true, false);<a name="line.395"></a> +<span class="sourceLineNo">396</span> }<a name="line.396"></a> +<span class="sourceLineNo">397</span><a name="line.397"></a> +<span class="sourceLineNo">398</span> /**<a name="line.398"></a> +<span class="sourceLineNo">399</span> * Test for HFileOutputFormat2.LOCALITY_SENSITIVE_CONF_KEY = true<a name="line.399"></a> +<span class="sourceLineNo">400</span> * This test could only check the correctness of original logic if LOCALITY_SENSITIVE_CONF_KEY<a name="line.400"></a> +<span class="sourceLineNo">401</span> * is set to true. Because MiniHBaseCluster always run with single hostname (and different ports),<a name="line.401"></a> +<span class="sourceLineNo">402</span> * it's not possible to check the region locality by comparing region locations and DN hostnames.<a name="line.402"></a> +<span class="sourceLineNo">403</span> * When MiniHBaseCluster supports explicit hostnames parameter (just like MiniDFSCluster does),<a name="line.403"></a> +<span class="sourceLineNo">404</span> * we could test region locality features more easily.<a name="line.404"></a> +<span class="sourceLineNo">405</span> */<a name="line.405"></a> +<span class="sourceLineNo">406</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.406"></a> +<span class="sourceLineNo">407</span> public void testMRIncrementalLoadWithLocality() throws Exception {<a name="line.407"></a> +<span class="sourceLineNo">408</span> LOG.info("\nStarting test testMRIncrementalLoadWithLocality\n");<a name="line.408"></a> +<span class="sourceLineNo">409</span> doIncrementalLoadTest(false, true);<a name="line.409"></a> +<span class="sourceLineNo">410</span> doIncrementalLoadTest(true, true);<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> private void doIncrementalLoadTest(boolean shouldChangeRegions, boolean shouldKeepLocality)<a name="line.413"></a> +<span class="sourceLineNo">414</span> throws Exception {<a name="line.414"></a> +<span class="sourceLineNo">415</span> util = new HBaseTestingUtility();<a name="line.415"></a> +<span class="sourceLineNo">416</span> Configuration conf = util.getConfiguration();<a name="line.416"></a> +<span class="sourceLineNo">417</span> conf.setBoolean(HFileOutputFormat2.LOCALITY_SENSITIVE_CONF_KEY, shouldKeepLocality);<a name="line.417"></a> +<span class="sourceLineNo">418</span> int hostCount = 1;<a name="line.418"></a> +<span class="sourceLineNo">419</span> int regionNum = 5;<a name="line.419"></a> +<span class="sourceLineNo">420</span> if(shouldKeepLocality) {<a name="line.420"></a> +<span class="sourceLineNo">421</span> // We should change host count higher than hdfs replica count when MiniHBaseCluster supports<a name="line.421"></a> +<span class="sourceLineNo">422</span> // explicit hostnames parameter just like MiniDFSCluster does.<a name="line.422"></a> +<span class="sourceLineNo">423</span> hostCount = 3;<a name="line.423"></a> +<span class="sourceLineNo">424</span> regionNum = 20;<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> byte[][] splitKeys = generateRandomSplitKeys(regionNum - 1);<a name="line.427"></a> +<span class="sourceLineNo">428</span> String[] hostnames = new String[hostCount];<a name="line.428"></a> +<span class="sourceLineNo">429</span> for(int i = 0; i < hostCount; ++i) {<a name="line.429"></a> +<span class="sourceLineNo">430</span> hostnames[i] = "datanode_" + i;<a name="line.430"></a> +<span class="sourceLineNo">431</span> }<a name="line.431"></a> +<span class="sourceLineNo">432</span> util.startMiniCluster(1, hostCount, hostnames);<a name="line.432"></a> +<span class="sourceLineNo">433</span><a name="line.433"></a> +<span class="sourceLineNo">434</span> Table table = util.createTable(TABLE_NAME, FAMILIES, splitKeys);<a name="line.434"></a> +<span class="sourceLineNo">435</span> Path testDir = util.getDataTestDirOnTestFS("testLocalMRIncrementalLoad");<a name="line.435"></a> +<span class="sourceLineNo">436</span> try (RegionLocator r = util.getConnection().getRegionLocator(TABLE_NAME); Admin admin =<a name="line.436"></a> +<span class="sourceLineNo">437</span> util.getConnection().getAdmin();) {<a name="line.437"></a> +<span class="sourceLineNo">438</span> assertEquals("Should start with empty table", 0, util.countRows(table));<a name="line.438"></a> +<span class="sourceLineNo">439</span> int numRegions = r.getStartKeys().length;<a name="line.439"></a> +<span class="sourceLineNo">440</span> assertEquals("Should make " + regionNum + " regions", numRegions, regionNum);<a name="line.440"></a> +<span class="sourceLineNo">441</span><a name="line.441"></a> +<span class="sourceLineNo">442</span> // Generate the bulk load files<a name="line.442"></a> +<span class="sourceLineNo">443</span> runIncrementalPELoad(conf, table.getTableDescriptor(), r, testDir);<a name="line.443"></a> +<span class="sourceLineNo">444</span> // This doesn't write into the table, just makes files<a name="line.444"></a> +<span class="sourceLineNo">445</span> assertEquals("HFOF should not touch actual table", 0, util.countRows(table));<a name="line.445"></a> +<span class="sourceLineNo">446</span><a name="line.446"></a> +<span class="sourceLineNo">447</span> // Make sure that a directory was created for every CF<a name="line.447"></a> +<span class="sourceLineNo">448</span> int dir = 0;<a name="line.448"></a> +<span class="sourceLineNo">449</span> for (FileStatus f : testDir.getFileSystem(conf).listStatus(testDir)) {<a name="line.449"></a> +<span class="sourceLineNo">450</span> for (byte[] family : FAMILIES) {<a name="line.450"></a> +<span class="sourceLineNo">451</span> if (Bytes.toString(family).equals(f.getPath().getName())) {<a name="line.451"></a> +<span class="sourceLineNo">452</span> ++dir;<a name="line.452"></a> +<span class="sourceLineNo">453</span> }<a name="line.453"></a> +<span class="sourceLineNo">454</span> }<a name="line.454"></a> +<span class="sourceLineNo">455</span> }<a name="line.455"></a> +<span class="sourceLineNo">456</span> assertEquals("Column family not found in FS.", FAMILIES.length, dir);<a name="line.456"></a> +<span class="sourceLineNo">457</span><a name="line.457"></a> +<span class="sourceLineNo">458</span> // handle the split case<a name="line.458"></a> +<span class="sourceLineNo">459</span> if (shouldChangeRegions) {<a name="line.459"></a> +<span class="sourceLineNo">460</span> LOG.info("Changing regions in table");<a name="line.460"></a> +<span class="sourceLineNo">461</span> admin.disableTable(table.getName());<a name="line.461"></a> +<span class="sourceLineNo">462</span> while(util.getMiniHBaseCluster().getMaster().getAssignmentManager().<a name="line.462"></a> +<span class="sourceLineNo">463</span> getRegionStates().isRegionsInTransition()) {<a name="line.463"></a> +<span class="sourceLineNo">464</span> Threads.sleep(200);<a name="line.464"></a> +<span class="sourceLineNo">465</span> LOG.info("Waiting on table to finish disabling");<a name="line.465"></a> +<span class="sourceLineNo">466</span> }<a name="line.466"></a> +<span class="sourceLineNo">467</span> util.deleteTable(table.getName());<a name="line.467"></a> +<span class="sourceLineNo">468</span> byte[][] newSplitKeys = generateRandomSplitKeys(14);<a name="line.468"></a> +<span class="sourceLineNo">469</span> table = util.createTable(TABLE_NAME, FAMILIES, newSplitKeys);<a name="line.469"></a> +<span class="sourceLineNo">470</span><a name="line.470"></a> +<span class="sourceLineNo">471</span> while (util.getConnection().getRegionLocator(TABLE_NAME)<a name="line.471"></a> +<span class="sourceLineNo">472</span> .getAllRegionLocations().size() != 15 ||<a name="line.472"></a> +<span class="sourceLineNo">473</span> !admin.isTableAvailable(table.getName())) {<a name="line.473"></a> +<span class="sourceLineNo">474</span> Thread.sleep(200);<a name="line.474"></a> +<span class="sourceLineNo">475</span> LOG.info("Waiting for new region assignment to happen");<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><a name="line.478"></a> +<span class="sourceLineNo">479</span> // Perform the actual load<a name="line.479"></a> +<span class="sourceLineNo">480</span> new LoadIncrementalHFiles(conf).doBulkLoad(testDir, admin, table, r);<a name="line.480"></a> +<span class="sourceLineNo">481</span><a name="line.481"></a> +<span class="sourceLineNo">482</span> // Ensure data shows up<a name="line.482"></a> +<span class="sourceLineNo">483</span> int expectedRows = NMapInputFormat.getNumMapTasks(conf) * ROWSPERSPLIT;<a name="line.483"></a> +<span class="sourceLineNo">484</span> assertEquals("LoadIncrementalHFiles should put expected data in table",<a name="line.484"></a> +<span class="sourceLineNo">485</span> expectedRows, util.countRows(table));<a name="line.485"></a> +<span class="sourceLineNo">486</span> Scan scan = new Scan();<a name="line.486"></a> +<span class="sourceLineNo">487</span> ResultScanner results = table.getScanner(scan);<a name="line.487"></a> +<span class="sourceLineNo">488</span> for (Result res : results) {<a name="line.488"></a> +<span class="sourceLineNo">489</span> assertEquals(FAMILIES.length, res.rawCells().length);<a name="line.489"></a> +<span class="sourceLineNo">490</span> Cell first = res.rawCells()[0];<a name="line.490"></a> +<span class="sourceLineNo">491</span> for (Cell kv : res.rawCells()) {<a name="line.491"></a> +<span class="sourceLineNo">492</span> assertTrue(CellUtil.matchingRow(first, kv));<a name="line.492"></a> +<span class="sourceLineNo">493</span> assertTrue(Bytes.equals(CellUtil.cloneValue(first), CellUtil.cloneValue(kv)));<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> results.close();<a name="line.496"></a> +<span class="sourceLineNo">497</span> String tableDigestBefore = util.checksumRows(table);<a name="line.497"></a> +<span class="sourceLineNo">498</span><a name="line.498"></a> +<span class="sourceLineNo">499</span> // Check region locality<a name="line.499"></a> +<span class="sourceLineNo">500</span> HDFSBlocksDistribution hbd = new HDFSBlocksDistribution();<a name="line.500"></a> +<span class="sourceLineNo">501</span> for (HRegion region : util.getHBaseCluster().getRegions(TABLE_NAME)) {<a name="line.501"></a> +<span class="sourceLineNo">502</span> hbd.add(region.getHDFSBlocksDistribution());<a name="line.502"></a> +<span class="sourceLineNo">503</span> }<a name="line.503"></a> +<span class="sourceLineNo">504</span> for (String hostname : hostnames) {<a name="line.504"></a> +<span class="sourceLineNo">505</span> float locality = hbd.getBlockLocalityIndex(hostname);<a name="line.505"></a> +<span class="sourceLineNo">506</span> LOG.info("locality of [" + hostname + "]: " + locality);<a name="line.506"></a> +<span class="sourceLineNo">507</span> assertEquals(100, (int) (locality * 100));<a name="line.507"></a> +<span class="sourceLineNo">508</span> }<a name="line.508"></a> +<span class="sourceLineNo">509</span><a name="line.509"></a> +<span class="sourceLineNo">510</span> // Cause regions to reopen<a name="line.510"></a> +<span class="sourceLineNo">511</span> admin.disableTable(TABLE_NAME);<a name="line.511"></a> +<span class="sourceLineNo">512</span> while (!admin.isTableDisabled(TABLE_NAME)) {<a name="line.512"></a> +<span class="sourceLineNo">513</span> Thread.sleep(200);<a name="line.513"></a> +<span class="sourceLineNo">514</span> LOG.info("Waiting for table to disable");<a name="line.514"></a> +<span class="sourceLineNo">515</span> }<a name="line.515"></a> +<span class="sourceLineNo">516</span> admin.enableTable(TABLE_NAME);<a name="line.516"></a> +<span class="sourceLineNo">517</span> util.waitTableAvailable(TABLE_NAME);<a name="line.517"></a> +<span class="sourceLineNo">518</span> assertEquals("Data should remain after reopening of regions",<a name="line.518"></a> +<span class="sourceLineNo">519</span> tableDigestBefore, util.checksumRows(table));<a name="line.519"></a> +<span class="sourceLineNo">520</span> } finally {<a name="line.520"></a> +<span class="sourceLineNo">521</span> testDir.getFileSystem(conf).delete(testDir, true);<a name="line.521"></a> +<span class="sourceLineNo">522</span> util.deleteTable(TABLE_NAME);<a name="line.522"></a> +<span class="sourceLineNo">523</span> util.shutdownMiniCluster();<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> private void runIncrementalPELoad(Configuration conf, HTableDescriptor tableDescriptor,<a name="line.527"></a> +<span class="sourceLineNo">528</span> RegionLocator regionLocator, Path outDir) throws IOException, UnsupportedEncodingException,<a name="line.528"></a> +<span class="sourceLineNo">529</span> InterruptedException, ClassNotFoundException {<a name="line.529"></a> +<span class="sourceLineNo">530</span> Job job = new Job(conf, "testLocalMRIncrementalLoad");<a name="line.530"></a> +<span class="sourceLineNo">531</span> job.setWorkingDirectory(util.getDataTestDirOnTestFS("runIncrementalPELoad"));<a name="line.531"></a> +<span class="sourceLineNo">532</span> job.getConfiguration().setStrings("io.serializations", conf.get("io.serializations"),<a name="line.532"></a> +<span class="sourceLineNo">533</span> MutationSerialization.class.getName(), ResultSerialization.class.getName(),<a name="line.533"></a> +<span class="sourceLineNo">534</span> KeyValueSerialization.class.getName());<a name="line.534"></a> +<span class="sourceLineNo">535</span> setupRandomGeneratorMapper(job);<a name="line.535"></a> +<span class="sourceLineNo">536</span> HFileOutputFormat2.configureIncrementalLoad(job, tableDescriptor, regionLocator);<a name="line.536"></a> +<span class="sourceLineNo">537</span> FileOutputFormat.setOutputPath(job, outDir);<a name="line.537"></a> +<span class="sourceLineNo">538</span><a name="line.538"></a> +<span class="sourceLineNo">539</span> assertFalse(util.getTestFileSystem().exists(outDir)) ;<a name="line.539"></a> +<span class="sourceLineNo">540</span><a name="line.540"></a> +<span class="sourceLineNo">541</span> assertEquals(regionLocator.getAllRegionLocations().size(), job.getNumReduceTasks());<a name="line.541"></a> +<span class="sourceLineNo">542</span><a name="line.542"></a> +<span class="sourceLineNo">543</span> assertTrue(job.waitForCompletion(true));<a name="line.543"></a> +<span class="sourceLineNo">544</span> }<a name="line.544"></a> +<span class="sourceLineNo">545</span><a name="line.545"></a> +<span class="sourceLineNo">546</span> /**<a name="line.546"></a> +<span class="sourceLineNo">547</span> * Test for {@link HFileOutputFormat2#configureCompression(org.apache.hadoop.hbase.client.Table,<a name="line.547"></a> +<span class="sourceLineNo">548</span> * Configuration)} and {@link HFileOutputFormat2#createFamilyCompressionMap<a name="line.548"></a> +<span class="sourceLineNo">549</span> * (Configuration)}.<a name="line.549"></a> +<span class="sourceLineNo">550</span> * Tests that the compression map is correctly serialized into<a name="line.550"></a> +<span class="sourceLineNo">551</span> * and deserialized from configuration<a name="line.551"></a> +<span class="sourceLineNo">552</span> *<a name="line.552"></a> +<span class="sourceLineNo">553</span> * @throws IOException<a name="line.553"></a> +<span class="sourceLineNo">554</span> */<a name="line.554"></a> +<span class="sourceLineNo">555</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.555"></a> +<span class="sourceLineNo">556</span> public void testSerializeDeserializeFamilyCompressionMap() throws IOException {<a name="line.556"></a> +<span class="sourceLineNo">557</span> for (int numCfs = 0; numCfs <= 3; numCfs++) {<a name="line.557"></a> +<span class="sourceLineNo">558</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.558"></a> +<span class="sourceLineNo">559</span> Map<String, Compression.Algorithm> familyToCompression =<a name="line.559"></a> +<span class="sourceLineNo">560</span> getMockColumnFamiliesForCompression(numCfs);<a name="line.560"></a> +<span class="sourceLineNo">561</span> Table table = Mockito.mock(HTable.class);<a name="line.561"></a> +<span class="sourceLineNo">562</span> setupMockColumnFamiliesForCompression(table, familyToCompression);<a name="line.562"></a> +<span class="sourceLineNo">563</span> HFileOutputFormat2.configureCompression(conf, table.getTableDescriptor());<a name="line.563"></a> +<span class="sourceLineNo">564</span><a name="line.564"></a> +<span class="sourceLineNo">565</span> // read back family specific compression setting from the configuration<a name="line.565"></a> +<span class="sourceLineNo">566</span> Map<byte[], Algorithm> retrievedFamilyToCompressionMap = HFileOutputFormat2<a name="line.566"></a> +<span class="sourceLineNo">567</span> .createFamilyCompressionMap(conf);<a name="line.567"></a> +<span class="sourceLineNo">568</span><a name="line.568"></a> +<span class="sourceLineNo">569</span> // test that we have a value for all column families that matches with the<a name="line.569"></a> +<span class="sourceLineNo">570</span> // used mock values<a name="line.570"></a> +<span class="sourceLineNo">571</span> for (Entry<String, Algorithm> entry : familyToCompression.entrySet()) {<a name="line.571"></a> +<span class="sourceLineNo">572</span> assertEquals("Compression configuration incorrect for column family:"<a name="line.572"></a> +<span class="sourceLineNo">573</span> + entry.getKey(), entry.getValue(),<a name="line.573"></a> +<span class="sourceLineNo">574</span> retrievedFamilyToCompressionMap.get(entry.getKey().getBytes()));<a name="line.574"></a> +<span class="sourceLineNo">575</span> }<a name="line.575"></a> +<span class="sourceLineNo">576</span> }<a name="line.576"></a> +<span class="sourceLineNo">577</span> }<a name="line.577"></a> +<span class="sourceLineNo">578</span><a name="line.578"></a> +<span class="sourceLineNo">579</span> private void setupMockColumnFamiliesForCompression(Table table,<a name="line.579"></a> +<span class="sourceLineNo">580</span> Map<String, Compression.Algorithm> familyToCompression) throws IOException {<a name="line.580"></a> +<span class="sourceLineNo">581</span> HTableDescriptor mockTableDescriptor = new HTableDescriptor(TABLE_NAME);<a name="line.581"></a> +<span class="sourceLineNo">582</span> for (Entry<String, Compression.Algorithm> entry : familyToCompression.entrySet()) {<a name="line.582"></a> +<span class="sourceLineNo">583</span> mockTableDescriptor.addFamily(new HColumnDescriptor(entry.getKey())<a name="line.583"></a> +<span class="sourceLineNo">584</span> .setMaxVersions(1)<a name="line.584"></a> +<span class="sourceLineNo">585</span> .setCompressionType(entry.getValue())<a name="line.585"></a> +<span class="sourceLineNo">586</span> .setBlockCacheEnabled(false)<a name="line.586"></a> +<span class="sourceLineNo">587</span> .setTimeToLive(0));<a name="line.587"></a> +<span class="sourceLineNo">588</span> }<a name="line.588"></a> +<span class="sourceLineNo">589</span> Mockito.doReturn(mockTableDescriptor).when(table).getTableDescriptor();<a name="line.589"></a> +<span class="sourceLineNo">590</span> }<a name="line.590"></a> +<span class="sourceLineNo">591</span><a name="line.591"></a> +<span class="sourceLineNo">592</span> /**<a name="line.592"></a> +<span class="sourceLineNo">593</span> * @return a map from column family names to compression algorithms for<a name="line.593"></a> +<span class="sourceLineNo">594</span> * testing column family compression. Column family names have special characters<a name="line.594"></a> +<span class="sourceLineNo">595</span> */<a name="line.595"></a> +<span class="sourceLineNo">596</span> private Map<String, Compression.Algorithm><a name="line.596"></a> +<span class="sourceLineNo">597</span> getMockColumnFamiliesForCompression (int numCfs) {<a name="line.597"></a> +<span class="sourceLineNo">598</span> Map<String, Compression.Algorithm> familyToCompression<a name="line.598"></a> +<span class="sourceLineNo">599</span> = new HashMap<String, Compression.Algorithm>();<a name="line.599"></a> +<span class="sourceLineNo">600</span> // use column family names having special characters<a name="line.600"></a> +<span class="sourceLineNo">601</span> if (numCfs-- > 0) {<a name="line.601"></a> +<span class="sourceLineNo">602</span> familyToCompression.put("Family1!@#!@#&", Compression.Algorithm.LZO);<a name="line.602"></a> +<span class="sourceLineNo">603</span> }<a name="line.603"></a> +<span class="sourceLineNo">604</span> if (numCfs-- > 0) {<a name="line.604"></a> +<span class="sourceLineNo">605</span> familyToCompression.put("Family2=asdads&!AASD", Compression.Algorithm.SNAPPY);<a name="line.605"></a> +<span class="sourceLineNo">606</span> }<a name="line.606"></a> +<span class="sourceLineNo">607</span> if (numCfs-- > 0) {<a name="line.607"></a> +<span class="sourceLineNo">608</span> familyToCompression.put("Family2=asdads&!AASD", Compression.Algorithm.GZ);<a name="line.608"></a> +<span class="sourceLineNo">609</span> }<a name="line.609"></a> +<span class="sourceLineNo">610</span> if (numCfs-- > 0) {<a name="line.610"></a> +<span class="sourceLineNo">611</span> familyToCompression.put("Family3", Compression.Algorithm.NONE);<a name="line.611"></a> +<span class="sourceLineNo">612</span> }<a name="line.612"></a> +<span class="sourceLineNo">613</span> return familyToCompression;<a name="line.613"></a> +<span class="sourceLineNo">614</span> }<a name="line.614"></a> <span class="sourceLineNo">615</span><a name="line.615"></a> -<span class="sourceLineNo">616</span> /**<a name="line.616"></a> -<span class="sourceLineNo">617</span> * Test for {@link HFileOutputFormat2#configureBloomType(org.apache.hadoop.hbase.client.Table,<a name="line.617"></a> -<span class="sourceLineNo">618</span> * Configuration)} and {@link HFileOutputFormat2#createFamilyBloomTypeMap<a name="line.618"></a> -<span class="sourceLineNo">619</span> * (Configuration)}.<a name="line.619"></a> -<span class="sourceLineNo">620</span> * Tests that the compression map is correctly serialized into<a name="line.620"></a> -<span class="sourceLineNo">621</span> * and deserialized from configuration<a name="line.621"></a> -<span class="sourceLineNo">622</span> *<a name="line.622"></a> -<span class="sourceLineNo">623</span> * @throws IOException<a name="line.623"></a> -<span class="sourceLineNo">624</span> */<a name="line.624"></a> -<span class="sourceLineNo">625</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.625"></a> -<span class="sourceLineNo">626</span> public void testSerializeDeserializeFamilyBloomTypeMap() throws IOException {<a name="line.626"></a> -<span class="sourceLineNo">627</span> for (int numCfs = 0; numCfs <= 2; numCfs++) {<a name="line.627"></a> -<span class="sourceLineNo">628</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.628"></a> -<span class="sourceLineNo">629</span> Map<String, BloomType> familyToBloomType =<a name="line.629"></a> -<span class="sourceLineNo">630</span> getMockColumnFamiliesForBloomType(numCfs);<a name="line.630"></a> -<span class="sourceLineNo">631</span> Table table = Mockito.mock(HTable.class);<a name="line.631"></a> -<span class="sourceLineNo">632</span> setupMockColumnFamiliesForBloomType(table,<a name="line.632"></a> -<span class="sourceLineNo">633</span> familyToBloomType);<a name="line.633"></a> -<span class="sourceLineNo">634</span> HFileOutputFormat2.configureBloomType(table.getTableDescriptor(), conf);<a name="line.634"></a> -<span class="sourceLineNo">635</span><a name="line.635"></a> -<span class="sourceLineNo">636</span> // read back family specific data block encoding settings from the<a name="line.636"></a> -<span class="sourceLineNo">637</span> // configuration<a name="line.637"></a> -<span class="sourceLineNo">638</span> Map<byte[], BloomType> retrievedFamilyToBloomTypeMap =<a name="line.638"></a> -<span class="sourceLineNo">639</span> HFileOutputFormat2<a name="line.639"></a> -<span class="sourceLineNo">640</span> .createFamilyBloomTypeMap(conf);<a name="line.640"></a> -<span class="sourceLineNo">641</span><a name="line.641"></a> -<span class="sourceLineNo">642</span> // test that we have a value for all column families that matches with the<a name="line.642"></a> -<span class="sourceLineNo">643</span> // used mock values<a name="line.643"></a> -<span class="sourceLineNo">644</span> for (Entry<String, BloomType> entry : familyToBloomType.entrySet()) {<a name="line.644"></a> -<span class="sourceLineNo">645</span> assertEquals("BloomType configuration incorrect for column family:"<a name="line.645"></a> -<span class="sourceLineNo">646</span> + entry.getKey(), entry.getValue(),<a name="line.646"></a> -<span class="sourceLineNo">647</span> retrievedFamilyToBloomTypeMap.get(entry.getKey().getBytes()));<a name="line.647"></a> -<span class="sourceLineNo">648</span> }<a name="line.648"></a> -<span class="sourceLineNo">649</span> }<a name="line.649"></a> -<span class="sourceLineNo">650</span> }<a name="line.650"></a> -<span class="sourceLineNo">651</span><a name="line.651"></a> -<span class="sourceLineNo">652</span> private void setupMockColumnFamiliesForBloomType(Table table,<a name="line.652"></a> -<span class="sourceLineNo">653</span> Map<String, BloomType> familyToDataBlockEncoding) throws IOException {<a name="line.653"></a> -<span class="sourceLineNo">654</span> HTableDescriptor mockTableDescriptor = new HTableDescriptor(TABLE_NAME);<a name="line.654"></a> -<span class="sourceLineNo">655</span> for (Entry<String, BloomType> entry : familyToDataBlockEncoding.entrySet()) {<a name="line.655"></a> -<span class="sourceLineNo">656</span> mockTableDescriptor.addFamily(new HColumnDescriptor(entry.getKey())<a name="line.656"></a> -<span class="sourceLineNo">657</span> .setMaxVersions(1)<a name="line.657"></a> -<span class="sourceLineNo">658</span> .setBloomFilterType(entry.getValue())<a name="line.658"></a> -<span class="sourceLineNo">659</span> .setBlockCacheEnabled(false)<a name="line.659"></a> -<span class="sourceLineNo">660</span> .setTimeToLive(0));<a name="line.660"></a> -<span class="sourceLineNo">661</span> }<a name="line.661"></a> -<span class="sourceLineNo">662</span> Mockito.doReturn(mockTableDescriptor).when(table).getTableDescriptor();<a name="line.662"></a> -<span class="sourceLineNo">663</span> }<a name="line.663"></a> -<span class="sourceLineNo">664</span><a name="line.664"></a> -<span class="sourceLineNo">665</span> /**<a name="line.665"></a> -<span class="sourceLineNo">666</span> * @return a map from column family names to compression algorithms for<a name="line.666"></a> -<span class="sourceLineNo">667</span> * testing column family compression. Column family names have special characters<a name="line.667"></a> -<span class="sourceLineNo">668</span> */<a name="line.668"></a> -<span class="sourceLineNo">669</span> private Map<String, BloomType><a name="line.669"></a> -<span class="sourceLineNo">670</span> getMockColumnFamiliesForBloomType (int numCfs) {<a name="line.670"></a> -<span class="sourceLineNo">671</span> Map<String, BloomType> familyToBloomType =<a name="line.671"></a> -<span class="sourceLineNo">672</span> new HashMap<String, BloomType>();<a name="line.672"></a> -<span class="sourceLineNo">673</span> // use column family names having special characters<a name="line.673"></a> -<span class="sourceLineNo">674</span> if (numCfs-- > 0) {<a name="line.674"></a> -<span class="sourceLineNo">675</span> familyToBloomType.put("Family1!@#!@#&", BloomType.ROW);<a name="line.675"></a> -<span class="sourceLineNo">676</span> }<a name="line.676"></a> -<span class="sourceLineNo">677</span> if (numCfs-- > 0) {<a name="line.677"></a> -<span class="sourceLineNo">678</span> familyToBloomType.put("Family2=asdads&!AASD",<a name="line.678"></a> -<span class="sourceLineNo">679</span> BloomType.ROWCOL);<a name="line.679"></a> -<span class="sourceLineNo">680</span> }<a name="line.680"></a> -<span class="sourceLineNo">681</span> if (numCfs-- > 0) {<a name="line.681"></a> -<span class="sourceLineNo">682</span> familyToBloomType.put("Family3", BloomType.NONE);<a name="line.682"></a> -<span class="sourceLineNo">683</span> }<a name="line.683"></a> -<span class="sourceLineNo">684</span> return familyToBloomType;<a name="line.684"></a> -<span class="sourceLineNo">685</span> }<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> * Test for {@link HFileOutputFormat2#configureBlockSize(org.apache.hadoop.hbase.client.Table,<a name="line.688"></a> -<span class="sourceLineNo">689</span> * Configuration)} and {@link HFileOutputFormat2#createFamilyBlockSizeMap<a name="line.689"></a> -<span class="sourceLineNo">690</span> * (Configuration)}.<a name="line.690"></a> -<span class="sourceLineNo">691</span> * Tests that the compression map is correctly serialized into<a name="line.691"></a> -<span class="sourceLineNo">692</span> * and deserialized from configuration<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> */<a name="line.695"></a> -<span class="sourceLineNo">696</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.696"></a> -<span class="sourceLineNo">697</span> public void testSerializeDeserializeFamilyBlockSizeMap() throws IOException {<a name="line.697"></a> -<span class="sourceLineNo">698</span> for (int numCfs = 0; numCfs <= 3; numCfs++) {<a name="line.698"></a> -<span class="sourceLineNo">699</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.699"></a> -<span class="sourceLineNo">700</span> Map<String, Integer> familyToBlockSize =<a name="line.700"></a> -<span class="sourceLineNo">701</span> getMockColumnFamiliesForBlockSize(numCfs);<a name="line.701"></a> -<span class="sourceLineNo">702</span> Table table = Mockito.mock(HTable.class);<a name="line.702"></a> -<span class="sourceLineNo">703</span> setupMockColumnFamiliesForBlockSize(table,<a name="line.703"></a> -<span class="sourceLineNo">704</span> familyToBlockSize);<a name="line.704"></a> -<span class="sourceLineNo">705</span> HFileOutputFormat2.configureBlockSize(table.getTableDescriptor(), conf);<a name="line.705"></a> -<span class="sourceLineNo">706</span><a name="line.706"></a> -<span class="sourceLineNo">707</span> // read back family specific data block encoding settings from the<a name="line.707"></a> -<span class="sourceLineNo">708</span> // configuration<a name="line.708"></a> -<span class="sourceLineNo">709</span> Map<byte[], Integer> retrievedFamilyToBlockSizeMap =<a name="line.709"></a> -<span class="sourceLineNo">710</span> HFileOutputFormat2<a name="line.710"></a> -<span class="sourceLineNo">711</span> .createFamilyBlockSizeMap(conf);<a name="line.711"></a> -<span class="sourceLineNo">712</span><a name="line.712"></a> -<span class="sourceLineNo">713</span> // test that we have a value for all column families that matches with the<a name="line.713"></a> -<span class="sourceLineNo">714</span> // used mock values<a name="line.714"></a> -<span class="sourceLineNo">715</span> for (Entry<String, Integer> entry : familyToBlockSize.entrySet()<a name="line.715"></a> -<span class="sourceLineNo">716</span> ) {<a name="line.716"></a> -<span class="sourceLineNo">717</span> assertEquals("BlockSize configuration incorrect for column family:"<a name="line.717"></a> -<span class="sourceLineNo">718</span> + entry.getKey(), entry.getValue(),<a name="line.718"></a> -<span class="sourceLineNo">719</span> retrievedFamilyToBlockSizeMap.get(entry.getKey().getBytes()));<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> }<a name="line.722"></a> -<span class="sourceLineNo">723</span><a name="line.723"></a> -<span class="sourceLineNo">724</span> private void setupMockColumnFamiliesForBlockSize(Table table,<a name="line.724"></a> -<span class="sourceLineNo">725</span> Map<String, Integer> familyToDataBlockEncoding) throws IOException {<a name="line.725"></a> -<span class="sourceLineNo">726</span> HTableDescriptor mockTableDescriptor = new HTableDescriptor(TABLE_NAME);<a name="line.726"></a> -<span class="sourceLineNo">727</span> for (Entry<String, Integer> entry : familyToDataBlockEncoding.entrySet()) {<a name="line.727"></a> -<span class="sourceLineNo">728</span> mockTableDescriptor.addFamily(new HColumnDescriptor(entry.getKey())<a name="line.728"></a> -<span class="sourceLineNo">729</span> .setMaxVersions(1)<a name="line.729"></a> -<span class="sourceLineNo">730</span> .setBlocksize(entry.getValue())<a name="line.730"></a> -<span class="sourceLineNo">731</span> .setBlockCacheEnabled(false)<a name="line.731"></a> -<span class="sourceLineNo">732</span> .setTimeToLive(0));<a name="line.732"></a> -<span class="sourceLineNo">733</span> }<a name="line.733"></a> -<span class="sourceLineNo">734</span> Mockito.doReturn(mockTableDescriptor).when(table).getTableDescriptor();<a name="line.734"></a> -<span class="sourceLineNo">735</span> }<a name="line.735"></a> -<span class="sourceLineNo">736</span><a name="line.736"></a> -<span class="sourceLineNo">737</span> /**<a name="line.737"></a> -<span class="sourceLineNo">738</span> * @return a map from column family names to compression algorithms for<a name="line.738"></a> -<span class="sourceLineNo">739</span> * testing column family compression. Column family names have special characters<a name="line.739"></a> -<span class="sourceLineNo">740</span> */<a name="line.740"></a> -<span class="sourceLineNo">741</span> private Map<String, Integer><a name="line.741"></a> -<span class="sourceLineNo">742</span> getMockColumnFamiliesForBlockSize (int numCfs) {<a name="line.742"></a> -<span class="sourceLineNo">743</span> Map<String, Integer> familyToBlockSize =<a name="line.743"></a> -<span class="sourceLineNo">744</span> new HashMap<String, Integer>();<a name="line.744"></a> -<span class="sourceLineNo">745</span> // use column family names having special characters<a name="line.745"></a> -<span class="sourceLineNo">746</span> if (numCfs-- > 0) {<a name="line.746"></a> -<span class="sourceLineNo">747</span> familyToBlockSize.put("Family1!@#!@#&", 1234);<a name="line.747"></a> -<span class="sourceLineNo">748</span> }<a name="line.748"></a> -<span class="sourceLineNo">749</span> if (numCfs-- > 0) {<a name="line.749"></a> -<span class="sourceLineNo">750</span> familyToBlockSize.put("Family2=asdads&!AASD",<a name="line.750"></a> -<span class="sourceLineNo">751</span> Integer.MAX_VALUE);<a name="line.751"></a> -<span class="sourceLineNo">752</span> }<a name="line.752"></a> -<span class="sourceLineNo">753</span> if (numCfs-- > 0) {<a name="line.753"></a> -<span class="sourceLineNo">754</span> familyToBlockSize.put("Family2=asdads&!AASD",<a name="line.754"></a> -<span class="sourceLineNo">755</span> Integer.MAX_VALUE);<a name="line.755"></a> -<span class="sourceLineNo">756</span> }<a name="line.756"></a> -<span class="sourceLineNo">757</span> if (numCfs-- > 0) {<a name="line.757"></a> -<span class="sourceLineNo">758</span> familyToBlockSize.put("Family3", 0);<a name="line.758"></a> -<span class="sourceLineNo">759</span> }<a name="line.759"></a> -<span class="sourceLineNo">760</span> return familyToBlockSize;<a name="line.760"></a> -<span class="sourceLineNo">761</span> }<a name="line.761"></a> -<span class="sourceLineNo">762</span><a name="line.762"></a> -<span class="sourceLineNo">763</span> /**<a name="line.763"></a> -<span class="sourceLineNo">764</span> * Test for {@link HFileOutputFormat2#configureDataBlockEncoding(HTableDescriptor, Configuration)}<a name="line.764"></a> -<span class="sourceLineNo">765</span> * and {@link HFileOutputFormat2#createFamilyDataBlockEncodingMap(Configuration)}.<a name="line.765"></a> -<span class="sourceLineNo">766</span> * Tests that the compression map is correctly serialized into<a name="line.766"></a> -<span class="sourceLineNo">767</span> * and deserialized from configuration<a name="line.767"></a> -<span class="sourceLineNo">768</span> *<a name="line.768"></a> -<span class="sourceLineNo">769</span> * @throws IOException<a name="line.769"></a> -<span class="sourceLineNo">770</span> */<a name="line.770"></a> -<span class="sourceLineNo">771</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.771"></a> -<span class="sourceLineNo">772</span> public void testSerializeDeserializeFamilyDataBlockEncodingMap() throws IOException {<a name="line.772"></a> -<span class="sourceLineNo">773</span> for (int numCfs = 0; numCfs <= 3; numCfs++) {<a name="line.773"></a> -<span class="sourceLineNo">774</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.774"></a> -<span class="sourceLineNo">775</span> Map<String, DataBlockEncoding> familyToDataBlockEncoding =<a name="line.775"></a> -<span class="sourceLineNo">776</span> getMockColumnFamiliesForDataBlockEncoding(numCfs);<a name="line.776"></a> -<span class="sourceLineNo">777</span> Table table = Mockito.mock(HTable.class);<a name="line.777"></a> -<span class="sourceLineNo">778</span> setupMockColumnFamiliesForDataBlockEncoding(table,<a name="line.778"></a> -<span class="sourceLineNo">779</span> familyToDataBlockEncoding);<a name="line.779"></a> -<span class="sourceLineNo">780</span> HTableDescriptor tableDescriptor = table.getTableDescriptor();<a name="line.780"></a> -<span class="sourceLineNo">781</span> HFileOutputFormat2.configureDataBlockEncoding(tableDescriptor, conf);<a name="line.781"></a> -<span class="sourceLineNo">782</span><a name="line.782"></a> -<span class="sourceLineNo">783</span> // read back family specific data block encoding settings from the<a name="line.783"></a> -<span class="sourceLineNo">784</span> // configuration<a name="line.784"></a> -<span class="sourceLineNo">785</span> Map<byte[], DataBlockEncoding> retrievedFamilyToDataBlockEncodingMap =<a name="line.785"></a> -<span class="sourceLineNo">786</span> HFileOutputFormat2<a name="line.786"></a> -<span class="sourceLineNo">787</span> .createFamilyDataBlockEncodingMap(conf);<a name="line.787"></a> -<span class="sourceLineNo">788</span><a name="line.788"></a> -<span class="sourceLineNo">789</span> // test that we have a value for all column families that matches with the<a name="line.789"></a> -<span class="sourceLineNo">790</span> // used mock values<a name="line.790"></a> -<span class="sourceLineNo">791</span> for (Entry<String, DataBlockEncoding> entry : familyToDataBlockEncoding.entrySet()) {<a name="line.791"></a> -<span class="sourceLineNo">792</span> assertEquals("DataBlockEncoding configuration incorrect for column family:"<a name="line.792"></a> -<span class="sourceLineNo">793</span> + entry.getKey(), entry.getValue(),<a name="line.793"></a> -<span class="sourceLineNo">794</span> retrievedFamilyToDataBlockEncodingMap.get(entry.getKey().getBytes()));<a name="line.794"></a> -<span class="sourceLineNo">795</span> }<a name="line.795"></a> -<span class="sourceLineNo">796</span> }<a name="line.796"></a> -<span class="sourceLineNo">797</span> }<a name="line.797"></a> -<span class="sourceLineNo">798</span><a name="line.798"></a> -<span class="sourceLineNo">799</span> private void setupMockColumnFamiliesForDataBlockEncoding(Table table,<a name="line.799"></a> -<span class="sourceLineNo">800</span> Map<String, DataBlockEncoding> familyToDataBlockEncoding) throws IOException {<a name="line.800"></a> -<span class="sourceLineNo">801</span> HTableDescriptor mockTableDescriptor = new HTableDescriptor(TABLE_NAME);<a name="line.801"></a> -<span class="sourceLineNo">802</span> for (Entry<String, DataBlockEncoding> entry : familyToDataBlockEncoding.entrySet()) {<a name="line.802"></a> -<span class="sourceLineNo">803</span> mockTableDescriptor.addFamily(new HColumnDescriptor(entry.getKey())<a name="line.803"></a> -<span class="sourceLineNo">804</span> .setMaxVersions(1)<a name="line.804"></a> -<span class="sourceLineNo">805</span> .setDataBlockEncoding(entry.getValue())<a name="line.805"></a> -<span class="sourceLineNo">806</span> .setBlockCacheEnabled(false)<a name="line.806"></a> -<span class="sourceLineNo">807</span> .setTimeToLive(0));<a name="line.807"></a> -<span class="sourceLineNo">808</span> }<a name="line.808"></a> -<span class="sourceLineNo">809</span> Mockito.doReturn(mockTableDescriptor).when(table).getTableDescriptor();<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> /**<a name="line.812"></a> -<span class="sourceLineNo">813</span> * @return a map from column family names to compression algorithms for<a name="line.813"></a> -<span class="sourceLineNo">814</span> * testing column family compression. Column family names have special characters<a name="line.814"></a> -<span class="sourceLineNo">815</span> */<a name="line.815"></a> -<span class="sourceLineNo">816</span> private Map<String, DataBlockEncoding><a name="line.816"></a> -<span class="sourceLineNo">817</span> getMockColumnFamiliesForDataBlockEncoding (int numCfs) {<a name="line.817"></a> -<span class="sourceLineNo">818</span> Map<String, DataBlockEncoding> familyToDataBlockEncoding =<a name="line.818"></a> -<span class="sourceLineNo">819</span> new HashMap<String, DataBlockEncoding>();<a name="line.819"></a> -<span class="sourceLineNo">820</span> // use column family names having special characters<a name="line.820"></a> -<span class="sourceLineNo">821</span> if (numCfs-- > 0) {<a name="line.821"></a> -<span class="sourceLineNo">822</span> familyToDataBlockEncoding.put("Family1!@#!@#&", DataBlockEncoding.DIFF);<a name="line.822"></a> -<span class="sourceLineNo">823</span> }<a name="line.823"></a> -<span class="sourceLineNo">824</span> if (numCfs-- > 0) {<a name="line.824"></a> -<span class="sourceLineNo">825</span> familyToDataBlockEncoding.put("Family2=asdads&!AASD",<a name="line.825"></a> -<span class="sourceLineNo">826</span> DataBlockEncoding.FAST_DIFF);<a name="line.826"></a> -<span class="sourceLineNo">827</span> }<a name="line.827"></a> -<span class="sourceLineNo">828</span> if (numCfs-- > 0) {<a name="line.828"></a> -<span class="sourceLineNo">829</span> familyToDataBlockEncoding.put("Family2=asdads&!AASD",<a name="line.829"></a> -<span class="sourceLineNo">830</span> DataBlockEncoding.PREFIX);<a name="line.830"></a> -<span class="sourceLineNo">831</span> }<a name="line.831"></a> -<span class="sourceLineNo">832</span> if (numCfs-- > 0) {<a name="line.832"></a> -<span class="sourceLineNo">833</span> familyToDataBlockEncoding.put("Family3", DataBlockEncoding.NONE);<a name="line.833"></a> -<span class="sourceLineNo">834</span> }<a name="line.834"></a> -<span class="sourceLineNo">835</span> return familyToDataBlockEncoding;<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> private void setupMockStartKeys(RegionLocator table) throws IOException {<a name="line.838"></a> -<span class="sourceLineNo">839</span> byte[][] mockKeys = new byte[][] {<a name="line.839"></a> -<span class="sourceLineNo">840</span> HConstants.EMPTY_BYTE_ARRAY,<a name="line.840"></a> -<span class="sourceLineNo">841</span> Bytes.toBytes("aaa"),<a name="line.841"></a> -<span class="sourceLineNo">842</span> Bytes.toBytes("ggg"),<a name="line.842"></a> -<span class="sourceLineNo">843</span> Bytes.toBytes("zzz")<a name="line.843"></a> -<span class="sourceLineNo">844</span> };<a name="line.844"></a> -<span class="sourceLineNo">845</span> Mockito.doReturn(mockKeys).when(table).getStartKeys();<a name="line.845"></a> -<span class="sourceLineNo">846</span> }<a name="line.846"></a> -<span class="sourceLineNo">847</span><a name="line.847"></a> -<span class="sourceLineNo">848</span> private void setupMockTableName(RegionLocator table) throws IOException {<a name="line.848"></a> -<span class="sourceLineNo">849</span> TableName mockTableName = TableName.valueOf("mock_table");<a name="line.849"></a> -<span class="sourceLineNo">850</span> Mockito.doReturn(mockTableName).when(table).getName();<a name="line.850"></a> -<span class="sourceLineNo">851</span> }<a name="line.851"></a> -<span class="sourceLineNo">852</span><a name="line.852"></a> -<span class="sourceLineNo">853</span> /**<a name="line.853"></a> -<span class="sourceLineNo">854</span> * Test that {@link HFileOutputFormat2} RecordWriter uses compression and<a name="line.854"></a> -<span class="sourceLineNo">855</span> * bloom filter settings from the column family descriptor<a name="line.855"></a> -<span class="sourceLineNo">856</span> */<a name="line.856"></a> -<span class="sourceLineNo">857</span> @Ignore("Goes zombie too frequently; needs work. See HBASE-14563") @Test<a name="line.857"></a> -<span class="sourceLineNo">858</span> public void testColumnFamilySettings() throws Exception {<a name="line.858"></a> -<span class="sourceLineNo">859</span> Configuration conf = new Configuration(this.util.getConfiguration());<a name="line.859"></a> -<span class="sourceLineNo">860</span> RecordWriter<ImmutableBytesWritable, Cell> writer = null;<a name="line.860"></a> -<span class="sourceLineNo">861</span> TaskAttemptContext context = null;<a name="line.861"></a> -<span class="sourceLineNo">862</span> Path dir = util.getDataTestDir("testColumnFamilySettings");<a name="line.862"></a> -<span class="sourceLineNo">863</span><a name="line.863"></a> -<span class="sourceLineNo">864</span> // Setup table descriptor<a name="line.864"></a> -<span class="sourceLineNo">865</span> Table table = Mockito.mock(Table.class);<a name="line.865"></a> -<span class="sourceLineNo">866</span> RegionLocator regionLocator = Mockito.mock(RegionLocator.class);<a name="line.866"></a> -<span class="sourceLineNo">867</span> HTableDescriptor htd = new HTableDescriptor(TABLE_NAME);<a name="line.867"></a> -<span class="sourceLineNo">868</span> Mockito.doReturn(htd).when(table).getTableDescriptor();<a name="line.868"></a> -<span class="sourceLineNo">869</span> for (HColumnDescriptor hcd: HBaseTestingUtility.generateColumnDescriptors()) {<a name="line.869"></a> -<span class="sourceLineNo">870</span> htd.addFamily(hcd);<a name="line.870"></a> -<span class="sourceLineNo">871</span> }<a name="line.871"></a> -<span class="sourceLineNo">872</span><a name="line.872"></a> -<span class="sourceLineNo">873</span> // set up the table to return some mock keys<a name="line.873"></a> -<span class="sourceLineNo">874</span> setupMockStartKeys(regionLocator);<a name="line.874"></a> -<span class="sourceLineNo">875</span><a name="line.875"></a> -<span class="sourceLineNo">876</span> try {<a name="line.876"></a> -<span class="sourceLineNo">877</span> // partial map red setup to get an operational writer for testing<a name="line.877"></a> -<span class="sourceLineNo">878</span> // We turn off the sequence file compression, because DefaultCodec<a name="line.878"></a> -<span class="sourceLineNo">879</span> // pollutes the GZip codec pool with an incompatible compressor.<a name="line.879"></a> -<span class="sourceLineNo">880</span> conf.set("io.seqfile.compression.type", "NONE");<a name="line.880"></a> -<span class="sourceLineNo">881</span> conf.set("hbase.fs.tmp.dir", dir.toString());<a name="line.881"></a> -<span class="sourceLineNo">882</span> // turn locality off to eliminate getRegionLocation fail-and-retry time when writing kvs<a name="line.882"></a> -<span class="sourceLineNo">883</span> conf.setBoolean(HFileOutputFormat2.LOCALITY_SENSITIVE_CONF_KEY, false);<a name="line.883"></a> -<span class="sourceLineNo">884</span><a name="line.884"></a> -<span class="sourceLineNo">885</span> Job job = new Job(conf, "testLocalMRIncrementalLoad");<a name="line.885"></a> -<span class="sourceLineNo">886</span> job.setWorkingDirectory(util.getDataTestDirOnTestFS("testColumnFamilySettings"));<a name="line.886"></a> -<span class="sourceLineNo">887</span> setupRandomGeneratorMapper(job);<a name="line.887"></a> -<span class="sourceLineNo">888</span> HFileOutputFormat2.configureIncrementalLoad(job, table.getTableDescriptor(), regionLocator);<a name="line.888"></a> -<span class="sourceLineNo">889</span> FileOutputFormat.setOutputPath(job, dir);<a name="line.889"></a> -<span class="sourceLineNo">890</span> context = createTestTaskAttemptContext(job);<a name="line.890"></a> -<span class="sourceLineNo">891</span> HFileOutputFormat2 hof = new HFileOutputFormat2();<a name="line.891"></a> -<span class="sourceLineNo">892</span> writer = hof.getRecordWriter(context);<a name="line.892"></a> -<span class="sourceLineNo">893</span><a name="line.893"></a> -<span class="sourceLineNo">894</span> // write out random rows<a name="line.894"></a> -<span class="sourceLineNo">895</span> writeRandomKeyValues(writer, context, htd.getFamiliesKeys(), ROWSPERSPLIT);<a name="line.895"></a> -<span class="sourceLineNo">896</span> writer.close(context);<a name="line.896"></a> -<span class="sourceLineNo">897</span><a name="line.897"></a> -<span class="sourceLineNo">898</span> // Make sure that a directory was created for every CF<a name="line.898"></a> -<span class="sourceLineNo">899</span> FileSystem fs = dir.getFileSystem(conf);<a name="line.899"></a> -<span class="sourceLineNo">900</span><a name="line.900"></a> -<span class="sourceLineNo">901</span> // commit so that the filesystem has one directory per column family<a name="line.901"></a> -<span class="sourceLineNo">902</span> hof.getOutputCommitter(context).commitTask(context);<a name="line.902"></a> -<span class="sourceLineNo">903</span> hof.getOutputCommitter(context).commitJob(context);<a name="line.903"></a> -<span class="sourceLineNo">904</span> FileStatus[] families = FSUtils.listStatus(fs, dir, new FSUtils.FamilyDirFilter(fs));<a name="line.904"></a> -<span class="sourceLineNo">905</span> assertEquals(htd.getFamilies().size(), families.length);<a name="line.905"></a> -<span class="sourceLineNo">906</span> for (FileStatus f : families) {<a name="line.906"></a> -<span class="sourceLineNo">907</span> String familyStr = f.getPath().getName();<a name="line.907"></a> -<span class="sourceLineNo">908</span> HColumnDescriptor hcd = htd.getFamily(Bytes.toBytes(familyStr));<a name="line.908"></a> -<span class="sourceLineNo">909</span> // verify that the compression on this file matches the configured<a name="line.909"></a> -<span class="sourceLineNo">910</span> // compression<a name="line.910"></a> -<span class="sourceLineNo">911</span> Path dataFilePath = fs.listStatus(f.getPath())[0].getPath();<a name="line.911"></a> -<span class="sourceLineNo">912</span> Reader reader = HFile.createReader(fs, dataFilePath, new CacheConfig(conf), conf);<a name="line.912"></a> -<span class="sourceLineNo">913</span> Map<byte[], byte[]> fileInfo = reader.loadFileInfo();<a name="line.913"></a> -<span class="sourceLineNo">914</span><a name="line.914"></a> -<span class="sourceLineNo">915</span> byte[] bloomFilter = fileInfo.get(StoreFile.BLOOM_FILTER_TYPE_KEY);<a name="line.915"></a> -<span class="sourceLineNo">916</span> if (bloomFilter == null) bloomFilter = Bytes.toBytes("NONE");<a name="line.916"></a> -<span class="sourceLineNo">917</span> assertEquals("Incorrect bloom filter used for column family " + familyStr +<a name="line.917"></a> -<span class="sourceLineNo">918</span> "(reader: " + reader + ")",<a name="line.918"></a> -<span class="sourceLineNo">919</span> hcd.getBloomFilterType(), BloomType.valueOf(Bytes.toString(bloomFilter)));<a name="line.919"></a> -<span class="sourceLineNo">920</span> assertEquals("Incorrect compression used for column family " + familyStr +<a name="line.920"></a> -<span class="sourceLineNo">921</span> "(reader: " + reader + ")", hcd.getCompressionType(), reader.getFileContext().getCompression());<a name="line.921"></a> -<span class="sourceLineNo">922</span> }<a name="line.922"></a> -<span class="sourceLineNo">923</span> } finally {<a name="line.923"></a> -<span class="sourceLineNo">924</span> dir.getFileSystem(conf).delete(dir, true);<a name="line.924"></a> -<span class="sourceLineNo">925</span> }<a name="line.925"></a> -<span class="sourceLineNo">926</span> }<a name="line.926"></a> -<span class="sourceLineNo">927</span><a name="line.927"></a> -<span class="sourceLineNo">928</span> /**<a name="line.928"></a> -<span class="sourceLineNo">929</span> * Write random values to the writer assuming a table created using<a name="line.929"></a> -<span class="sourceLineNo">930</span> * {@link #FAMILIES} as column family descriptors<a name="line.930"></a> -<span class="sourceLineNo">931</span> */<a name="line.931"></a> -<span class="sourceLineNo">932</span> private void writeRandomKeyValues(RecordWriter<ImmutableBytesWritable, Cell> writer,<a name="line.932"></a> -<span class="sourceLineNo">933</span> TaskAttemptContext context, Set<byte[]> families, int numRows)<a name="line.933"></a> -<span class="sourceLin
<TRUNCATED>