Examples with FileSlice org.apache.hudi.common.model.FileSlice used on opensource projects

Example 76 with FileSlice

use of org.apache.hudi.common.model.FileSlice in project hudi by apache.

the class JavaSizeBasedClusteringPlanStrategy method buildClusteringGroupsForPartition.

@Override
protected Stream<HoodieClusteringGroup> buildClusteringGroupsForPartition(String partitionPath, List<FileSlice> fileSlices) {
    List<Pair<List<FileSlice>, Integer>> fileSliceGroups = new ArrayList<>();
    List<FileSlice> currentGroup = new ArrayList<>();
    long totalSizeSoFar = 0;
    HoodieWriteConfig writeConfig = getWriteConfig();
    for (FileSlice currentSlice : fileSlices) {
        // assume each filegroup size is ~= parquet.max.file.size
        totalSizeSoFar += currentSlice.getBaseFile().isPresent() ? currentSlice.getBaseFile().get().getFileSize() : writeConfig.getParquetMaxFileSize();
        // check if max size is reached and create new group, if needed.
        if (totalSizeSoFar >= writeConfig.getClusteringMaxBytesInGroup() && !currentGroup.isEmpty()) {
            int numOutputGroups = getNumberOfOutputFileGroups(totalSizeSoFar, writeConfig.getClusteringTargetFileMaxBytes());
            LOG.info("Adding one clustering group " + totalSizeSoFar + " max bytes: " + writeConfig.getClusteringMaxBytesInGroup() + " num input slices: " + currentGroup.size() + " output groups: " + numOutputGroups);
            fileSliceGroups.add(Pair.of(currentGroup, numOutputGroups));
            currentGroup = new ArrayList<>();
            totalSizeSoFar = 0;
        }
        currentGroup.add(currentSlice);
        // reset to the size of current slice, otherwise the number of output file group will become 0 even though current slice is present.
        if (totalSizeSoFar == 0) {
            totalSizeSoFar += currentSlice.getBaseFile().isPresent() ? currentSlice.getBaseFile().get().getFileSize() : writeConfig.getParquetMaxFileSize();
        }
    }
    if (!currentGroup.isEmpty()) {
        int numOutputGroups = getNumberOfOutputFileGroups(totalSizeSoFar, writeConfig.getClusteringTargetFileMaxBytes());
        LOG.info("Adding final clustering group " + totalSizeSoFar + " max bytes: " + writeConfig.getClusteringMaxBytesInGroup() + " num input slices: " + currentGroup.size() + " output groups: " + numOutputGroups);
        fileSliceGroups.add(Pair.of(currentGroup, numOutputGroups));
    }
    return fileSliceGroups.stream().map(fileSliceGroup -> HoodieClusteringGroup.newBuilder().setSlices(getFileSliceInfo(fileSliceGroup.getLeft())).setNumOutputFileGroups(fileSliceGroup.getRight()).setMetrics(buildMetrics(fileSliceGroup.getLeft())).build());
}

Example 77 with FileSlice

use of org.apache.hudi.common.model.FileSlice in project hudi by apache.

the class TestHoodieTableFileSystemView method testPendingCompactionWithDuplicateFileIdsAcrossPartitions.

@Test
public void testPendingCompactionWithDuplicateFileIdsAcrossPartitions() throws Exception {
    // Put some files in the partition
    String partitionPath1 = "2016/05/01";
    String partitionPath2 = "2016/05/02";
    String partitionPath3 = "2016/05/03";
    String fullPartitionPath1 = basePath + "/" + partitionPath1 + "/";
    new File(fullPartitionPath1).mkdirs();
    String fullPartitionPath2 = basePath + "/" + partitionPath2 + "/";
    new File(fullPartitionPath2).mkdirs();
    String fullPartitionPath3 = basePath + "/" + partitionPath3 + "/";
    new File(fullPartitionPath3).mkdirs();
    String instantTime1 = "1";
    String deltaInstantTime1 = "2";
    String deltaInstantTime2 = "3";
    String fileId = UUID.randomUUID().toString();
    String dataFileName = FSUtils.makeDataFileName(instantTime1, TEST_WRITE_TOKEN, fileId);
    new File(fullPartitionPath1 + dataFileName).createNewFile();
    String fileName1 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, instantTime1, 0, TEST_WRITE_TOKEN);
    new File(fullPartitionPath1 + fileName1).createNewFile();
    new File(fullPartitionPath2 + FSUtils.makeDataFileName(instantTime1, TEST_WRITE_TOKEN, fileId)).createNewFile();
    new File(fullPartitionPath2 + fileName1).createNewFile();
    new File(fullPartitionPath3 + FSUtils.makeDataFileName(instantTime1, TEST_WRITE_TOKEN, fileId)).createNewFile();
    new File(fullPartitionPath3 + fileName1).createNewFile();
    HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
    HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, instantTime1);
    HoodieInstant deltaInstant2 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime1);
    HoodieInstant deltaInstant3 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime2);
    saveAsComplete(commitTimeline, instant1, Option.empty());
    saveAsComplete(commitTimeline, deltaInstant2, Option.empty());
    saveAsComplete(commitTimeline, deltaInstant3, Option.empty());
    // Now we list all partitions
    FileStatus[] statuses = metaClient.getFs().listStatus(new Path[] { new Path(fullPartitionPath1), new Path(fullPartitionPath2), new Path(fullPartitionPath3) });
    assertEquals(6, statuses.length);
    refreshFsView();
    Arrays.asList(partitionPath1, partitionPath2, partitionPath3).forEach(p -> fsView.getAllFileGroups(p).count());
    List<HoodieFileGroup> groups = Stream.of(partitionPath1, partitionPath2, partitionPath3).flatMap(p -> fsView.getAllFileGroups(p)).collect(Collectors.toList());
    assertEquals(3, groups.size(), "Expected number of file-groups");
    assertEquals(3, groups.stream().map(HoodieFileGroup::getPartitionPath).collect(Collectors.toSet()).size(), "Partitions must be different for file-groups");
    Set<String> fileIds = groups.stream().map(HoodieFileGroup::getFileGroupId).map(HoodieFileGroupId::getFileId).collect(Collectors.toSet());
    assertEquals(1, fileIds.size(), "File Id must be same");
    assertTrue(fileIds.contains(fileId), "Expected FileId");
    // Setup Pending compaction for all of these fileIds.
    List<Pair<String, FileSlice>> partitionFileSlicesPairs = new ArrayList<>();
    List<FileSlice> fileSlices = rtView.getLatestFileSlices(partitionPath1).collect(Collectors.toList());
    partitionFileSlicesPairs.add(Pair.of(partitionPath1, fileSlices.get(0)));
    fileSlices = rtView.getLatestFileSlices(partitionPath2).collect(Collectors.toList());
    partitionFileSlicesPairs.add(Pair.of(partitionPath2, fileSlices.get(0)));
    fileSlices = rtView.getLatestFileSlices(partitionPath3).collect(Collectors.toList());
    partitionFileSlicesPairs.add(Pair.of(partitionPath3, fileSlices.get(0)));
    String compactionRequestedTime = "2";
    String compactDataFileName = FSUtils.makeDataFileName(compactionRequestedTime, TEST_WRITE_TOKEN, fileId);
    HoodieCompactionPlan compactionPlan = CompactionUtils.buildFromFileSlices(partitionFileSlicesPairs, Option.empty(), Option.empty());
    // Create a Data-file for some of the partitions but this should be skipped by view
    new File(basePath + "/" + partitionPath1 + "/" + compactDataFileName).createNewFile();
    new File(basePath + "/" + partitionPath2 + "/" + compactDataFileName).createNewFile();
    HoodieInstant compactionInstant = new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
    HoodieInstant requested = HoodieTimeline.getCompactionRequestedInstant(compactionInstant.getTimestamp());
    metaClient.getActiveTimeline().saveToCompactionRequested(requested, TimelineMetadataUtils.serializeCompactionPlan(compactionPlan));
    metaClient.getActiveTimeline().transitionCompactionRequestedToInflight(requested);
    // Fake delta-ingestion after compaction-requested
    String deltaInstantTime4 = "4";
    String deltaInstantTime5 = "6";
    String fileName3 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, compactionRequestedTime, 0, TEST_WRITE_TOKEN);
    String fileName4 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, compactionRequestedTime, 1, TEST_WRITE_TOKEN);
    new File(basePath + "/" + partitionPath1 + "/" + fileName3).createNewFile();
    new File(basePath + "/" + partitionPath1 + "/" + fileName4).createNewFile();
    new File(basePath + "/" + partitionPath2 + "/" + fileName3).createNewFile();
    new File(basePath + "/" + partitionPath2 + "/" + fileName4).createNewFile();
    new File(basePath + "/" + partitionPath3 + "/" + fileName3).createNewFile();
    new File(basePath + "/" + partitionPath3 + "/" + fileName4).createNewFile();
    HoodieInstant deltaInstant4 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime4);
    HoodieInstant deltaInstant5 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime5);
    saveAsComplete(commitTimeline, deltaInstant4, Option.empty());
    saveAsComplete(commitTimeline, deltaInstant5, Option.empty());
    refreshFsView();
    // Test Data Files
    List<HoodieBaseFile> dataFiles = roView.getAllBaseFiles(partitionPath1).collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "One data-file is expected as there is only one file-group");
    assertEquals("1", dataFiles.get(0).getCommitTime(), "Expect only valid commit");
    dataFiles = roView.getAllBaseFiles(partitionPath2).collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "One data-file is expected as there is only one file-group");
    assertEquals("1", dataFiles.get(0).getCommitTime(), "Expect only valid commit");
    // Merge API Tests
    Arrays.asList(partitionPath1, partitionPath2, partitionPath3).forEach(partitionPath -> {
        List<FileSlice> fileSliceList = rtView.getLatestMergedFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
        assertEquals(1, fileSliceList.size(), "Expect file-slice to be merged");
        FileSlice fileSlice = fileSliceList.get(0);
        assertEquals(fileId, fileSlice.getFileId());
        assertEquals(dataFileName, fileSlice.getBaseFile().get().getFileName(), "Data file must be present");
        assertEquals(instantTime1, fileSlice.getBaseInstantTime(), "Base Instant of penultimate file-slice must be base instant");
        List<HoodieLogFile> logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
        assertEquals(3, logFiles.size(), "Log files must include those after compaction request");
        assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
        assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
        assertEquals(fileName1, logFiles.get(2).getFileName(), "Log File Order check");
        fileSliceList = rtView.getLatestFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5, true).collect(Collectors.toList());
        assertEquals(1, fileSliceList.size(), "Expect only one file-id");
        fileSlice = fileSliceList.get(0);
        assertEquals(fileId, fileSlice.getFileId());
        assertFalse(fileSlice.getBaseFile().isPresent(), "No data-file expected in latest file-slice");
        assertEquals(compactionRequestedTime, fileSlice.getBaseInstantTime(), "Compaction requested instant must be base instant");
        logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
        assertEquals(2, logFiles.size(), "Log files must include only those after compaction request");
        assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
        assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
        // Check getLatestFileSlicesBeforeOrOn excluding fileIds in pending compaction
        fileSliceList = rtView.getLatestFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5, false).collect(Collectors.toList());
        assertEquals(0, fileSliceList.size(), "Expect empty list as file-id is in pending compaction");
    });
    assertEquals(3, fsView.getPendingCompactionOperations().count());
    Set<String> partitionsInCompaction = fsView.getPendingCompactionOperations().map(Pair::getValue).map(CompactionOperation::getPartitionPath).collect(Collectors.toSet());
    assertEquals(3, partitionsInCompaction.size());
    assertTrue(partitionsInCompaction.contains(partitionPath1));
    assertTrue(partitionsInCompaction.contains(partitionPath2));
    assertTrue(partitionsInCompaction.contains(partitionPath3));
    Set<String> fileIdsInCompaction = fsView.getPendingCompactionOperations().map(Pair::getValue).map(CompactionOperation::getFileId).collect(Collectors.toSet());
    assertEquals(1, fileIdsInCompaction.size());
    assertTrue(fileIdsInCompaction.contains(fileId));
}

Example 78 with FileSlice

use of org.apache.hudi.common.model.FileSlice in project hudi by apache.

the class TestHoodieTableFileSystemView method testViewForFileSlicesWithAsyncCompaction.

/**
 * Helper method to test Views in the presence of concurrent compaction.
 *
 * @param skipCreatingDataFile if set, first File Slice will not have data-file set. This would simulate inserts going
 *        directly to log files
 * @param isCompactionInFlight if set, compaction was inflight (running) when view was tested first time, otherwise
 *        compaction was in requested state
 * @param expTotalFileSlices Total number of file-slices across file-groups in the partition path
 * @param expTotalDataFiles Total number of data-files across file-groups in the partition path
 * @param includeInvalidAndInflight Whether view includes inflight and invalid file-groups.
 * @param testBootstrap enable Bootstrap and test
 * @throws Exception -
 */
protected void testViewForFileSlicesWithAsyncCompaction(boolean skipCreatingDataFile, boolean isCompactionInFlight, int expTotalFileSlices, int expTotalDataFiles, boolean includeInvalidAndInflight, boolean testBootstrap) throws Exception {
    if (testBootstrap) {
        metaClient = HoodieTestUtils.init(tempDir.toAbsolutePath().toString(), getTableType(), BOOTSTRAP_SOURCE_PATH, testBootstrap);
    }
    String partitionPath = "2016/05/01";
    new File(basePath + "/" + partitionPath).mkdirs();
    String fileId = UUID.randomUUID().toString();
    String srcName = "part_0000" + metaClient.getTableConfig().getBaseFileFormat().getFileExtension();
    HoodieFileStatus srcFileStatus = HoodieFileStatus.newBuilder().setPath(HoodiePath.newBuilder().setUri(BOOTSTRAP_SOURCE_PATH + partitionPath + "/" + srcName).build()).setLength(256 * 1024 * 1024L).setAccessTime(new Date().getTime()).setModificationTime(new Date().getTime() + 99999).setBlockReplication(2).setOwner("hudi").setGroup("hudi").setBlockSize(128 * 1024 * 1024L).setPermission(HoodieFSPermission.newBuilder().setUserAction(FsAction.ALL.name()).setGroupAction(FsAction.READ.name()).setOtherAction(FsAction.NONE.name()).setStickyBit(true).build()).build();
    // if skipCreatingDataFile, then instantTime1 below acts like delta-commit, otherwise it is base-commit
    String instantTime1 = testBootstrap && !skipCreatingDataFile ? HoodieTimeline.METADATA_BOOTSTRAP_INSTANT_TS : "1";
    String deltaInstantTime1 = "2";
    String deltaInstantTime2 = "3";
    String dataFileName = null;
    if (!skipCreatingDataFile) {
        dataFileName = FSUtils.makeDataFileName(instantTime1, TEST_WRITE_TOKEN, fileId);
        new File(basePath + "/" + partitionPath + "/" + dataFileName).createNewFile();
    }
    String fileName1 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, instantTime1, 0, TEST_WRITE_TOKEN);
    String fileName2 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, instantTime1, 1, TEST_WRITE_TOKEN);
    new File(basePath + "/" + partitionPath + "/" + fileName1).createNewFile();
    new File(basePath + "/" + partitionPath + "/" + fileName2).createNewFile();
    HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
    HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, instantTime1);
    HoodieInstant deltaInstant2 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime1);
    HoodieInstant deltaInstant3 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime2);
    if (testBootstrap && !skipCreatingDataFile) {
        try (IndexWriter writer = new HFileBootstrapIndex(metaClient).createWriter(BOOTSTRAP_SOURCE_PATH)) {
            writer.begin();
            BootstrapFileMapping mapping = new BootstrapFileMapping(BOOTSTRAP_SOURCE_PATH, partitionPath, partitionPath, srcFileStatus, fileId);
            List<BootstrapFileMapping> b = new ArrayList<>();
            b.add(mapping);
            writer.appendNextPartition(partitionPath, b);
            writer.finish();
        }
    }
    saveAsComplete(commitTimeline, instant1, Option.empty());
    saveAsComplete(commitTimeline, deltaInstant2, Option.empty());
    saveAsComplete(commitTimeline, deltaInstant3, Option.empty());
    refreshFsView();
    List<FileSlice> fileSlices = rtView.getLatestFileSlices(partitionPath).collect(Collectors.toList());
    assertEquals(1, fileSlices.size());
    FileSlice fileSlice = fileSlices.get(0);
    assertEquals(instantTime1, fileSlice.getBaseInstantTime());
    if (!skipCreatingDataFile) {
        assertTrue(fileSlice.getBaseFile().isPresent());
        checkExternalFile(srcFileStatus, fileSlice.getBaseFile().get().getBootstrapBaseFile(), testBootstrap);
    }
    String compactionRequestedTime = "4";
    String compactDataFileName = FSUtils.makeDataFileName(compactionRequestedTime, TEST_WRITE_TOKEN, fileId);
    List<Pair<String, FileSlice>> partitionFileSlicesPairs = new ArrayList<>();
    partitionFileSlicesPairs.add(Pair.of(partitionPath, fileSlices.get(0)));
    HoodieCompactionPlan compactionPlan = CompactionUtils.buildFromFileSlices(partitionFileSlicesPairs, Option.empty(), Option.empty());
    HoodieInstant compactionInstant;
    if (isCompactionInFlight) {
        // Create a Data-file but this should be skipped by view
        new File(basePath + "/" + partitionPath + "/" + compactDataFileName).createNewFile();
        compactionInstant = new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
        HoodieInstant requested = HoodieTimeline.getCompactionRequestedInstant(compactionInstant.getTimestamp());
        commitTimeline.saveToCompactionRequested(requested, TimelineMetadataUtils.serializeCompactionPlan(compactionPlan));
        commitTimeline.transitionCompactionRequestedToInflight(requested);
    } else {
        compactionInstant = new HoodieInstant(State.REQUESTED, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
        commitTimeline.saveToCompactionRequested(compactionInstant, TimelineMetadataUtils.serializeCompactionPlan(compactionPlan));
    }
    // View immediately after scheduling compaction
    refreshFsView();
    List<FileSlice> slices = rtView.getLatestFileSlices(partitionPath).collect(Collectors.toList());
    assertEquals(1, slices.size(), "Expected latest file-slices");
    assertEquals(compactionRequestedTime, slices.get(0).getBaseInstantTime(), "Base-Instant must be compaction Instant");
    assertFalse(slices.get(0).getBaseFile().isPresent(), "Latest File Slice must not have data-file");
    assertEquals(0, slices.get(0).getLogFiles().count(), "Latest File Slice must not have any log-files");
    // Fake delta-ingestion after compaction-requested
    String deltaInstantTime4 = "5";
    String deltaInstantTime5 = "6";
    List<String> allInstantTimes = Arrays.asList(instantTime1, deltaInstantTime1, deltaInstantTime2, compactionRequestedTime, deltaInstantTime4, deltaInstantTime5);
    String fileName3 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, compactionRequestedTime, 0, TEST_WRITE_TOKEN);
    String fileName4 = FSUtils.makeLogFileName(fileId, HoodieLogFile.DELTA_EXTENSION, compactionRequestedTime, 1, TEST_WRITE_TOKEN);
    new File(basePath + "/" + partitionPath + "/" + fileName3).createNewFile();
    new File(basePath + "/" + partitionPath + "/" + fileName4).createNewFile();
    HoodieInstant deltaInstant4 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime4);
    HoodieInstant deltaInstant5 = new HoodieInstant(true, HoodieTimeline.DELTA_COMMIT_ACTION, deltaInstantTime5);
    saveAsComplete(commitTimeline, deltaInstant4, Option.empty());
    saveAsComplete(commitTimeline, deltaInstant5, Option.empty());
    refreshFsView();
    List<HoodieBaseFile> dataFiles = roView.getAllBaseFiles(partitionPath).collect(Collectors.toList());
    if (skipCreatingDataFile) {
        assertTrue(dataFiles.isEmpty(), "No data file expected");
    } else {
        assertEquals(1, dataFiles.size(), "One data-file is expected as there is only one file-group");
        assertEquals(dataFileName, dataFiles.get(0).getFileName(), "Expect only valid data-file");
    }
    // Merge API Tests
    List<FileSlice> fileSliceList = rtView.getLatestMergedFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
    assertEquals(1, fileSliceList.size(), "Expect file-slice to be merged");
    fileSlice = fileSliceList.get(0);
    assertEquals(fileId, fileSlice.getFileId());
    if (!skipCreatingDataFile) {
        assertEquals(dataFileName, fileSlice.getBaseFile().get().getFileName(), "Data file must be present");
        checkExternalFile(srcFileStatus, fileSlice.getBaseFile().get().getBootstrapBaseFile(), testBootstrap);
    } else {
        assertFalse(fileSlice.getBaseFile().isPresent(), "No data-file expected as it was not created");
    }
    assertEquals(instantTime1, fileSlice.getBaseInstantTime(), "Base Instant of penultimate file-slice must be base instant");
    List<HoodieLogFile> logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
    assertEquals(4, logFiles.size(), "Log files must include those after compaction request");
    assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
    assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
    assertEquals(fileName2, logFiles.get(2).getFileName(), "Log File Order check");
    assertEquals(fileName1, logFiles.get(3).getFileName(), "Log File Order check");
    fileSliceList = rtView.getLatestFileSlicesBeforeOrOn(partitionPath, deltaInstantTime5, true).collect(Collectors.toList());
    assertEquals(1, fileSliceList.size(), "Expect only one file-id");
    fileSlice = fileSliceList.get(0);
    assertEquals(fileId, fileSlice.getFileId());
    assertFalse(fileSlice.getBaseFile().isPresent(), "No data-file expected in latest file-slice");
    assertEquals(compactionRequestedTime, fileSlice.getBaseInstantTime(), "Compaction requested instant must be base instant");
    logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
    assertEquals(2, logFiles.size(), "Log files must include only those after compaction request");
    assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
    assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
    // Data Files API tests
    dataFiles = roView.getLatestBaseFiles().collect(Collectors.toList());
    if (skipCreatingDataFile) {
        assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
    } else {
        assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
        dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
        checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
    }
    dataFiles = roView.getLatestBaseFiles(partitionPath).collect(Collectors.toList());
    if (skipCreatingDataFile) {
        assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
    } else {
        assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
        dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
        checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
    }
    dataFiles = roView.getLatestBaseFilesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
    if (skipCreatingDataFile) {
        assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
    } else {
        assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
        dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
        checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
    }
    dataFiles = roView.getLatestBaseFilesInRange(allInstantTimes).collect(Collectors.toList());
    if (skipCreatingDataFile) {
        assertEquals(0, dataFiles.size(), "Expect no data file to be returned");
    } else {
        assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
        dataFiles.forEach(df -> assertEquals(df.getCommitTime(), instantTime1, "Expect data-file for instant 1 be returned"));
        checkExternalFile(srcFileStatus, dataFiles.get(0).getBootstrapBaseFile(), testBootstrap);
    }
    // Inflight/Orphan File-groups needs to be in the view
    // There is a data-file with this inflight file-id
    final String inflightFileId1 = UUID.randomUUID().toString();
    // There is a log-file with this inflight file-id
    final String inflightFileId2 = UUID.randomUUID().toString();
    // There is an orphan data file with this file-id
    final String orphanFileId1 = UUID.randomUUID().toString();
    // There is an orphan log data file with this file-id
    final String orphanFileId2 = UUID.randomUUID().toString();
    final String invalidInstantId = "INVALIDTIME";
    String inflightDeltaInstantTime = "7";
    String orphanDataFileName = FSUtils.makeDataFileName(invalidInstantId, TEST_WRITE_TOKEN, orphanFileId1);
    new File(basePath + "/" + partitionPath + "/" + orphanDataFileName).createNewFile();
    String orphanLogFileName = FSUtils.makeLogFileName(orphanFileId2, HoodieLogFile.DELTA_EXTENSION, invalidInstantId, 0, TEST_WRITE_TOKEN);
    new File(basePath + "/" + partitionPath + "/" + orphanLogFileName).createNewFile();
    String inflightDataFileName = FSUtils.makeDataFileName(inflightDeltaInstantTime, TEST_WRITE_TOKEN, inflightFileId1);
    new File(basePath + "/" + partitionPath + "/" + inflightDataFileName).createNewFile();
    String inflightLogFileName = FSUtils.makeLogFileName(inflightFileId2, HoodieLogFile.DELTA_EXTENSION, inflightDeltaInstantTime, 0, TEST_WRITE_TOKEN);
    new File(basePath + "/" + partitionPath + "/" + inflightLogFileName).createNewFile();
    // Mark instant as inflight
    commitTimeline.createNewInstant(new HoodieInstant(State.REQUESTED, HoodieTimeline.DELTA_COMMIT_ACTION, inflightDeltaInstantTime));
    commitTimeline.transitionRequestedToInflight(new HoodieInstant(State.REQUESTED, HoodieTimeline.DELTA_COMMIT_ACTION, inflightDeltaInstantTime), Option.empty());
    refreshFsView();
    List<FileSlice> allRawFileSlices = getAllRawFileSlices(partitionPath).collect(Collectors.toList());
    dataFiles = allRawFileSlices.stream().flatMap(slice -> {
        if (slice.getBaseFile().isPresent()) {
            return Stream.of(slice.getBaseFile().get());
        }
        return Stream.empty();
    }).collect(Collectors.toList());
    if (includeInvalidAndInflight) {
        assertEquals(2 + (isCompactionInFlight ? 1 : 0) + (skipCreatingDataFile ? 0 : 1), dataFiles.size(), "Inflight/Orphan data-file is also expected");
        Set<String> fileNames = dataFiles.stream().map(HoodieBaseFile::getFileName).collect(Collectors.toSet());
        assertTrue(fileNames.contains(orphanDataFileName), "Expect orphan data-file to be present");
        assertTrue(fileNames.contains(inflightDataFileName), "Expect inflight data-file to be present");
        if (!skipCreatingDataFile) {
            assertTrue(fileNames.contains(dataFileName), "Expect old committed data-file");
        }
        if (isCompactionInFlight) {
            assertTrue(fileNames.contains(compactDataFileName), "Expect inflight compacted data file to be present");
        }
        fileSliceList = getLatestRawFileSlices(partitionPath).collect(Collectors.toList());
        assertEquals(includeInvalidAndInflight ? 5 : 1, fileSliceList.size(), "Expect both inflight and orphan file-slice to be included");
        Map<String, FileSlice> fileSliceMap = fileSliceList.stream().collect(Collectors.toMap(FileSlice::getFileId, r -> r));
        FileSlice orphanFileSliceWithDataFile = fileSliceMap.get(orphanFileId1);
        FileSlice orphanFileSliceWithLogFile = fileSliceMap.get(orphanFileId2);
        FileSlice inflightFileSliceWithDataFile = fileSliceMap.get(inflightFileId1);
        FileSlice inflightFileSliceWithLogFile = fileSliceMap.get(inflightFileId2);
        assertEquals(invalidInstantId, orphanFileSliceWithDataFile.getBaseInstantTime(), "Orphan File Slice with data-file check base-commit");
        assertEquals(orphanDataFileName, orphanFileSliceWithDataFile.getBaseFile().get().getFileName(), "Orphan File Slice with data-file check data-file");
        assertEquals(0, orphanFileSliceWithDataFile.getLogFiles().count(), "Orphan File Slice with data-file check data-file");
        assertEquals(inflightDeltaInstantTime, inflightFileSliceWithDataFile.getBaseInstantTime(), "Inflight File Slice with data-file check base-commit");
        assertEquals(inflightDataFileName, inflightFileSliceWithDataFile.getBaseFile().get().getFileName(), "Inflight File Slice with data-file check data-file");
        assertEquals(0, inflightFileSliceWithDataFile.getLogFiles().count(), "Inflight File Slice with data-file check data-file");
        assertEquals(invalidInstantId, orphanFileSliceWithLogFile.getBaseInstantTime(), "Orphan File Slice with log-file check base-commit");
        assertFalse(orphanFileSliceWithLogFile.getBaseFile().isPresent(), "Orphan File Slice with log-file check data-file");
        logFiles = orphanFileSliceWithLogFile.getLogFiles().collect(Collectors.toList());
        assertEquals(1, logFiles.size(), "Orphan File Slice with log-file check data-file");
        assertEquals(orphanLogFileName, logFiles.get(0).getFileName(), "Orphan File Slice with log-file check data-file");
        assertEquals(inflightDeltaInstantTime, inflightFileSliceWithLogFile.getBaseInstantTime(), "Inflight File Slice with log-file check base-commit");
        assertFalse(inflightFileSliceWithLogFile.getBaseFile().isPresent(), "Inflight File Slice with log-file check data-file");
        logFiles = inflightFileSliceWithLogFile.getLogFiles().collect(Collectors.toList());
        assertEquals(1, logFiles.size(), "Inflight File Slice with log-file check data-file");
        assertEquals(inflightLogFileName, logFiles.get(0).getFileName(), "Inflight File Slice with log-file check data-file");
    }
    compactionInstant = new HoodieInstant(State.INFLIGHT, HoodieTimeline.COMPACTION_ACTION, compactionRequestedTime);
    // Now simulate Compaction completing - Check the view
    if (!isCompactionInFlight) {
        // For inflight compaction, we already create a data-file to test concurrent inflight case.
        // If we skipped creating data file corresponding to compaction commit, create it now
        new File(basePath + "/" + partitionPath + "/" + compactDataFileName).createNewFile();
        commitTimeline.createNewInstant(compactionInstant);
    }
    commitTimeline.saveAsComplete(compactionInstant, Option.empty());
    refreshFsView();
    // populate the cache
    roView.getAllBaseFiles(partitionPath);
    fileSliceList = rtView.getLatestFileSlices(partitionPath).collect(Collectors.toList());
    LOG.info("FILESLICE LIST=" + fileSliceList);
    dataFiles = fileSliceList.stream().map(FileSlice::getBaseFile).filter(Option::isPresent).map(Option::get).collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "Expect only one data-files in latest view as there is only one file-group");
    assertEquals(compactDataFileName, dataFiles.get(0).getFileName(), "Data Filename must match");
    assertEquals(1, fileSliceList.size(), "Only one latest file-slice in the partition");
    assertFalse(dataFiles.get(0).getBootstrapBaseFile().isPresent(), "No external data file must be present");
    fileSlice = fileSliceList.get(0);
    assertEquals(fileId, fileSlice.getFileId(), "Check file-Id is set correctly");
    assertEquals(compactDataFileName, fileSlice.getBaseFile().get().getFileName(), "Check data-filename is set correctly");
    assertEquals(compactionRequestedTime, fileSlice.getBaseInstantTime(), "Ensure base-instant is now compaction request instant");
    logFiles = fileSlice.getLogFiles().collect(Collectors.toList());
    assertEquals(2, logFiles.size(), "Only log-files after compaction request shows up");
    assertEquals(fileName4, logFiles.get(0).getFileName(), "Log File Order check");
    assertEquals(fileName3, logFiles.get(1).getFileName(), "Log File Order check");
    // Data Files API tests
    dataFiles = roView.getLatestBaseFiles().collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
    assertFalse(dataFiles.get(0).getBootstrapBaseFile().isPresent(), "No external data file must be present");
    dataFiles.forEach(df -> {
        assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
        assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
    });
    dataFiles = roView.getLatestBaseFiles(partitionPath).collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
    dataFiles.forEach(df -> {
        assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
        assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
    });
    dataFiles = roView.getLatestBaseFilesBeforeOrOn(partitionPath, deltaInstantTime5).collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
    dataFiles.forEach(df -> {
        assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
        assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
    });
    dataFiles = roView.getLatestBaseFilesInRange(allInstantTimes).collect(Collectors.toList());
    assertEquals(1, dataFiles.size(), "Expect only one data-file to be sent");
    dataFiles.forEach(df -> {
        assertEquals(df.getCommitTime(), compactionRequestedTime, "Expect data-file created by compaction be returned");
        assertFalse(df.getBootstrapBaseFile().isPresent(), "No external data file must be present");
    });
    assertEquals(expTotalFileSlices, rtView.getAllFileSlices(partitionPath).count(), "Total number of file-slices in partitions matches expected");
    assertEquals(expTotalDataFiles, roView.getAllBaseFiles(partitionPath).count(), "Total number of data-files in partitions matches expected");
    // file-groups includes inflight/invalid file-ids
    assertEquals(5, fsView.getAllFileGroups(partitionPath).count(), "Total number of file-groups in partitions matches expected");
}

Example 79 with FileSlice

use of org.apache.hudi.common.model.FileSlice in project hudi by apache.

the class TestHoodieTableFileSystemView method testStreamLatestVersions.

protected void testStreamLatestVersions(boolean isLatestFileSliceOnly) throws IOException {
    // Put some files in the partition
    String partitionPath = "2016/05/01";
    String fullPartitionPath = basePath + "/" + partitionPath;
    new File(fullPartitionPath).mkdirs();
    String commitTime1 = "1";
    String commitTime2 = "2";
    String commitTime3 = "3";
    String commitTime4 = "4";
    String fileId1 = UUID.randomUUID().toString();
    String fileId2 = UUID.randomUUID().toString();
    String fileId3 = UUID.randomUUID().toString();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId1)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeLogFileName(fileId1, HoodieLogFile.DELTA_EXTENSION, commitTime1, 0, TEST_WRITE_TOKEN)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime4, TEST_WRITE_TOKEN, fileId1)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeLogFileName(fileId1, HoodieLogFile.DELTA_EXTENSION, commitTime4, 0, TEST_WRITE_TOKEN)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId2)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime2, TEST_WRITE_TOKEN, fileId2)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeLogFileName(fileId2, HoodieLogFile.DELTA_EXTENSION, commitTime2, 0, TEST_WRITE_TOKEN)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId2)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId3)).createNewFile();
    new File(fullPartitionPath + "/" + FSUtils.makeDataFileName(commitTime4, TEST_WRITE_TOKEN, fileId3)).createNewFile();
    new File(basePath + "/.hoodie/" + commitTime1 + ".commit").createNewFile();
    new File(basePath + "/.hoodie/" + commitTime2 + ".commit").createNewFile();
    new File(basePath + "/.hoodie/" + commitTime3 + ".commit").createNewFile();
    new File(basePath + "/.hoodie/" + commitTime4 + ".commit").createNewFile();
    // Now we list the entire partition
    FileStatus[] statuses = metaClient.getFs().listStatus(new Path(fullPartitionPath));
    assertEquals(10, statuses.length);
    refreshFsView();
    fsView.getAllBaseFiles(partitionPath);
    List<HoodieFileGroup> fileGroups = fsView.getAllFileGroups(partitionPath).collect(Collectors.toList());
    assertEquals(3, fileGroups.size());
    for (HoodieFileGroup fileGroup : fileGroups) {
        List<FileSlice> slices = fileGroup.getAllFileSlices().collect(Collectors.toList());
        String fileId = fileGroup.getFileGroupId().getFileId();
        if (fileId.equals(fileId1)) {
            assertEquals(isLatestFileSliceOnly ? 1 : 2, slices.size());
            assertEquals(commitTime4, slices.get(0).getBaseInstantTime());
            if (!isLatestFileSliceOnly) {
                assertEquals(commitTime1, slices.get(1).getBaseInstantTime());
            }
        } else if (fileId.equals(fileId2)) {
            assertEquals(isLatestFileSliceOnly ? 1 : 3, slices.size());
            assertEquals(commitTime3, slices.get(0).getBaseInstantTime());
            if (!isLatestFileSliceOnly) {
                assertEquals(commitTime2, slices.get(1).getBaseInstantTime());
                assertEquals(commitTime1, slices.get(2).getBaseInstantTime());
            }
        } else if (fileId.equals(fileId3)) {
            assertEquals(isLatestFileSliceOnly ? 1 : 2, slices.size());
            assertEquals(commitTime4, slices.get(0).getBaseInstantTime());
            if (!isLatestFileSliceOnly) {
                assertEquals(commitTime3, slices.get(1).getBaseInstantTime());
            }
        }
    }
    List<HoodieBaseFile> statuses1 = roView.getLatestBaseFiles().collect(Collectors.toList());
    assertEquals(3, statuses1.size());
    Set<String> filenames = new HashSet<>();
    for (HoodieBaseFile status : statuses1) {
        filenames.add(status.getFileName());
    }
    assertTrue(filenames.contains(FSUtils.makeDataFileName(commitTime4, TEST_WRITE_TOKEN, fileId1)));
    assertTrue(filenames.contains(FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId2)));
    assertTrue(filenames.contains(FSUtils.makeDataFileName(commitTime4, TEST_WRITE_TOKEN, fileId3)));
}

Example 80 with FileSlice

use of org.apache.hudi.common.model.FileSlice in project hudi by apache.

the class TestHoodieTableFileSystemView method testStreamLatestVersionInRange.

protected void testStreamLatestVersionInRange(boolean isLatestFileSliceOnly) throws IOException {
    // Put some files in the partition
    String fullPartitionPath = basePath + "/2016/05/01/";
    new File(fullPartitionPath).mkdirs();
    String commitTime1 = "1";
    String commitTime2 = "2";
    String commitTime3 = "3";
    String commitTime4 = "4";
    String fileId1 = UUID.randomUUID().toString();
    String fileId2 = UUID.randomUUID().toString();
    String fileId3 = UUID.randomUUID().toString();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId1)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeLogFileName(fileId1, HoodieLogFile.DELTA_EXTENSION, commitTime1, 0, TEST_WRITE_TOKEN)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId1)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId2)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime2, TEST_WRITE_TOKEN, fileId2)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId2)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeLogFileName(fileId2, HoodieLogFile.DELTA_EXTENSION, commitTime3, 0, TEST_WRITE_TOKEN)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId3)).createNewFile();
    new File(fullPartitionPath + FSUtils.makeDataFileName(commitTime4, TEST_WRITE_TOKEN, fileId3)).createNewFile();
    new File(basePath + "/.hoodie/" + commitTime1 + ".commit").createNewFile();
    new File(basePath + "/.hoodie/" + commitTime2 + ".commit").createNewFile();
    new File(basePath + "/.hoodie/" + commitTime3 + ".commit").createNewFile();
    new File(basePath + "/.hoodie/" + commitTime4 + ".commit").createNewFile();
    // Now we list the entire partition
    FileStatus[] statuses = metaClient.getFs().listStatus(new Path(fullPartitionPath));
    assertEquals(9, statuses.length);
    refreshFsView();
    // Populate view for partition
    roView.getAllBaseFiles("2016/05/01/");
    List<HoodieBaseFile> dataFiles = roView.getLatestBaseFilesInRange(Arrays.asList(commitTime2, commitTime3)).collect(Collectors.toList());
    assertEquals(isLatestFileSliceOnly ? 2 : 3, dataFiles.size());
    Set<String> filenames = new HashSet<>();
    for (HoodieBaseFile status : dataFiles) {
        filenames.add(status.getFileName());
    }
    assertTrue(filenames.contains(FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId1)));
    assertTrue(filenames.contains(FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId2)));
    if (!isLatestFileSliceOnly) {
        assertTrue(filenames.contains(FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId3)));
    }
    List<FileSlice> slices = rtView.getLatestFileSliceInRange(Arrays.asList(commitTime3, commitTime4)).collect(Collectors.toList());
    assertEquals(3, slices.size());
    for (FileSlice slice : slices) {
        if (slice.getFileId().equals(fileId1)) {
            assertEquals(slice.getBaseInstantTime(), commitTime3);
            assertTrue(slice.getBaseFile().isPresent());
            assertEquals(slice.getLogFiles().count(), 0);
        } else if (slice.getFileId().equals(fileId2)) {
            assertEquals(slice.getBaseInstantTime(), commitTime3);
            assertTrue(slice.getBaseFile().isPresent());
            assertEquals(slice.getLogFiles().count(), 1);
        } else if (slice.getFileId().equals(fileId3)) {
            assertEquals(slice.getBaseInstantTime(), commitTime4);
            assertTrue(slice.getBaseFile().isPresent());
            assertEquals(slice.getLogFiles().count(), 0);
        }
    }
}