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

Example 36 with HoodieCommitMetadata

use of org.apache.hudi.common.model.HoodieCommitMetadata 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 37 with HoodieCommitMetadata

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

the class TestHoodieTableFileSystemView method testHoodieTableFileSystemViewWithPendingClustering.

/**
 * create hoodie table like
 * .
 * ├── .hoodie
 * │   ├── .aux
 * │   │   └── .bootstrap
 * │   │       ├── .fileids
 * │   │       └── .partitions
 * │   ├── .temp
 * │   ├── 1.commit
 * │   ├── 1.commit.requested
 * │   ├── 1.inflight
 * │   ├── 2.replacecommit
 * │   ├── 2.replacecommit.inflight
 * │   ├── 2.replacecommit.requested
 * │   ├── 3.commit
 * │   ├── 3.commit.requested
 * │   ├── 3.inflight
 * │   ├── archived
 * │   └── hoodie.properties
 * └── 2020
 *     └── 06
 *         └── 27
 *             ├── 5fe477d2-0150-46d4-833c-1e9cc8da9948_1-0-1_3.parquet
 *             ├── 7e3208c8-fdec-4254-9682-8fff1e51ee8d_1-0-1_2.parquet
 *             ├── e04b0e2d-1467-46b2-8ea6-f4fe950965a5_1-0-1_1.parquet
 *             └── f3936b66-b3db-4fc8-a6d0-b1a7559016e6_1-0-1_1.parquet
 *
 * First test fsView API with finished clustering:
 *  1. getLatestBaseFilesBeforeOrOn
 *  2. getBaseFileOn
 *  3. getLatestBaseFilesInRange
 *  4. getAllBaseFiles
 *  5. getLatestBaseFiles
 *
 * Then remove 2.replacecommit, 1.commit, 1.commit.requested, 1.inflight to simulate
 * pending clustering at the earliest position in the active timeline and test these APIs again.
 *
 * @throws IOException
 */
@Test
public void testHoodieTableFileSystemViewWithPendingClustering() throws IOException {
    List<String> latestBaseFilesBeforeOrOn;
    Option<HoodieBaseFile> baseFileOn;
    List<String> latestBaseFilesInRange;
    List<String> allBaseFiles;
    List<String> latestBaseFiles;
    List<String> latestBaseFilesPerPartition;
    String partitionPath = "2020/06/27";
    new File(basePath + "/" + partitionPath).mkdirs();
    HoodieActiveTimeline commitTimeline = metaClient.getActiveTimeline();
    // will create 5 fileId in partition.
    // fileId1 and fileId2 will be replaced by fileID3
    // fileId4 and fileId5 will be committed after clustering finished.
    String fileId1 = UUID.randomUUID().toString();
    String fileId2 = UUID.randomUUID().toString();
    String fileId3 = UUID.randomUUID().toString();
    String fileId4 = UUID.randomUUID().toString();
    String fileId5 = UUID.randomUUID().toString();
    assertFalse(roView.getLatestBaseFiles(partitionPath).anyMatch(dfile -> dfile.getFileId().equals(fileId1) || dfile.getFileId().equals(fileId2) || dfile.getFileId().equals(fileId3) || dfile.getFileId().equals(fileId4) || dfile.getFileId().equals(fileId5)), "No commit, should not find any data file");
    // first insert commit
    String commitTime1 = "1";
    String fileName1 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId1);
    String fileName2 = FSUtils.makeDataFileName(commitTime1, TEST_WRITE_TOKEN, fileId2);
    new File(basePath + "/" + partitionPath + "/" + fileName1).createNewFile();
    new File(basePath + "/" + partitionPath + "/" + fileName2).createNewFile();
    HoodieInstant instant1 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, commitTime1);
    // build writeStats
    HashMap<String, List<String>> partitionToFile1 = new HashMap<>();
    ArrayList<String> files1 = new ArrayList<>();
    files1.add(fileId1);
    files1.add(fileId2);
    partitionToFile1.put(partitionPath, files1);
    List<HoodieWriteStat> writeStats1 = buildWriteStats(partitionToFile1, commitTime1);
    HoodieCommitMetadata commitMetadata1 = CommitUtils.buildMetadata(writeStats1, new HashMap<>(), Option.empty(), WriteOperationType.INSERT, "", HoodieTimeline.COMMIT_ACTION);
    saveAsComplete(commitTimeline, instant1, Option.of(commitMetadata1.toJsonString().getBytes(StandardCharsets.UTF_8)));
    commitTimeline.reload();
    // replace commit
    String commitTime2 = "2";
    String fileName3 = FSUtils.makeDataFileName(commitTime2, TEST_WRITE_TOKEN, fileId3);
    new File(basePath + "/" + partitionPath + "/" + fileName3).createNewFile();
    HoodieInstant instant2 = new HoodieInstant(true, HoodieTimeline.REPLACE_COMMIT_ACTION, commitTime2);
    Map<String, List<String>> partitionToReplaceFileIds = new HashMap<>();
    List<String> replacedFileIds = new ArrayList<>();
    replacedFileIds.add(fileId1);
    replacedFileIds.add(fileId2);
    partitionToReplaceFileIds.put(partitionPath, replacedFileIds);
    HashMap<String, List<String>> partitionToFile2 = new HashMap<>();
    ArrayList<String> files2 = new ArrayList<>();
    files2.add(fileId3);
    partitionToFile2.put(partitionPath, files2);
    List<HoodieWriteStat> writeStats2 = buildWriteStats(partitionToFile2, commitTime2);
    HoodieCommitMetadata commitMetadata2 = CommitUtils.buildMetadata(writeStats2, partitionToReplaceFileIds, Option.empty(), WriteOperationType.INSERT_OVERWRITE, "", HoodieTimeline.REPLACE_COMMIT_ACTION);
    saveAsComplete(commitTimeline, instant2, Option.of(commitMetadata2.toJsonString().getBytes(StandardCharsets.UTF_8)));
    // another insert commit
    String commitTime3 = "3";
    String fileName4 = FSUtils.makeDataFileName(commitTime3, TEST_WRITE_TOKEN, fileId4);
    new File(basePath + "/" + partitionPath + "/" + fileName4).createNewFile();
    HoodieInstant instant3 = new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, commitTime3);
    // build writeStats
    HashMap<String, List<String>> partitionToFile3 = new HashMap<>();
    ArrayList<String> files3 = new ArrayList<>();
    files3.add(fileId4);
    partitionToFile3.put(partitionPath, files3);
    List<HoodieWriteStat> writeStats3 = buildWriteStats(partitionToFile3, commitTime3);
    HoodieCommitMetadata commitMetadata3 = CommitUtils.buildMetadata(writeStats3, new HashMap<>(), Option.empty(), WriteOperationType.INSERT, "", HoodieTimeline.COMMIT_ACTION);
    saveAsComplete(commitTimeline, instant3, Option.of(commitMetadata3.toJsonString().getBytes(StandardCharsets.UTF_8)));
    metaClient.reloadActiveTimeline();
    refreshFsView();
    ArrayList<String> commits = new ArrayList<>();
    commits.add(commitTime1);
    commits.add(commitTime2);
    commits.add(commitTime3);
    // do check
    latestBaseFilesBeforeOrOn = fsView.getLatestBaseFilesBeforeOrOn(partitionPath, commitTime3).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(2, latestBaseFilesBeforeOrOn.size());
    assertTrue(latestBaseFilesBeforeOrOn.contains(fileId3));
    assertTrue(latestBaseFilesBeforeOrOn.contains(fileId4));
    // could see fileId3 because clustering is committed.
    baseFileOn = fsView.getBaseFileOn(partitionPath, commitTime2, fileId3);
    assertTrue(baseFileOn.isPresent());
    assertEquals(baseFileOn.get().getFileId(), fileId3);
    latestBaseFilesInRange = fsView.getLatestBaseFilesInRange(commits).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(2, latestBaseFilesInRange.size());
    assertTrue(latestBaseFilesInRange.contains(fileId3));
    assertTrue(latestBaseFilesInRange.contains(fileId4));
    allBaseFiles = fsView.getAllBaseFiles(partitionPath).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(2, allBaseFiles.size());
    assertTrue(allBaseFiles.contains(fileId3));
    assertTrue(allBaseFiles.contains(fileId4));
    // could see fileId3 because clustering is committed.
    latestBaseFiles = fsView.getLatestBaseFiles().map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(2, latestBaseFiles.size());
    assertTrue(allBaseFiles.contains(fileId3));
    assertTrue(allBaseFiles.contains(fileId4));
    // could see fileId3 because clustering is committed.
    latestBaseFilesPerPartition = fsView.getLatestBaseFiles(partitionPath).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(2, latestBaseFiles.size());
    assertTrue(latestBaseFilesPerPartition.contains(fileId3));
    assertTrue(latestBaseFilesPerPartition.contains(fileId4));
    HoodieWrapperFileSystem fs = metaClient.getFs();
    fs.delete(new Path(basePath + "/.hoodie", "1.commit"), false);
    fs.delete(new Path(basePath + "/.hoodie", "1.inflight"), false);
    fs.delete(new Path(basePath + "/.hoodie", "1.commit.requested"), false);
    fs.delete(new Path(basePath + "/.hoodie", "2.replacecommit"), false);
    metaClient.reloadActiveTimeline();
    refreshFsView();
    // do check after delete some commit file
    latestBaseFilesBeforeOrOn = fsView.getLatestBaseFilesBeforeOrOn(partitionPath, commitTime3).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(3, latestBaseFilesBeforeOrOn.size());
    assertTrue(latestBaseFilesBeforeOrOn.contains(fileId1));
    assertTrue(latestBaseFilesBeforeOrOn.contains(fileId2));
    assertTrue(latestBaseFilesBeforeOrOn.contains(fileId4));
    // couldn't see fileId3 because clustering is not committed.
    baseFileOn = fsView.getBaseFileOn(partitionPath, commitTime2, fileId3);
    assertFalse(baseFileOn.isPresent());
    latestBaseFilesInRange = fsView.getLatestBaseFilesInRange(commits).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(3, latestBaseFilesInRange.size());
    assertTrue(latestBaseFilesInRange.contains(fileId1));
    assertTrue(latestBaseFilesInRange.contains(fileId2));
    assertTrue(latestBaseFilesInRange.contains(fileId4));
    allBaseFiles = fsView.getAllBaseFiles(partitionPath).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(3, allBaseFiles.size());
    assertTrue(allBaseFiles.contains(fileId1));
    assertTrue(allBaseFiles.contains(fileId2));
    assertTrue(allBaseFiles.contains(fileId4));
    // couldn't see fileId3 because clustering is not committed.
    latestBaseFiles = fsView.getLatestBaseFiles().map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(3, latestBaseFiles.size());
    assertTrue(allBaseFiles.contains(fileId1));
    assertTrue(allBaseFiles.contains(fileId2));
    assertTrue(allBaseFiles.contains(fileId4));
    // couldn't see fileId3 because clustering is not committed.
    latestBaseFilesPerPartition = fsView.getLatestBaseFiles(partitionPath).map(HoodieBaseFile::getFileId).collect(Collectors.toList());
    assertEquals(3, latestBaseFiles.size());
    assertTrue(latestBaseFilesPerPartition.contains(fileId1));
    assertTrue(latestBaseFilesPerPartition.contains(fileId2));
    assertTrue(latestBaseFilesPerPartition.contains(fileId4));
}

Example 38 with HoodieCommitMetadata

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

the class TestIncrementalFSViewSync method addInstant.

private List<String> addInstant(HoodieTableMetaClient metaClient, String instant, boolean deltaCommit, String baseInstant) throws IOException {
    List<Pair<String, HoodieWriteStat>> writeStats = generateDataForInstant(baseInstant, instant, deltaCommit);
    HoodieCommitMetadata metadata = new HoodieCommitMetadata();
    writeStats.forEach(e -> metadata.addWriteStat(e.getKey(), e.getValue()));
    HoodieInstant inflightInstant = new HoodieInstant(true, deltaCommit ? HoodieTimeline.DELTA_COMMIT_ACTION : HoodieTimeline.COMMIT_ACTION, instant);
    metaClient.getActiveTimeline().createNewInstant(inflightInstant);
    metaClient.getActiveTimeline().saveAsComplete(inflightInstant, Option.of(metadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    /*
    // Delete pending compaction if present
    metaClient.getFs().delete(new Path(metaClient.getMetaPath(),
        new HoodieInstant(State.REQUESTED, HoodieTimeline.COMPACTION_ACTION, instant).getFileName()));
     */
    return writeStats.stream().map(e -> e.getValue().getPath()).collect(Collectors.toList());
}

Example 39 with HoodieCommitMetadata

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

the class TestIncrementalFSViewSync method testIngestion.

@Test
public void testIngestion() throws IOException {
    SyncableFileSystemView view = getFileSystemView(metaClient);
    // Add an empty ingestion
    String firstEmptyInstantTs = "11";
    HoodieCommitMetadata metadata = new HoodieCommitMetadata();
    metaClient.getActiveTimeline().createNewInstant(new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, firstEmptyInstantTs));
    metaClient.getActiveTimeline().saveAsComplete(new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, firstEmptyInstantTs), Option.of(metadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    view.sync();
    assertTrue(view.getLastInstant().isPresent());
    assertEquals("11", view.getLastInstant().get().getTimestamp());
    assertEquals(State.COMPLETED, view.getLastInstant().get().getState());
    assertEquals(HoodieTimeline.COMMIT_ACTION, view.getLastInstant().get().getAction());
    partitions.forEach(p -> assertEquals(0, view.getLatestFileSlices(p).count()));
    metaClient.reloadActiveTimeline();
    SyncableFileSystemView newView = getFileSystemView(metaClient);
    areViewsConsistent(view, newView, 0L);
    // Add 3 non-empty ingestions to COW table
    Map<String, List<String>> instantsToFiles = testMultipleWriteSteps(view, Arrays.asList("12", "13", "14"));
    // restore instants in reverse order till we rollback all
    testRestore(view, Arrays.asList("15", "16", "17"), instantsToFiles, Arrays.asList(getHoodieCommitInstant("14", false), getHoodieCommitInstant("13", false), getHoodieCommitInstant("12", false)), "17", true);
    // Add 5 non-empty ingestions back-to-back
    instantsToFiles = testMultipleWriteSteps(view, Arrays.asList("18", "19", "20"));
    // Clean instants.
    testCleans(view, Arrays.asList("21", "22"), instantsToFiles, Arrays.asList("18", "19"), 0, 0);
}

Example 40 with HoodieCommitMetadata

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

the class TestIncrementalFSViewSync method testReplaceCommits.

@Test
public void testReplaceCommits() throws IOException {
    SyncableFileSystemView view = getFileSystemView(metaClient);
    // Add an empty ingestion
    String firstEmptyInstantTs = "11";
    HoodieCommitMetadata metadata = new HoodieCommitMetadata();
    metaClient.getActiveTimeline().createNewInstant(new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, firstEmptyInstantTs));
    metaClient.getActiveTimeline().saveAsComplete(new HoodieInstant(true, HoodieTimeline.COMMIT_ACTION, firstEmptyInstantTs), Option.of(metadata.toJsonString().getBytes(StandardCharsets.UTF_8)));
    view.sync();
    assertTrue(view.getLastInstant().isPresent());
    assertEquals("11", view.getLastInstant().get().getTimestamp());
    assertEquals(State.COMPLETED, view.getLastInstant().get().getState());
    assertEquals(HoodieTimeline.COMMIT_ACTION, view.getLastInstant().get().getAction());
    partitions.forEach(p -> assertEquals(0, view.getLatestFileSlices(p).count()));
    metaClient.reloadActiveTimeline();
    SyncableFileSystemView newView = getFileSystemView(metaClient);
    areViewsConsistent(view, newView, 0L);
    // Add 1 non-empty ingestions to COW table
    Map<String, List<String>> instantsToFiles = testMultipleWriteSteps(view, Arrays.asList("12"));
    // ADD replace instants
    testMultipleReplaceSteps(instantsToFiles, view, Arrays.asList("13", "14"), NUM_FILE_IDS_PER_PARTITION);
    // restore instants in reverse order till we rollback all replace instants
    testRestore(view, Arrays.asList("15", "16"), instantsToFiles, Arrays.asList(getHoodieReplaceInstant("14"), getHoodieReplaceInstant("13")), "17", true, 1, fileIdsPerPartition.size());
    // clear files from inmemory view for replaced instants
    instantsToFiles.remove("14");
    instantsToFiles.remove("13");
    // add few more replace instants
    testMultipleReplaceSteps(instantsToFiles, view, Arrays.asList("18", "19", "20"), NUM_FILE_IDS_PER_PARTITION);
    // Clean instants.
    testCleans(view, Arrays.asList("21", "22"), instantsToFiles, Arrays.asList("18", "19"), NUM_FILE_IDS_PER_PARTITION, 1);
}