Examples with HoodieTable org.apache.hudi.table.HoodieTable used on opensource projects

Example 26 with HoodieTable

use of org.apache.hudi.table.HoodieTable in project hudi by apache.

the class TestCopyOnWriteRollbackActionExecutor method testCopyOnWriteRollbackActionExecutorForFileListingAsGenerateFile.

@Test
public void testCopyOnWriteRollbackActionExecutorForFileListingAsGenerateFile() throws Exception {
    final String p1 = "2015/03/16";
    final String p2 = "2015/03/17";
    final String p3 = "2016/03/15";
    // Let's create some commit files and base files
    HoodieTestTable testTable = HoodieTestTable.of(metaClient).withPartitionMetaFiles(p1, p2, p3).addCommit("001").withBaseFilesInPartition(p1, "id11").withBaseFilesInPartition(p2, "id12").withLogFile(p1, "id11", 3).addCommit("002").withBaseFilesInPartition(p1, "id21").withBaseFilesInPartition(p2, "id22");
    HoodieWriteConfig writeConfig = getConfigBuilder().withRollbackUsingMarkers(false).build();
    HoodieTable table = this.getHoodieTable(metaClient, writeConfig);
    HoodieInstant needRollBackInstant = new HoodieInstant(false, HoodieTimeline.COMMIT_ACTION, "002");
    // execute CopyOnWriteRollbackActionExecutor with filelisting mode
    BaseRollbackPlanActionExecutor copyOnWriteRollbackPlanActionExecutor = new BaseRollbackPlanActionExecutor(context, table.getConfig(), table, "003", needRollBackInstant, false, table.getConfig().shouldRollbackUsingMarkers());
    HoodieRollbackPlan rollbackPlan = (HoodieRollbackPlan) copyOnWriteRollbackPlanActionExecutor.execute().get();
    CopyOnWriteRollbackActionExecutor copyOnWriteRollbackActionExecutor = new CopyOnWriteRollbackActionExecutor(context, table.getConfig(), table, "003", needRollBackInstant, true, false);
    List<HoodieRollbackStat> hoodieRollbackStats = copyOnWriteRollbackActionExecutor.executeRollback(rollbackPlan);
    // assert hoodieRollbackStats
    assertEquals(hoodieRollbackStats.size(), 3);
    for (HoodieRollbackStat stat : hoodieRollbackStats) {
        switch(stat.getPartitionPath()) {
            case p1:
                assertEquals(1, stat.getSuccessDeleteFiles().size());
                assertEquals(0, stat.getFailedDeleteFiles().size());
                assertEquals(Collections.EMPTY_MAP, stat.getCommandBlocksCount());
                assertEquals(testTable.forCommit("002").getBaseFilePath(p1, "id21").toString(), this.fs.getScheme() + ":" + stat.getSuccessDeleteFiles().get(0));
                break;
            case p2:
                assertEquals(1, stat.getSuccessDeleteFiles().size());
                assertEquals(0, stat.getFailedDeleteFiles().size());
                assertEquals(Collections.EMPTY_MAP, stat.getCommandBlocksCount());
                assertEquals(testTable.forCommit("002").getBaseFilePath(p2, "id22").toString(), this.fs.getScheme() + ":" + stat.getSuccessDeleteFiles().get(0));
                break;
            case p3:
                assertEquals(0, stat.getSuccessDeleteFiles().size());
                assertEquals(0, stat.getFailedDeleteFiles().size());
                assertEquals(Collections.EMPTY_MAP, stat.getCommandBlocksCount());
                break;
            default:
                fail("Unexpected partition: " + stat.getPartitionPath());
        }
    }
    assertTrue(testTable.inflightCommitExists("001"));
    assertTrue(testTable.commitExists("001"));
    assertTrue(testTable.baseFileExists(p1, "001", "id11"));
    assertTrue(testTable.baseFileExists(p2, "001", "id12"));
    assertFalse(testTable.inflightCommitExists("002"));
    assertFalse(testTable.commitExists("002"));
    assertFalse(testTable.baseFileExists(p1, "002", "id21"));
    assertFalse(testTable.baseFileExists(p2, "002", "id22"));
}

Example 27 with HoodieTable

use of org.apache.hudi.table.HoodieTable in project hudi by apache.

the class TestCopyOnWriteRollbackActionExecutor method testCopyOnWriteRollbackWithReplaceCommits.

// Verify that rollback works with replacecommit
@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testCopyOnWriteRollbackWithReplaceCommits(boolean isUsingMarkers) throws IOException {
    // 1. prepare data and assert data result
    List<FileSlice> firstPartitionCommit2FileSlices = new ArrayList<>();
    List<FileSlice> secondPartitionCommit2FileSlices = new ArrayList<>();
    HoodieWriteConfig cfg = getConfigBuilder().withRollbackUsingMarkers(isUsingMarkers).withAutoCommit(false).build();
    this.insertOverwriteCommitDataWithTwoPartitions(firstPartitionCommit2FileSlices, secondPartitionCommit2FileSlices, cfg, !isUsingMarkers);
    HoodieTable table = this.getHoodieTable(metaClient, cfg);
    performRollbackAndValidate(isUsingMarkers, cfg, table, firstPartitionCommit2FileSlices, secondPartitionCommit2FileSlices);
}

Example 28 with HoodieTable

use of org.apache.hudi.table.HoodieTable in project hudi by apache.

the class TestHoodieKeyLocationFetchHandle method testFetchHandle.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testFetchHandle(boolean populateMetaFields) throws Exception {
    metaClient = HoodieTestUtils.init(hadoopConf, basePath, HoodieTableType.COPY_ON_WRITE, populateMetaFields ? new Properties() : getPropertiesForKeyGen());
    config = getConfigBuilder().withProperties(getPropertiesForKeyGen()).withIndexConfig(HoodieIndexConfig.newBuilder().build()).build();
    List<HoodieRecord> records = dataGen.generateInserts(makeNewCommitTime(), 100);
    Map<String, List<HoodieRecord>> partitionRecordsMap = recordsToPartitionRecordsMap(records);
    HoodieTable hoodieTable = HoodieSparkTable.create(config, context, metaClient);
    HoodieSparkWriteableTestTable testTable = HoodieSparkWriteableTestTable.of(hoodieTable, AVRO_SCHEMA_WITH_METADATA_FIELDS);
    Map<Tuple2<String, String>, List<Tuple2<HoodieKey, HoodieRecordLocation>>> expectedList = writeToParquetAndGetExpectedRecordLocations(partitionRecordsMap, testTable);
    List<Tuple2<String, HoodieBaseFile>> partitionPathFileIdPairs = loadAllFilesForPartitions(new ArrayList<>(partitionRecordsMap.keySet()), context, hoodieTable);
    BaseKeyGenerator keyGenerator = (BaseKeyGenerator) HoodieSparkKeyGeneratorFactory.createKeyGenerator(new TypedProperties(getPropertiesForKeyGen()));
    for (Tuple2<String, HoodieBaseFile> entry : partitionPathFileIdPairs) {
        HoodieKeyLocationFetchHandle fetcherHandle = new HoodieKeyLocationFetchHandle(config, hoodieTable, Pair.of(entry._1, entry._2), populateMetaFields ? Option.empty() : Option.of(keyGenerator));
        Iterator<Pair<HoodieKey, HoodieRecordLocation>> result = fetcherHandle.locations().iterator();
        List<Tuple2<HoodieKey, HoodieRecordLocation>> actualList = new ArrayList<>();
        result.forEachRemaining(x -> actualList.add(new Tuple2<>(x.getLeft(), x.getRight())));
        assertEquals(expectedList.get(new Tuple2<>(entry._1, entry._2.getFileId())), actualList);
    }
}

Example 29 with HoodieTable

use of org.apache.hudi.table.HoodieTable in project hudi by apache.

the class TestHoodieTimelineArchiver method testMergeSmallArchiveFilesRecoverFromMergeFailed.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testMergeSmallArchiveFilesRecoverFromMergeFailed(boolean enableArchiveMerge) throws Exception {
    HoodieWriteConfig writeConfig = initTestTableAndGetWriteConfig(true, 2, 3, 2, enableArchiveMerge, 3, 209715200);
    // do ingestion and trigger archive actions here.
    for (int i = 1; i < 8; i++) {
        testTable.doWriteOperation("0000000" + i, WriteOperationType.UPSERT, i == 1 ? Arrays.asList("p1", "p2") : Collections.emptyList(), Arrays.asList("p1", "p2"), 2);
        archiveAndGetCommitsList(writeConfig);
    }
    // do a single merge small archive files
    HoodieTable table = HoodieSparkTable.create(writeConfig, context, metaClient);
    HoodieTimelineArchiver archiver = new HoodieTimelineArchiver(writeConfig, table);
    FileStatus[] fsStatuses = metaClient.getFs().globStatus(new Path(metaClient.getArchivePath() + "/.commits_.archive*"));
    List<String> candidateFiles = Arrays.stream(fsStatuses).map(fs -> fs.getPath().toString()).collect(Collectors.toList());
    archiver.reOpenWriter();
    archiver.buildArchiveMergePlan(candidateFiles, new Path(metaClient.getArchivePath(), HoodieArchivedTimeline.MERGE_ARCHIVE_PLAN_NAME), ".commits_.archive.3_1-0-1");
    archiver.mergeArchiveFiles(Arrays.stream(fsStatuses).collect(Collectors.toList()));
    HoodieLogFormat.Writer writer = archiver.reOpenWriter();
    // check loading archived and active timeline success
    HoodieActiveTimeline rawActiveTimeline = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine = metaClient.getArchivedTimeline().reload();
    assertEquals(7 * 3, rawActiveTimeline.countInstants() + archivedTimeLine.reload().countInstants());
    String s = "Dummy Content";
    // stain the current merged archive file.
    FileIOUtils.createFileInPath(metaClient.getFs(), writer.getLogFile().getPath(), Option.of(s.getBytes()));
    // do another archive actions with merge small archive files.
    for (int i = 1; i < 10; i++) {
        testTable.doWriteOperation("1000000" + i, WriteOperationType.UPSERT, i == 1 ? Arrays.asList("p1", "p2") : Collections.emptyList(), Arrays.asList("p1", "p2"), 2);
        archiveAndGetCommitsList(writeConfig);
    }
    // check result.
    // we need to load archived timeline successfully and ignore the parsing damage merged archive files exception.
    HoodieActiveTimeline rawActiveTimeline1 = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine1 = metaClient.getArchivedTimeline().reload();
    assertEquals(16 * 3, archivedTimeLine1.countInstants() + rawActiveTimeline1.countInstants());
    // if there are a damaged merged archive files and other common damaged archive file.
    // hoodie need throw ioe while loading archived timeline because of parsing the damaged archive file.
    Path damagedFile = new Path(metaClient.getArchivePath(), ".commits_.archive.300_1-0-1");
    FileIOUtils.createFileInPath(metaClient.getFs(), damagedFile, Option.of(s.getBytes()));
    assertThrows(HoodieException.class, () -> metaClient.getArchivedTimeline().reload());
}

Example 30 with HoodieTable

use of org.apache.hudi.table.HoodieTable in project hudi by apache.

the class TestHoodieTimelineArchiver method testArchiveCompletedRollbackAndClean.

@ParameterizedTest
@CsvSource({ "true,true", "true,false", "false,true", "false,false" })
public void testArchiveCompletedRollbackAndClean(boolean isEmpty, boolean enableMetadataTable) throws Exception {
    init();
    int minInstantsToKeep = 2;
    int maxInstantsToKeep = 10;
    HoodieWriteConfig cfg = HoodieWriteConfig.newBuilder().withPath(basePath).withSchema(HoodieTestDataGenerator.TRIP_EXAMPLE_SCHEMA).withParallelism(2, 2).forTable("test-trip-table").withCompactionConfig(HoodieCompactionConfig.newBuilder().retainCommits(1).archiveCommitsWith(minInstantsToKeep, maxInstantsToKeep).build()).withFileSystemViewConfig(FileSystemViewStorageConfig.newBuilder().withRemoteServerPort(timelineServicePort).build()).withMetadataConfig(HoodieMetadataConfig.newBuilder().enable(enableMetadataTable).build()).build();
    metaClient = HoodieTableMetaClient.reload(metaClient);
    int startInstant = 1;
    for (int i = 0; i < maxInstantsToKeep + 1; i++, startInstant++) {
        createCleanMetadata(startInstant + "", false, isEmpty || i % 2 == 0);
    }
    for (int i = 0; i < maxInstantsToKeep + 1; i++, startInstant += 2) {
        createCommitAndRollbackFile(startInstant + 1 + "", startInstant + "", false, isEmpty || i % 2 == 0);
    }
    if (enableMetadataTable) {
        // Simulate a compaction commit in metadata table timeline
        // so the archival in data table can happen
        createCompactionCommitInMetadataTable(hadoopConf, wrapperFs, basePath, Integer.toString(99));
    }
    HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
    HoodieTimelineArchiver archiver = new HoodieTimelineArchiver(cfg, table);
    archiver.archiveIfRequired(context);
    Stream<HoodieInstant> currentInstants = metaClient.getActiveTimeline().reload().getInstants();
    Map<Object, List<HoodieInstant>> actionInstantMap = currentInstants.collect(Collectors.groupingBy(HoodieInstant::getAction));
    assertTrue(actionInstantMap.containsKey("clean"), "Clean Action key must be preset");
    assertEquals(minInstantsToKeep, actionInstantMap.get("clean").size(), "Should have min instant");
    assertTrue(actionInstantMap.containsKey("rollback"), "Rollback Action key must be preset");
    assertEquals(minInstantsToKeep, actionInstantMap.get("rollback").size(), "Should have min instant");
}