Examples with HoodieInstant org.apache.hudi.common.table.timeline.HoodieInstant used on opensource projects

Example 11 with HoodieInstant

use of org.apache.hudi.common.table.timeline.HoodieInstant in project hudi by apache.

the class TestRepairUtils method testGetBaseAndLogFilePathsFromTimeline.

@Test
public void testGetBaseAndLogFilePathsFromTimeline() throws IOException {
    setupTimelineInFS();
    HoodieTimeline timeline = metaClient.getActiveTimeline();
    HoodieInstant commitInstant = new HoodieInstant(HoodieInstant.State.COMPLETED, HoodieTimeline.COMMIT_ACTION, "001");
    HoodieInstant inflightInstant = new HoodieInstant(HoodieInstant.State.INFLIGHT, HoodieTimeline.COMMIT_ACTION, "005");
    HoodieInstant compactionInstant = new HoodieInstant(HoodieInstant.State.COMPLETED, HoodieTimeline.COMPACTION_ACTION, "006");
    Map<String, List<Pair<String, String>>> partitionToFileIdAndNameMap = instantInfoMap.get(commitInstant.getTimestamp());
    Set<String> expectedPaths = partitionToFileIdAndNameMap.entrySet().stream().flatMap(entry -> entry.getValue().stream().map(fileInfo -> new Path(entry.getKey(), fileInfo.getValue()).toString()).collect(Collectors.toList()).stream()).collect(Collectors.toSet());
    assertEquals(Option.of(expectedPaths), RepairUtils.getBaseAndLogFilePathsFromTimeline(timeline, commitInstant));
    assertThrows(HoodieException.class, () -> RepairUtils.getBaseAndLogFilePathsFromTimeline(timeline, inflightInstant));
    assertEquals(Option.empty(), RepairUtils.getBaseAndLogFilePathsFromTimeline(timeline, compactionInstant));
}

Example 12 with HoodieInstant

use of org.apache.hudi.common.table.timeline.HoodieInstant in project hudi by apache.

the class TestCleaner method testKeepLatestCommits.

/**
 * Test HoodieTable.clean() Cleaning by commit logic for COW table.
 */
@ParameterizedTest
@MethodSource("argumentsForTestKeepLatestCommits")
public void testKeepLatestCommits(boolean simulateFailureRetry, boolean enableIncrementalClean, boolean enableBootstrapSourceClean) throws Exception {
    HoodieWriteConfig config = HoodieWriteConfig.newBuilder().withPath(basePath).withMetadataConfig(HoodieMetadataConfig.newBuilder().withAssumeDatePartitioning(true).build()).withCompactionConfig(HoodieCompactionConfig.newBuilder().withIncrementalCleaningMode(enableIncrementalClean).withFailedWritesCleaningPolicy(HoodieFailedWritesCleaningPolicy.EAGER).withCleanBootstrapBaseFileEnabled(enableBootstrapSourceClean).withCleanerPolicy(HoodieCleaningPolicy.KEEP_LATEST_COMMITS).retainCommits(2).build()).build();
    HoodieTestTable testTable = HoodieTestTable.of(metaClient);
    String p0 = "2020/01/01";
    String p1 = "2020/01/02";
    Map<String, List<BootstrapFileMapping>> bootstrapMapping = enableBootstrapSourceClean ? generateBootstrapIndexAndSourceData(p0, p1) : null;
    // make 1 commit, with 1 file per partition
    String file1P0C0 = enableBootstrapSourceClean ? bootstrapMapping.get(p0).get(0).getFileId() : UUID.randomUUID().toString();
    String file1P1C0 = enableBootstrapSourceClean ? bootstrapMapping.get(p1).get(0).getFileId() : UUID.randomUUID().toString();
    testTable.addInflightCommit("00000000000001").withBaseFilesInPartition(p0, file1P0C0).withBaseFilesInPartition(p1, file1P1C0);
    HoodieCommitMetadata commitMetadata = generateCommitMetadata("00000000000001", Collections.unmodifiableMap(new HashMap<String, List<String>>() {

        {
            put(p0, CollectionUtils.createImmutableList(file1P0C0));
            put(p1, CollectionUtils.createImmutableList(file1P1C0));
        }
    }));
    metaClient.getActiveTimeline().saveAsComplete(new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMMIT_ACTION, "00000000000001"), Option.of(commitMetadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    metaClient = HoodieTableMetaClient.reload(metaClient);
    List<HoodieCleanStat> hoodieCleanStatsOne = runCleaner(config, simulateFailureRetry);
    assertEquals(0, hoodieCleanStatsOne.size(), "Must not scan any partitions and clean any files");
    assertTrue(testTable.baseFileExists(p0, "00000000000001", file1P0C0));
    assertTrue(testTable.baseFileExists(p1, "00000000000001", file1P1C0));
    // make next commit, with 1 insert & 1 update per partition
    Map<String, String> partitionAndFileId002 = testTable.addInflightCommit("00000000000002").getFileIdsWithBaseFilesInPartitions(p0, p1);
    String file2P0C1 = partitionAndFileId002.get(p0);
    String file2P1C1 = partitionAndFileId002.get(p1);
    testTable.forCommit("00000000000002").withBaseFilesInPartition(p0, file1P0C0).withBaseFilesInPartition(p1, file1P1C0);
    commitMetadata = generateCommitMetadata("00000000000002", new HashMap<String, List<String>>() {

        {
            put(p0, CollectionUtils.createImmutableList(file1P0C0, file2P0C1));
            put(p1, CollectionUtils.createImmutableList(file1P1C0, file2P1C1));
        }
    });
    metaClient.getActiveTimeline().saveAsComplete(new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMMIT_ACTION, "00000000000002"), Option.of(commitMetadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    List<HoodieCleanStat> hoodieCleanStatsTwo = runCleaner(config, simulateFailureRetry);
    assertEquals(0, hoodieCleanStatsTwo.size(), "Must not scan any partitions and clean any files");
    assertTrue(testTable.baseFileExists(p0, "00000000000002", file2P0C1));
    assertTrue(testTable.baseFileExists(p1, "00000000000002", file2P1C1));
    assertTrue(testTable.baseFileExists(p0, "00000000000001", file1P0C0));
    assertTrue(testTable.baseFileExists(p1, "00000000000001", file1P1C0));
    // make next commit, with 2 updates to existing files, and 1 insert
    String file3P0C2 = testTable.addInflightCommit("00000000000003").withBaseFilesInPartition(p0, file1P0C0).withBaseFilesInPartition(p0, file2P0C1).getFileIdsWithBaseFilesInPartitions(p0).get(p0);
    commitMetadata = generateCommitMetadata("00000000000003", CollectionUtils.createImmutableMap(p0, CollectionUtils.createImmutableList(file1P0C0, file2P0C1, file3P0C2)));
    metaClient.getActiveTimeline().saveAsComplete(new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMMIT_ACTION, "00000000000003"), Option.of(commitMetadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    List<HoodieCleanStat> hoodieCleanStatsThree = runCleaner(config, simulateFailureRetry);
    assertEquals(0, hoodieCleanStatsThree.size(), "Must not clean any file. We have to keep 1 version before the latest commit time to keep");
    assertTrue(testTable.baseFileExists(p0, "00000000000001", file1P0C0));
    // make next commit, with 2 updates to existing files, and 1 insert
    String file4P0C3 = testTable.addInflightCommit("00000000000004").withBaseFilesInPartition(p0, file1P0C0).withBaseFilesInPartition(p0, file2P0C1).getFileIdsWithBaseFilesInPartitions(p0).get(p0);
    commitMetadata = generateCommitMetadata("00000000000004", CollectionUtils.createImmutableMap(p0, CollectionUtils.createImmutableList(file1P0C0, file2P0C1, file4P0C3)));
    metaClient.getActiveTimeline().saveAsComplete(new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMMIT_ACTION, "00000000000004"), Option.of(commitMetadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    List<HoodieCleanStat> hoodieCleanStatsFour = runCleaner(config, simulateFailureRetry);
    // enableBootstrapSourceClean would delete the bootstrap base file as the same time
    HoodieCleanStat partitionCleanStat = getCleanStat(hoodieCleanStatsFour, p0);
    assertEquals(enableBootstrapSourceClean ? 2 : 1, partitionCleanStat.getSuccessDeleteFiles().size() + (partitionCleanStat.getSuccessDeleteBootstrapBaseFiles() == null ? 0 : partitionCleanStat.getSuccessDeleteBootstrapBaseFiles().size()), "Must clean at least one old file");
    assertFalse(testTable.baseFileExists(p0, "00000000000001", file1P0C0));
    assertTrue(testTable.baseFileExists(p0, "00000000000002", file1P0C0));
    assertTrue(testTable.baseFileExists(p0, "00000000000003", file1P0C0));
    assertTrue(testTable.baseFileExists(p0, "00000000000002", file2P0C1));
    assertTrue(testTable.baseFileExists(p0, "00000000000003", file2P0C1));
    assertTrue(testTable.baseFileExists(p0, "00000000000003", file3P0C2));
    assertTrue(testTable.baseFileExists(p0, "00000000000004", file4P0C3));
    if (enableBootstrapSourceClean) {
        assertFalse(Files.exists(Paths.get(bootstrapMapping.get(p0).get(0).getBootstrapFileStatus().getPath().getUri())));
    }
    // No cleaning on partially written file, with no commit.
    testTable.forCommit("00000000000005").withBaseFilesInPartition(p0, file3P0C2);
    commitMetadata = generateCommitMetadata("00000000000005", CollectionUtils.createImmutableMap(p0, CollectionUtils.createImmutableList(file3P0C2)));
    metaClient.getActiveTimeline().createNewInstant(new HoodieInstant(State.REQUESTED, HoodieTimeline.COMMIT_ACTION, "00000000000005"));
    metaClient.getActiveTimeline().transitionRequestedToInflight(new HoodieInstant(State.REQUESTED, HoodieTimeline.COMMIT_ACTION, "00000000000005"), Option.of(commitMetadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    List<HoodieCleanStat> hoodieCleanStatsFive = runCleaner(config, simulateFailureRetry);
    HoodieCleanStat cleanStat = getCleanStat(hoodieCleanStatsFive, p0);
    assertNull(cleanStat, "Must not clean any files");
    assertTrue(testTable.baseFileExists(p0, "00000000000002", file1P0C0));
    assertTrue(testTable.baseFileExists(p0, "00000000000002", file2P0C1));
    assertTrue(testTable.baseFileExists(p0, "00000000000005", file3P0C2));
}

Example 13 with HoodieInstant

use of org.apache.hudi.common.table.timeline.HoodieInstant in project hudi by apache.

the class TestCleaner method testFailedInsertAndCleanByCommits.

/**
 * Test Helper for Cleaning failed commits by commits logic from HoodieWriteClient API perspective.
 *
 * @param insertFn Insert API to be tested
 * @param isPreppedAPI Flag to indicate if a prepped-version is used. If true, a wrapper function will be used during
 *        record generation to also tag the regards (de-dupe is implicit as we use uniq record-gen APIs)
 * @throws Exception in case of errors
 */
private void testFailedInsertAndCleanByCommits(Function3<JavaRDD<WriteStatus>, SparkRDDWriteClient, JavaRDD<HoodieRecord>, String> insertFn, boolean isPreppedAPI) throws Exception {
    // keep upto 3 commits from the past
    int maxCommits = 3;
    HoodieWriteConfig cfg = getConfigBuilder().withAutoCommit(false).withHeartbeatIntervalInMs(3000).withCompactionConfig(HoodieCompactionConfig.newBuilder().withFailedWritesCleaningPolicy(HoodieFailedWritesCleaningPolicy.LAZY).withCleanerPolicy(HoodieCleaningPolicy.KEEP_LATEST_COMMITS).retainCommits(maxCommits).build()).withParallelism(1, 1).withBulkInsertParallelism(1).withFinalizeWriteParallelism(1).withDeleteParallelism(1).withConsistencyGuardConfig(ConsistencyGuardConfig.newBuilder().withConsistencyCheckEnabled(true).build()).build();
    SparkRDDWriteClient client = getHoodieWriteClient(cfg);
    final Function2<List<HoodieRecord>, String, Integer> recordInsertGenWrappedFunction = generateWrapRecordsFn(isPreppedAPI, cfg, dataGen::generateInserts);
    Pair<String, JavaRDD<WriteStatus>> result = insertFirstBigBatchForClientCleanerTest(cfg, client, recordInsertGenWrappedFunction, insertFn, HoodieCleaningPolicy.KEEP_LATEST_COMMITS);
    client.commit(result.getLeft(), result.getRight());
    HoodieTable table = HoodieSparkTable.create(client.getConfig(), context, metaClient);
    assertTrue(table.getCompletedCleanTimeline().empty());
    insertFirstFailedBigBatchForClientCleanerTest(cfg, client, recordInsertGenWrappedFunction, insertFn, HoodieCleaningPolicy.KEEP_LATEST_COMMITS);
    insertFirstFailedBigBatchForClientCleanerTest(cfg, client, recordInsertGenWrappedFunction, insertFn, HoodieCleaningPolicy.KEEP_LATEST_COMMITS);
    Pair<String, JavaRDD<WriteStatus>> ret = insertFirstFailedBigBatchForClientCleanerTest(cfg, client, recordInsertGenWrappedFunction, insertFn, HoodieCleaningPolicy.KEEP_LATEST_COMMITS);
    // Await till enough time passes such that the last failed commits heartbeats are expired
    await().atMost(10, TimeUnit.SECONDS).until(() -> client.getHeartbeatClient().isHeartbeatExpired(ret.getLeft()));
    List<HoodieCleanStat> cleanStats = runCleaner(cfg);
    assertEquals(0, cleanStats.size(), "Must not clean any files");
    HoodieActiveTimeline timeline = metaClient.reloadActiveTimeline();
    assertTrue(timeline.getTimelineOfActions(CollectionUtils.createSet(HoodieTimeline.ROLLBACK_ACTION)).filterCompletedInstants().countInstants() == 3);
    Option<HoodieInstant> rollBackInstantForFailedCommit = timeline.getTimelineOfActions(CollectionUtils.createSet(HoodieTimeline.ROLLBACK_ACTION)).filterCompletedInstants().lastInstant();
    HoodieRollbackMetadata rollbackMetadata = TimelineMetadataUtils.deserializeAvroMetadata(timeline.getInstantDetails(rollBackInstantForFailedCommit.get()).get(), HoodieRollbackMetadata.class);
    // Rollback of one of the failed writes should have deleted 3 files
    assertEquals(3, rollbackMetadata.getTotalFilesDeleted());
}

Example 14 with HoodieInstant

use of org.apache.hudi.common.table.timeline.HoodieInstant in project hudi by apache.

the class TestHoodieMergeOnReadTable method testUpsertPartitioner.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testUpsertPartitioner(boolean populateMetaFields) throws Exception {
    HoodieWriteConfig.Builder cfgBuilder = getConfigBuilder(true);
    addConfigsForPopulateMetaFields(cfgBuilder, populateMetaFields);
    HoodieWriteConfig cfg = cfgBuilder.build();
    try (SparkRDDWriteClient client = getHoodieWriteClient(cfg)) {
        /**
         * Write 1 (only inserts, written as base file)
         */
        String newCommitTime = "001";
        client.startCommitWithTime(newCommitTime);
        List<HoodieRecord> records = dataGen.generateInserts(newCommitTime, 20);
        JavaRDD<HoodieRecord> writeRecords = jsc().parallelize(records, 1);
        List<WriteStatus> statuses = client.upsert(writeRecords, newCommitTime).collect();
        assertNoWriteErrors(statuses);
        HoodieTable hoodieTable = HoodieSparkTable.create(cfg, context(), metaClient);
        Option<HoodieInstant> deltaCommit = metaClient.getActiveTimeline().getDeltaCommitTimeline().firstInstant();
        assertTrue(deltaCommit.isPresent());
        assertEquals("001", deltaCommit.get().getTimestamp(), "Delta commit should be 001");
        Option<HoodieInstant> commit = metaClient.getActiveTimeline().getCommitTimeline().firstInstant();
        assertFalse(commit.isPresent());
        FileStatus[] allFiles = listAllBaseFilesInPath(hoodieTable);
        BaseFileOnlyView roView = getHoodieTableFileSystemView(metaClient, metaClient.getCommitsTimeline().filterCompletedInstants(), allFiles);
        Stream<HoodieBaseFile> dataFilesToRead = roView.getLatestBaseFiles();
        Map<String, Long> fileIdToSize = dataFilesToRead.collect(Collectors.toMap(HoodieBaseFile::getFileId, HoodieBaseFile::getFileSize));
        roView = getHoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
        dataFilesToRead = roView.getLatestBaseFiles();
        List<HoodieBaseFile> dataFilesList = dataFilesToRead.collect(Collectors.toList());
        assertTrue(dataFilesList.size() > 0, "Should list the base files we wrote in the delta commit");
        /**
         * Write 2 (only updates + inserts, written to .log file + correction of existing base file size)
         */
        newCommitTime = "002";
        client.startCommitWithTime(newCommitTime);
        List<HoodieRecord> newRecords = dataGen.generateUpdates(newCommitTime, records);
        newRecords.addAll(dataGen.generateInserts(newCommitTime, 20));
        statuses = client.upsert(jsc().parallelize(newRecords), newCommitTime).collect();
        // Verify there are no errors
        assertNoWriteErrors(statuses);
        metaClient = HoodieTableMetaClient.reload(metaClient);
        deltaCommit = metaClient.getActiveTimeline().getDeltaCommitTimeline().lastInstant();
        assertTrue(deltaCommit.isPresent());
        assertEquals("002", deltaCommit.get().getTimestamp(), "Latest Delta commit should be 002");
        commit = metaClient.getActiveTimeline().getCommitTimeline().firstInstant();
        assertFalse(commit.isPresent());
        allFiles = listAllBaseFilesInPath(hoodieTable);
        roView = getHoodieTableFileSystemView(metaClient, hoodieTable.getActiveTimeline().reload().getCommitsTimeline().filterCompletedInstants(), allFiles);
        dataFilesToRead = roView.getLatestBaseFiles();
        List<HoodieBaseFile> newDataFilesList = dataFilesToRead.collect(Collectors.toList());
        Map<String, Long> fileIdToNewSize = newDataFilesList.stream().collect(Collectors.toMap(HoodieBaseFile::getFileId, HoodieBaseFile::getFileSize));
        assertTrue(fileIdToNewSize.entrySet().stream().anyMatch(entry -> fileIdToSize.get(entry.getKey()) < entry.getValue()));
        List<String> inputPaths = roView.getLatestBaseFiles().map(baseFile -> new Path(baseFile.getPath()).getParent().toString()).collect(Collectors.toList());
        List<GenericRecord> recordsRead = HoodieMergeOnReadTestUtils.getRecordsUsingInputFormat(hadoopConf(), inputPaths, basePath(), new JobConf(hadoopConf()), true, false);
        // Wrote 20 records in 2 batches
        assertEquals(40, recordsRead.size(), "Must contain 40 records");
    }
}

Example 15 with HoodieInstant

use of org.apache.hudi.common.table.timeline.HoodieInstant in project hudi by apache.

the class TestHoodieMergeOnReadTable method testMetadataStatsOnCommit.

/**
 * Test to ensure metadata stats are correctly written to metadata file.
 */
@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testMetadataStatsOnCommit(Boolean rollbackUsingMarkers) throws Exception {
    HoodieWriteConfig cfg = getConfigBuilder(false, rollbackUsingMarkers, IndexType.INMEMORY).withAutoCommit(false).build();
    try (SparkRDDWriteClient client = getHoodieWriteClient(cfg)) {
        HoodieTable table = HoodieSparkTable.create(cfg, context(), metaClient);
        // Create a commit without metadata stats in metadata to test backwards compatibility
        HoodieActiveTimeline activeTimeline = table.getActiveTimeline();
        String commitActionType = table.getMetaClient().getCommitActionType();
        HoodieInstant instant = new HoodieInstant(State.REQUESTED, commitActionType, "000");
        activeTimeline.createNewInstant(instant);
        activeTimeline.transitionRequestedToInflight(instant, Option.empty());
        instant = new HoodieInstant(State.INFLIGHT, commitActionType, "000");
        activeTimeline.saveAsComplete(instant, Option.empty());
        String instantTime = "001";
        client.startCommitWithTime(instantTime);
        List<HoodieRecord> records = dataGen.generateInserts(instantTime, 200);
        JavaRDD<HoodieRecord> writeRecords = jsc().parallelize(records, 1);
        JavaRDD<WriteStatus> statuses = client.insert(writeRecords, instantTime);
        assertTrue(client.commit(instantTime, statuses), "Commit should succeed");
        // Read from commit file
        table = HoodieSparkTable.create(cfg, context());
        HoodieCommitMetadata metadata = HoodieCommitMetadata.fromBytes(table.getActiveTimeline().getInstantDetails(table.getActiveTimeline().getDeltaCommitTimeline().lastInstant().get()).get(), HoodieCommitMetadata.class);
        int inserts = 0;
        for (Map.Entry<String, List<HoodieWriteStat>> pstat : metadata.getPartitionToWriteStats().entrySet()) {
            for (HoodieWriteStat stat : pstat.getValue()) {
                inserts += stat.getNumInserts();
            }
        }
        assertEquals(200, inserts);
        instantTime = "002";
        client.startCommitWithTime(instantTime);
        records = dataGen.generateUpdates(instantTime, records);
        writeRecords = jsc().parallelize(records, 1);
        statuses = client.upsert(writeRecords, instantTime);
        // assertTrue(client.commit(instantTime, statuses), "Commit should succeed");
        inserts = 0;
        int upserts = 0;
        List<WriteStatus> writeStatusList = statuses.collect();
        for (WriteStatus ws : writeStatusList) {
            inserts += ws.getStat().getNumInserts();
            upserts += ws.getStat().getNumUpdateWrites();
        }
        // Read from commit file
        assertEquals(0, inserts);
        assertEquals(200, upserts);
        client.rollback(instantTime);
        // Read from commit file
        table = HoodieSparkTable.create(cfg, context());
        metadata = HoodieCommitMetadata.fromBytes(table.getActiveTimeline().getInstantDetails(table.getActiveTimeline().getDeltaCommitTimeline().lastInstant().get()).get(), HoodieCommitMetadata.class);
        inserts = 0;
        upserts = 0;
        for (Map.Entry<String, List<HoodieWriteStat>> pstat : metadata.getPartitionToWriteStats().entrySet()) {
            for (HoodieWriteStat stat : pstat.getValue()) {
                inserts += stat.getNumInserts();
                upserts += stat.getNumUpdateWrites();
            }
        }
        assertEquals(200, inserts);
        assertEquals(0, upserts);
    }
}