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

Example 16 with HoodieActiveTimeline

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

the class ITTestSavepointsCommand method testSavepoint.

/**
 * Test case of command 'savepoint create'.
 */
@Test
public void testSavepoint() {
    // generate four savepoints
    for (int i = 100; i < 104; i++) {
        String instantTime = String.valueOf(i);
        HoodieTestDataGenerator.createCommitFile(tablePath, instantTime, jsc.hadoopConfiguration());
    }
    String savepoint = "102";
    CommandResult cr = getShell().executeCommand(String.format("savepoint create --commit %s --sparkMaster %s", savepoint, "local"));
    assertAll("Command run failed", () -> assertTrue(cr.isSuccess()), () -> assertEquals(String.format("The commit \"%s\" has been savepointed.", savepoint), cr.getResult().toString()));
    // there is 1 savepoint instant
    HoodieActiveTimeline timeline = HoodieCLI.getTableMetaClient().getActiveTimeline();
    assertEquals(1, timeline.getSavePointTimeline().countInstants());
}

Example 17 with HoodieActiveTimeline

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

the class ITTestCompactionCommand method prepareScheduleCompaction.

private String prepareScheduleCompaction() {
    // generate requested compaction
    CommandResult cr = getShell().executeCommand(String.format("compaction schedule --hoodieConfigs hoodie.compact.inline.max.delta.commits=1 --sparkMaster %s", "local"));
    assertTrue(cr.isSuccess());
    // get compaction instance
    HoodieActiveTimeline timeline = HoodieCLI.getTableMetaClient().getActiveTimeline();
    Option<String> instance = timeline.filterPendingCompactionTimeline().firstInstant().map(HoodieInstant::getTimestamp);
    assertTrue(instance.isPresent(), "Must have pending compaction.");
    return instance.get();
}

Example 18 with HoodieActiveTimeline

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

the class TestHoodieTableFileSystemView method testReplaceFileIdIsExcludedInView.

@Test
public void testReplaceFileIdIsExcludedInView() throws IOException {
    String partitionPath1 = "2020/06/27";
    String partitionPath2 = "2020/07/14";
    new File(basePath + "/" + partitionPath1).mkdirs();
    new File(basePath + "/" + partitionPath2).mkdirs();
    // create 2 fileId in partition1 - fileId1 is replaced later on.
    String fileId1 = UUID.randomUUID().toString();
    String fileId2 = UUID.randomUUID().toString();
    // create 2 fileId in partition2 - fileId3, fileId4 is replaced later on.
    String fileId3 = UUID.randomUUID().toString();
    String fileId4 = UUID.randomUUID().toString();
    assertFalse(roView.getLatestBaseFiles(partitionPath1).anyMatch(dfile -> dfile.getFileId().equals(fileId1) || dfile.getFileId().equals(fileId2)), "No commit, should not find any data file");
    assertFalse(roView.getLatestBaseFiles(partitionPath2).anyMatch(dfile -> dfile.getFileId().equals(fileId3) || dfile.getFileId().equals(fileId4)), "No commit, should not find any data file");
    // Only one commit
    String commitTime1 = "1";
    String fileName1 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId1);
    String fileName2 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId2);
    String fileName3 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId3);
    String fileName4 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId4);
    new File(basePath + "/" + partitionPath1 + "/" + fileName1).createNewFile();
    new File(basePath + "/" + partitionPath1 + "/" + fileName2).createNewFile();
    new File(basePath + "/" + partitionPath2 + "/" + fileName3).createNewFile();
    new File(basePath + "/" + partitionPath2 + "/" + fileName4).createNewFile();
    Map<String, List<String>> partitionToReplaceFileIds = new HashMap<>();
    List<String> replacedFileIdsP1 = new ArrayList<>();
    replacedFileIdsP1.add(fileId1);
    partitionToReplaceFileIds.put(partitionPath1, replacedFileIdsP1);
    List<String> replacedFileIdsP2 = new ArrayList<>();
    replacedFileIdsP2.add(fileId3);
    replacedFileIdsP2.add(fileId4);
    partitionToReplaceFileIds.put(partitionPath2, replacedFileIdsP2);
    HoodieCommitMetadata commitMetadata = CommitUtils.buildMetadata(Collections.emptyList(), partitionToReplaceFileIds, Option.empty(), WriteOperationType.INSERT_OVERWRITE, "", HoodieTimeline.REPLACE_COMMIT_ACTION);
    HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
    HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.REPLACE_COMMIT_ACTION, commitTime1);
    saveAsComplete(commitTimeline, instant1, Option.of(commitMetadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    refreshFsView();
    assertEquals(0, roView.getLatestBaseFiles(partitionPath1).filter(dfile -> dfile.getFileId().equals(fileId1)).count());
    assertEquals(fileName2, roView.getLatestBaseFiles(partitionPath1).filter(dfile -> dfile.getFileId().equals(fileId2)).findFirst().get().getFileName());
    assertEquals(0, roView.getLatestBaseFiles(partitionPath2).filter(dfile -> dfile.getFileId().equals(fileId3)).count());
    assertEquals(0, roView.getLatestBaseFiles(partitionPath2).filter(dfile -> dfile.getFileId().equals(fileId4)).count());
    // ensure replacedFileGroupsBefore works with all instants
    List<HoodieFileGroup> replacedOnInstant1 = fsView.getReplacedFileGroupsBeforeOrOn("0", partitionPath1).collect(Collectors.toList());
    assertEquals(0, replacedOnInstant1.size());
    List<HoodieFileGroup> allReplaced = fsView.getReplacedFileGroupsBeforeOrOn("2", partitionPath1).collect(Collectors.toList());
    allReplaced.addAll(fsView.getReplacedFileGroupsBeforeOrOn("2", partitionPath2).collect(Collectors.toList()));
    assertEquals(3, allReplaced.size());
    Set<String> allReplacedFileIds = allReplaced.stream().map(fg -> fg.getFileGroupId().getFileId()).collect(Collectors.toSet());
    Set<String> actualReplacedFileIds = Stream.of(fileId1, fileId3, fileId4).collect(Collectors.toSet());
    assertEquals(actualReplacedFileIds, allReplacedFileIds);
}

Example 19 with HoodieActiveTimeline

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

the class TestHoodieTableFileSystemView method testGetLatestDataFilesForFileId.

@Test
public void testGetLatestDataFilesForFileId() throws IOException {
    String partitionPath = "2016/05/01";
    new File(basePath + "/" + partitionPath).mkdirs();
    String fileId = UUID.randomUUID().toString();
    assertFalse(roView.getLatestBaseFiles(partitionPath).anyMatch(dfile -> dfile.getFileId().equals(fileId)), "No commit, should not find any data file");
    // Only one commit, but is not safe
    String commitTime1 = "1";
    String fileName1 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId);
    new File(basePath + "/" + partitionPath + "/" + fileName1).createNewFile();
    refreshFsView();
    assertFalse(roView.getLatestBaseFiles(partitionPath).anyMatch(dfile -> dfile.getFileId().equals(fileId)), "No commit, should not find any data file");
    // Make this commit safe
    HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
    HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, commitTime1);
    saveAsComplete(commitTimeline, instant1, Option.empty());
    refreshFsView();
    assertEquals(fileName1, roView.getLatestBaseFiles(partitionPath).filter(dfile -> dfile.getFileId().equals(fileId)).findFirst().get().getFileName());
    // Do another commit, but not safe
    String commitTime2 = "2";
    String fileName2 = FSUtils.makeDataFileName(commitTime2, TEST_WRITE_TOKEN, fileId);
    new File(basePath + "/" + partitionPath + "/" + fileName2).createNewFile();
    refreshFsView();
    assertEquals(fileName1, roView.getLatestBaseFiles(partitionPath).filter(dfile -> dfile.getFileId().equals(fileId)).findFirst().get().getFileName());
    // Make it safe
    HoodieInstant instant2 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, commitTime2);
    saveAsComplete(commitTimeline, instant2, Option.empty());
    refreshFsView();
    assertEquals(fileName2, roView.getLatestBaseFiles(partitionPath).filter(dfile -> dfile.getFileId().equals(fileId)).findFirst().get().getFileName());
}

Example 20 with HoodieActiveTimeline

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

the class TestHoodieTableFileSystemView method testPendingClusteringOperations.

@Test
public void testPendingClusteringOperations() throws IOException {
    String partitionPath1 = "2020/06/27";
    new File(basePath + "/" + partitionPath1).mkdirs();
    // create 2 fileId in partition1 - fileId1 is replaced later on.
    String fileId1 = UUID.randomUUID().toString();
    String fileId2 = UUID.randomUUID().toString();
    String fileId3 = UUID.randomUUID().toString();
    assertFalse(roView.getLatestBaseFiles(partitionPath1).anyMatch(dfile -> dfile.getFileId().equals(fileId1) || dfile.getFileId().equals(fileId2) || dfile.getFileId().equals(fileId3)), "No commit, should not find any data file");
    // Only one commit
    String commitTime1 = "1";
    String fileName1 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId1);
    String fileName2 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId2);
    String fileName3 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId3);
    new File(basePath + "/" + partitionPath1 + "/" + fileName1).createNewFile();
    new File(basePath + "/" + partitionPath1 + "/" + fileName2).createNewFile();
    new File(basePath + "/" + partitionPath1 + "/" + fileName3).createNewFile();
    HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
    HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, commitTime1);
    saveAsComplete(commitTimeline, instant1, Option.empty());
    refreshFsView();
    assertEquals(1, roView.getLatestBaseFiles(partitionPath1).filter(dfile -> dfile.getFileId().equals(fileId1)).count());
    assertEquals(1, roView.getLatestBaseFiles(partitionPath1).filter(dfile -> dfile.getFileId().equals(fileId2)).count());
    assertEquals(1, roView.getLatestBaseFiles(partitionPath1).filter(dfile -> dfile.getFileId().equals(fileId3)).count());
    List<FileSlice>[] fileSliceGroups = new List[] { Collections.singletonList(fsView.getLatestFileSlice(partitionPath1, fileId1).get()), Collections.singletonList(fsView.getLatestFileSlice(partitionPath1, fileId2).get()) };
    // create pending clustering operation - fileId1, fileId2 are being clustered in different groups
    HoodieClusteringPlan plan = ClusteringUtils.createClusteringPlan("strategy", new HashMap<>(), fileSliceGroups, Collections.emptyMap());
    String clusterTime = "2";
    HoodieInstant instant2 = new HoodieInstant(State.REQUESTED, HoodieTimeline.REPLACE_COMMIT_ACTION, clusterTime);
    HoodieRequestedReplaceMetadata requestedReplaceMetadata = HoodieRequestedReplaceMetadata.newBuilder().setClusteringPlan(plan).setOperationType(WriteOperationType.CLUSTER.name()).build();
    metaClient.getActiveTimeline().saveToPendingReplaceCommit(instant2, TimelineMetadataUtils.serializeRequestedReplaceMetadata(requestedReplaceMetadata));
    // make sure view doesnt include fileId1
    refreshFsView();
    Set<String> fileIds = fsView.getFileGroupsInPendingClustering().map(e -> e.getLeft().getFileId()).collect(Collectors.toSet());
    assertTrue(fileIds.contains(fileId1));
    assertTrue(fileIds.contains(fileId2));
    assertFalse(fileIds.contains(fileId3));
}