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

Example 6 with HoodieArchivedTimeline

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

the class TestHoodieTimelineArchiver method testMergeSmallArchiveFilesRecoverFromDeleteFailed.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testMergeSmallArchiveFilesRecoverFromDeleteFailed(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()));
    archiver.reOpenWriter();
    // delete only one of the small archive file to simulate delete action failed.
    metaClient.getFs().delete(fsStatuses[0].getPath());
    // loading archived timeline and active timeline success
    HoodieActiveTimeline rawActiveTimeline = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine = metaClient.getArchivedTimeline().reload();
    assertEquals(7 * 3, rawActiveTimeline.countInstants() + archivedTimeLine.countInstants());
    // 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.
    HoodieActiveTimeline rawActiveTimeline1 = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine1 = metaClient.getArchivedTimeline().reload();
    assertEquals(16 * 3, archivedTimeLine1.countInstants() + rawActiveTimeline1.countInstants());
}

Example 7 with HoodieArchivedTimeline

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

the class TestHoodieTimelineArchiver method testLoadArchiveTimelineWithUncompletedMergeArchiveFile.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testLoadArchiveTimelineWithUncompletedMergeArchiveFile(boolean enableArchiveMerge) throws Exception {
    HoodieWriteConfig writeConfig = initTestTableAndGetWriteConfig(true, 2, 3, 2, enableArchiveMerge, 3, 209715200);
    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);
    }
    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();
    String s = "Dummy Content";
    // stain the current merged archive file.
    FileIOUtils.createFileInPath(metaClient.getFs(), writer.getLogFile().getPath(), Option.of(s.getBytes()));
    // if there's only a damaged merged archive file, we need to ignore the exception while reading this damaged file.
    HoodieActiveTimeline rawActiveTimeline1 = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine1 = metaClient.getArchivedTimeline();
    assertEquals(7 * 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 8 with HoodieArchivedTimeline

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

the class TestHoodieTimelineArchiver method testArchiveCommitTimeline.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testArchiveCommitTimeline(boolean enableMetadataTable) throws Exception {
    init();
    HoodieWriteConfig cfg = HoodieWriteConfig.newBuilder().withPath(basePath).withSchema(HoodieTestDataGenerator.TRIP_EXAMPLE_SCHEMA).withParallelism(2, 2).forTable("test-trip-table").withCompactionConfig(HoodieCompactionConfig.newBuilder().retainCommits(1).archiveCommitsWith(2, 3).build()).withFileSystemViewConfig(FileSystemViewStorageConfig.newBuilder().withRemoteServerPort(timelineServicePort).build()).withMetadataConfig(HoodieMetadataConfig.newBuilder().enable(enableMetadataTable).build()).build();
    metaClient = HoodieTableMetaClient.reload(metaClient);
    HoodieTestDataGenerator.createCommitFile(basePath, "1", wrapperFs.getConf());
    HoodieInstant instant1 = new HoodieInstant(false, HoodieTimeline.COMMIT_ACTION, "1");
    HoodieTestDataGenerator.createCommitFile(basePath, "2", wrapperFs.getConf());
    Path markerPath = new Path(metaClient.getMarkerFolderPath("2"));
    wrapperFs.mkdirs(markerPath);
    HoodieInstant instant2 = new HoodieInstant(false, HoodieTimeline.COMMIT_ACTION, "2");
    HoodieTestDataGenerator.createCommitFile(basePath, "3", wrapperFs.getConf());
    HoodieInstant instant3 = new HoodieInstant(false, HoodieTimeline.COMMIT_ACTION, "3");
    // add 2 more instants to pass filter criteria set in compaction config above
    HoodieTestDataGenerator.createCommitFile(basePath, "4", wrapperFs.getConf());
    HoodieTestDataGenerator.createCommitFile(basePath, "5", wrapperFs.getConf());
    if (enableMetadataTable) {
        // Simulate a compaction commit in metadata table timeline
        // so the archival in data table can happen
        createCompactionCommitInMetadataTable(hadoopConf, wrapperFs, basePath, "5");
    }
    HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
    HoodieTimelineArchiver archiver = new HoodieTimelineArchiver(cfg, table);
    boolean result = archiver.archiveIfRequired(context);
    assertTrue(result);
    HoodieArchivedTimeline archivedTimeline = metaClient.getArchivedTimeline();
    List<HoodieInstant> archivedInstants = Arrays.asList(instant1, instant2, instant3);
    assertEquals(new HashSet<>(archivedInstants), archivedTimeline.filterCompletedInstants().getInstants().collect(Collectors.toSet()));
    assertFalse(wrapperFs.exists(markerPath));
}

Example 9 with HoodieArchivedTimeline

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

the class TestHoodieTimelineArchiver method testMergeSmallArchiveFilesRecoverFromBuildPlanFailed.

@ParameterizedTest
@ValueSource(booleans = { true, false })
public void testMergeSmallArchiveFilesRecoverFromBuildPlanFailed(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);
    }
    // build a merge small archive plan with dummy content
    // this plan can not be deserialized.
    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();
    Path plan = new Path(metaClient.getArchivePath(), HoodieArchivedTimeline.MERGE_ARCHIVE_PLAN_NAME);
    archiver.buildArchiveMergePlan(candidateFiles, plan, ".commits_.archive.3_1-0-1");
    String s = "Dummy Content";
    // stain the current merge plan file.
    FileIOUtils.createFileInPath(metaClient.getFs(), plan, Option.of(s.getBytes()));
    // check that damaged plan file will not block archived timeline loading.
    HoodieActiveTimeline rawActiveTimeline = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine = metaClient.getArchivedTimeline().reload();
    assertEquals(7 * 3, rawActiveTimeline.countInstants() + archivedTimeLine.countInstants());
    // trigger several archive after left damaged merge small archive file plan.
    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);
    }
    // loading archived timeline and active timeline success
    HoodieActiveTimeline rawActiveTimeline1 = new HoodieActiveTimeline(metaClient, false);
    HoodieArchivedTimeline archivedTimeLine1 = metaClient.getArchivedTimeline().reload();
    // check instant number
    assertEquals(16 * 3, archivedTimeLine1.countInstants() + rawActiveTimeline1.countInstants());
    // if there are damaged archive files and damaged plan, hoodie need throw ioe while loading archived timeline.
    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 10 with HoodieArchivedTimeline

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

the class CommitsCommand method showArchivedCommits.

@CliCommand(value = "commits showarchived", help = "Show the archived commits")
public String showArchivedCommits(@CliOption(key = { "includeExtraMetadata" }, help = "Include extra metadata", unspecifiedDefaultValue = "false") final boolean includeExtraMetadata, @CliOption(key = { "createView" }, mandatory = false, help = "view name to store output table", unspecifiedDefaultValue = "") final String exportTableName, @CliOption(key = { "startTs" }, mandatory = false, help = "start time for commits, default: now - 10 days") String startTs, @CliOption(key = { "endTs" }, mandatory = false, help = "end time for commits, default: now - 1 day") String endTs, @CliOption(key = { "limit" }, mandatory = false, help = "Limit commits", unspecifiedDefaultValue = "-1") final Integer limit, @CliOption(key = { "sortBy" }, help = "Sorting Field", unspecifiedDefaultValue = "") final String sortByField, @CliOption(key = { "desc" }, help = "Ordering", unspecifiedDefaultValue = "false") final boolean descending, @CliOption(key = { "headeronly" }, help = "Print Header Only", unspecifiedDefaultValue = "false") final boolean headerOnly) throws IOException {
    if (StringUtils.isNullOrEmpty(startTs)) {
        startTs = CommitUtil.getTimeDaysAgo(10);
    }
    if (StringUtils.isNullOrEmpty(endTs)) {
        endTs = CommitUtil.getTimeDaysAgo(1);
    }
    HoodieArchivedTimeline archivedTimeline = HoodieCLI.getTableMetaClient().getArchivedTimeline();
    try {
        archivedTimeline.loadInstantDetailsInMemory(startTs, endTs);
        HoodieDefaultTimeline timelineRange = archivedTimeline.findInstantsInRange(startTs, endTs);
        if (includeExtraMetadata) {
            return printCommitsWithMetadata(timelineRange, limit, sortByField, descending, headerOnly, exportTableName);
        } else {
            return printCommits(timelineRange, limit, sortByField, descending, headerOnly, exportTableName);
        }
    } finally {
        // clear the instant details from memory after printing to reduce usage
        archivedTimeline.clearInstantDetailsFromMemory(startTs, endTs);
    }
}