Examples with CheckpointV2 org.apache.samza.checkpoint.CheckpointV2 used on opensource projects

Example 1 with CheckpointV2

use of org.apache.samza.checkpoint.CheckpointV2 in project samza by apache.

the class TestKafkaCheckpointManager method testReadCheckpointV2.

@Test
public void testReadCheckpointV2() throws InterruptedException {
    Config config = config(ImmutableMap.of(TaskConfig.CHECKPOINT_READ_VERSIONS, "1,2"));
    setupSystemFactory(config);
    CheckpointV2 checkpointV2 = buildCheckpointV2(INPUT_SSP0, "0");
    List<IncomingMessageEnvelope> checkpointEnvelopes = ImmutableList.of(newCheckpointV2Envelope(TASK0, checkpointV2, "0"));
    setupConsumer(checkpointEnvelopes);
    KafkaCheckpointManager kafkaCheckpointManager = buildKafkaCheckpointManager(true, config);
    kafkaCheckpointManager.register(TASK0);
    Checkpoint actualCheckpoint = kafkaCheckpointManager.readLastCheckpoint(TASK0);
    assertEquals(checkpointV2, actualCheckpoint);
}

Example 2 with CheckpointV2

use of org.apache.samza.checkpoint.CheckpointV2 in project samza by apache.

the class TestCheckpointV2Serde method testCheckpointV2Serde.

@Test
public void testCheckpointV2Serde() {
    CheckpointV2Serde serde = new CheckpointV2Serde();
    Map<SystemStreamPartition, String> offsets = new HashMap<>();
    SystemStreamPartition systemStreamPartition = new SystemStreamPartition("test-system", "test-stream", new Partition(777));
    offsets.put(systemStreamPartition, "1");
    // State Checkpoint marker
    Map<String, Map<String, String>> factoryStateCheckpointMarkersMap = new HashMap<>();
    Map<String, String> stateCheckpointMarkersMap = new HashMap<>();
    stateCheckpointMarkersMap.put("store1", "marker1");
    stateCheckpointMarkersMap.put("store2", "marker2");
    Map<String, String> stateCheckpointMarkersMap2 = new HashMap<>();
    stateCheckpointMarkersMap2.put("store1", "marker3");
    stateCheckpointMarkersMap2.put("store2", "marker4");
    factoryStateCheckpointMarkersMap.put("factory1", stateCheckpointMarkersMap);
    factoryStateCheckpointMarkersMap.put("factory2", stateCheckpointMarkersMap2);
    CheckpointId checkpointId = CheckpointId.create();
    CheckpointV2 checkpoint = new CheckpointV2(checkpointId, offsets, factoryStateCheckpointMarkersMap);
    CheckpointV2 deserializedCheckpoint = serde.fromBytes(serde.toBytes(checkpoint));
    // Validate input checkpoints
    assertEquals(checkpointId, deserializedCheckpoint.getCheckpointId());
    assertEquals("1", deserializedCheckpoint.getOffsets().get(systemStreamPartition));
    assertEquals(1, deserializedCheckpoint.getOffsets().size());
    // Validate state checkpoints
    assertEquals(2, deserializedCheckpoint.getStateCheckpointMarkers().size());
    assertTrue(deserializedCheckpoint.getStateCheckpointMarkers().containsKey("factory1"));
    assertEquals(stateCheckpointMarkersMap, deserializedCheckpoint.getStateCheckpointMarkers().get("factory1"));
    assertTrue(deserializedCheckpoint.getStateCheckpointMarkers().containsKey("factory2"));
    assertEquals(stateCheckpointMarkersMap2, deserializedCheckpoint.getStateCheckpointMarkers().get("factory2"));
}

Example 3 with CheckpointV2

use of org.apache.samza.checkpoint.CheckpointV2 in project samza by apache.

the class TestTaskStorageCommitManager method testPersistToFileSystemCheckpointV1AndV2Checkpoint.

@Test
public void testPersistToFileSystemCheckpointV1AndV2Checkpoint() throws IOException {
    ContainerStorageManager containerStorageManager = mock(ContainerStorageManager.class);
    StorageEngine mockLPStore = mock(StorageEngine.class);
    StoreProperties lpStoreProps = mock(StoreProperties.class);
    when(mockLPStore.getStoreProperties()).thenReturn(lpStoreProps);
    when(lpStoreProps.isPersistedToDisk()).thenReturn(true);
    when(lpStoreProps.isDurableStore()).thenReturn(true);
    Path mockPath = mock(Path.class);
    when(mockLPStore.checkpoint(any())).thenReturn(Optional.of(mockPath));
    StorageEngine mockPStore = mock(StorageEngine.class);
    StoreProperties pStoreProps = mock(StoreProperties.class);
    when(mockPStore.getStoreProperties()).thenReturn(pStoreProps);
    when(pStoreProps.isPersistedToDisk()).thenReturn(true);
    when(pStoreProps.isDurableStore()).thenReturn(false);
    StorageEngine mockLIStore = mock(StorageEngine.class);
    StoreProperties liStoreProps = mock(StoreProperties.class);
    when(mockLIStore.getStoreProperties()).thenReturn(liStoreProps);
    when(liStoreProps.isPersistedToDisk()).thenReturn(false);
    when(liStoreProps.isDurableStore()).thenReturn(true);
    StorageEngine mockIStore = mock(StorageEngine.class);
    StoreProperties iStoreProps = mock(StoreProperties.class);
    when(mockIStore.getStoreProperties()).thenReturn(iStoreProps);
    when(iStoreProps.isPersistedToDisk()).thenReturn(false);
    when(iStoreProps.isDurableStore()).thenReturn(false);
    Map<String, StorageEngine> taskStores = ImmutableMap.of("loggedPersistentStore", mockLPStore, "persistentStore", mockPStore, "loggedInMemStore", mockLIStore, "inMemStore", mockIStore);
    Partition changelogPartition = new Partition(0);
    SystemStream changelogSystemStream = new SystemStream("changelogSystem", "changelogStream");
    SystemStreamPartition changelogSSP = new SystemStreamPartition(changelogSystemStream, changelogPartition);
    Map<String, SystemStream> storeChangelogsStreams = ImmutableMap.of("loggedPersistentStore", changelogSystemStream, "loggedInMemStore", new SystemStream("system", "stream"));
    StorageManagerUtil storageManagerUtil = mock(StorageManagerUtil.class);
    File durableStoreDir = new File("durableStorePath");
    when(storageManagerUtil.getTaskStoreDir(eq(durableStoreDir), any(), any(), any())).thenReturn(durableStoreDir);
    TaskName taskName = new TaskName("task");
    TaskInstanceMetrics metrics = mock(TaskInstanceMetrics.class);
    Timer checkpointTimer = mock(Timer.class);
    when(metrics.storeCheckpointNs()).thenReturn(checkpointTimer);
    when(containerStorageManager.getAllStores(taskName)).thenReturn(taskStores);
    TaskStorageCommitManager commitManager = spy(new TaskStorageCommitManager(taskName, Collections.emptyMap(), containerStorageManager, storeChangelogsStreams, changelogPartition, null, null, ForkJoinPool.commonPool(), storageManagerUtil, durableStoreDir, metrics));
    doNothing().when(commitManager).writeChangelogOffsetFile(any(), any(), any(), any());
    when(storageManagerUtil.getStoreCheckpointDir(any(File.class), any(CheckpointId.class))).thenAnswer((Answer<String>) invocation -> {
        File file = invocation.getArgumentAt(0, File.class);
        CheckpointId checkpointId = invocation.getArgumentAt(1, CheckpointId.class);
        return file + "-" + checkpointId;
    });
    CheckpointId newCheckpointId = CheckpointId.create();
    String newestOffset = "1";
    KafkaChangelogSSPOffset kafkaChangelogSSPOffset = new KafkaChangelogSSPOffset(newCheckpointId, newestOffset);
    Map<SystemStreamPartition, String> offsetsJava = ImmutableMap.of(changelogSSP, kafkaChangelogSSPOffset.toString());
    commitManager.init();
    // invoke persist to file system for v2 checkpoint
    commitManager.writeCheckpointToStoreDirectories(new CheckpointV1(offsetsJava));
    verify(commitManager).writeChangelogOffsetFiles(offsetsJava);
    // evoked twice, for OFFSET-V1 and OFFSET-V2
    verify(commitManager).writeChangelogOffsetFile(eq("loggedPersistentStore"), eq(changelogSSP), eq(newestOffset), eq(durableStoreDir));
    File checkpointFile = Paths.get(storageManagerUtil.getStoreCheckpointDir(durableStoreDir, kafkaChangelogSSPOffset.getCheckpointId())).toFile();
    verify(commitManager).writeChangelogOffsetFile(eq("loggedPersistentStore"), eq(changelogSSP), eq(newestOffset), eq(checkpointFile));
    Map<String, String> storeSCM = ImmutableMap.of("loggedPersistentStore", "system;loggedPersistentStoreStream;1", "persistentStore", "system;persistentStoreStream;1", "loggedInMemStore", "system;loggedInMemStoreStream;1", "inMemStore", "system;inMemStoreStream;1");
    CheckpointV2 checkpoint = new CheckpointV2(newCheckpointId, Collections.emptyMap(), Collections.singletonMap("factory", storeSCM));
    // invoke persist to file system for v2 checkpoint
    commitManager.writeCheckpointToStoreDirectories(checkpoint);
    // Validate only durable and persisted stores are persisted
    // This should be evoked twice, for checkpointV1 and checkpointV2
    verify(storageManagerUtil, times(2)).getTaskStoreDir(eq(durableStoreDir), eq("loggedPersistentStore"), eq(taskName), any());
    File checkpointPath = Paths.get(storageManagerUtil.getStoreCheckpointDir(durableStoreDir, newCheckpointId)).toFile();
    verify(storageManagerUtil).writeCheckpointV2File(eq(checkpointPath), eq(checkpoint));
}

Example 4 with CheckpointV2

use of org.apache.samza.checkpoint.CheckpointV2 in project samza by apache.

the class TestTaskStorageCommitManager method testWriteChangelogOffsetFilesV2andV1.

@Test
public void testWriteChangelogOffsetFilesV2andV1() throws IOException {
    Map<String, Map<SystemStreamPartition, String>> mockFileSystem = new HashMap<>();
    ContainerStorageManager containerStorageManager = mock(ContainerStorageManager.class);
    Map<String, CheckpointV2> mockCheckpointFileSystem = new HashMap<>();
    StorageEngine mockLPStore = mock(StorageEngine.class);
    StoreProperties lpStoreProps = mock(StoreProperties.class);
    when(mockLPStore.getStoreProperties()).thenReturn(lpStoreProps);
    when(lpStoreProps.isPersistedToDisk()).thenReturn(true);
    when(lpStoreProps.isDurableStore()).thenReturn(true);
    Path mockPath = mock(Path.class);
    when(mockLPStore.checkpoint(any())).thenReturn(Optional.of(mockPath));
    TaskInstanceMetrics metrics = mock(TaskInstanceMetrics.class);
    Timer checkpointTimer = mock(Timer.class);
    when(metrics.storeCheckpointNs()).thenReturn(checkpointTimer);
    java.util.Map<String, StorageEngine> taskStores = ImmutableMap.of("loggedPersistentStore", mockLPStore);
    Partition changelogPartition = new Partition(0);
    SystemStream changelogSystemStream = new SystemStream("changelogSystem", "changelogStream");
    SystemStreamPartition changelogSSP = new SystemStreamPartition(changelogSystemStream, changelogPartition);
    java.util.Map<String, SystemStream> storeChangelogsStreams = ImmutableMap.of("loggedPersistentStore", changelogSystemStream);
    StorageManagerUtil storageManagerUtil = mock(StorageManagerUtil.class);
    File tmpTestPath = new File("store-checkpoint-test");
    when(storageManagerUtil.getTaskStoreDir(eq(tmpTestPath), eq("loggedPersistentStore"), any(), any())).thenReturn(tmpTestPath);
    TaskName taskName = new TaskName("task");
    when(containerStorageManager.getAllStores(taskName)).thenReturn(taskStores);
    TaskStorageCommitManager commitManager = spy(new TaskStorageCommitManager(taskName, Collections.emptyMap(), containerStorageManager, storeChangelogsStreams, changelogPartition, null, null, ForkJoinPool.commonPool(), storageManagerUtil, tmpTestPath, metrics));
    doAnswer(invocation -> {
        String fileDir = invocation.getArgumentAt(3, File.class).getName();
        SystemStreamPartition ssp = invocation.getArgumentAt(1, SystemStreamPartition.class);
        String offset = invocation.getArgumentAt(2, String.class);
        if (mockFileSystem.containsKey(fileDir)) {
            mockFileSystem.get(fileDir).put(ssp, offset);
        } else {
            Map<SystemStreamPartition, String> sspOffsets = new HashMap<>();
            sspOffsets.put(ssp, offset);
            mockFileSystem.put(fileDir, sspOffsets);
        }
        return null;
    }).when(commitManager).writeChangelogOffsetFile(any(), any(), any(), any());
    doAnswer(invocation -> {
        String storeDir = invocation.getArgumentAt(0, File.class).getName();
        CheckpointV2 checkpointV2 = invocation.getArgumentAt(1, CheckpointV2.class);
        mockCheckpointFileSystem.put(storeDir, checkpointV2);
        return null;
    }).when(storageManagerUtil).writeCheckpointV2File(any(), any());
    when(storageManagerUtil.getStoreCheckpointDir(any(File.class), any(CheckpointId.class))).thenAnswer((Answer<String>) invocation -> {
        File file = invocation.getArgumentAt(0, File.class);
        CheckpointId checkpointId = invocation.getArgumentAt(1, CheckpointId.class);
        return file + "-" + checkpointId;
    });
    CheckpointId newCheckpointId = CheckpointId.create();
    String newestOffset = "1";
    KafkaChangelogSSPOffset kafkaChangelogSSPOffset = new KafkaChangelogSSPOffset(newCheckpointId, newestOffset);
    java.util.Map<SystemStreamPartition, String> offsetsJava = ImmutableMap.of(changelogSSP, kafkaChangelogSSPOffset.toString());
    commitManager.init();
    // invoke persist to file system for v1 checkpoint
    commitManager.writeCheckpointToStoreDirectories(new CheckpointV1(offsetsJava));
    assertEquals(2, mockFileSystem.size());
    // check if v2 offsets are written correctly
    String v2FilePath = storageManagerUtil.getStoreCheckpointDir(tmpTestPath, newCheckpointId);
    assertTrue(mockFileSystem.containsKey(v2FilePath));
    assertTrue(mockFileSystem.get(v2FilePath).containsKey(changelogSSP));
    assertEquals(1, mockFileSystem.get(v2FilePath).size());
    assertEquals(newestOffset, mockFileSystem.get(v2FilePath).get(changelogSSP));
    // check if v1 offsets are written correctly
    String v1FilePath = tmpTestPath.getPath();
    assertTrue(mockFileSystem.containsKey(v1FilePath));
    assertTrue(mockFileSystem.get(v1FilePath).containsKey(changelogSSP));
    assertEquals(1, mockFileSystem.get(v1FilePath).size());
    assertEquals(newestOffset, mockFileSystem.get(v1FilePath).get(changelogSSP));
    java.util.Map<String, String> storeSCM = ImmutableMap.of("loggedPersistentStore", "system;loggedPersistentStoreStream;1", "persistentStore", "system;persistentStoreStream;1", "loggedInMemStore", "system;loggedInMemStoreStream;1", "inMemStore", "system;inMemStoreStream;1");
    CheckpointV2 checkpoint = new CheckpointV2(newCheckpointId, Collections.emptyMap(), Collections.singletonMap("factory", storeSCM));
    // invoke persist to file system with checkpoint v2
    commitManager.writeCheckpointToStoreDirectories(checkpoint);
    assertTrue(mockCheckpointFileSystem.containsKey(v2FilePath));
    assertEquals(checkpoint, mockCheckpointFileSystem.get(v2FilePath));
    assertTrue(mockCheckpointFileSystem.containsKey(v1FilePath));
    assertEquals(checkpoint, mockCheckpointFileSystem.get(v1FilePath));
    assertEquals(2, mockCheckpointFileSystem.size());
    CheckpointV2 updatedCheckpoint = new CheckpointV2(newCheckpointId, ImmutableMap.of(new SystemStreamPartition("inputSystem", "inputStream", changelogPartition), "5"), Collections.singletonMap("factory", storeSCM));
    commitManager.writeCheckpointToStoreDirectories(updatedCheckpoint);
    assertEquals(updatedCheckpoint, mockCheckpointFileSystem.get(v2FilePath));
    assertEquals(updatedCheckpoint, mockCheckpointFileSystem.get(v1FilePath));
    assertEquals(2, mockCheckpointFileSystem.size());
}

Example 5 with CheckpointV2

use of org.apache.samza.checkpoint.CheckpointV2 in project samza by apache.

the class ContainerStorageManager method getBackendFactoryStoreNames.

/**
 * Return a map of backend factory names to set of stores that should be restored using it
 */
@VisibleForTesting
Map<String, Set<String>> getBackendFactoryStoreNames(Checkpoint checkpoint, Set<String> storeNames, StorageConfig storageConfig) {
    // backendFactoryName -> set(storeNames)
    Map<String, Set<String>> backendFactoryStoreNames = new HashMap<>();
    if (checkpoint != null && checkpoint.getVersion() == 1) {
        // Only restore stores with changelog streams configured
        Set<String> changelogStores = storeNames.stream().filter(storeName -> storageConfig.getChangelogStream(storeName).isPresent()).collect(Collectors.toSet());
        // Default to changelog backend factory when using checkpoint v1 for backwards compatibility
        if (!changelogStores.isEmpty()) {
            backendFactoryStoreNames.put(StorageConfig.KAFKA_STATE_BACKEND_FACTORY, changelogStores);
        }
        if (storeNames.size() > changelogStores.size()) {
            Set<String> nonChangelogStores = storeNames.stream().filter(storeName -> !changelogStores.contains(storeName)).collect(Collectors.toSet());
            LOG.info("non-Side input stores: {}, do not have a configured store changelogs for checkpoint V1," + "restore for the store will be skipped", nonChangelogStores);
        }
    } else if (checkpoint == null || checkpoint.getVersion() == 2) {
        // Extract the state checkpoint markers if checkpoint exists
        Map<String, Map<String, String>> stateCheckpointMarkers = checkpoint == null ? Collections.emptyMap() : ((CheckpointV2) checkpoint).getStateCheckpointMarkers();
        // Find stores associated to each state backend factory
        storeNames.forEach(storeName -> {
            List<String> storeFactories = storageConfig.getStoreRestoreFactories(storeName);
            if (storeFactories.isEmpty()) {
                // If the restore factory is not configured for the store and the store does not have a changelog topic
                LOG.info("non-Side input store: {}, does not have a configured restore factories nor store changelogs," + "restore for the store will be skipped", storeName);
            } else {
                // Search the ordered list for the first matched state backend factory in the checkpoint
                // If the checkpoint does not exist or state checkpoint markers does not exist, we match the first configured
                // restore manager
                Optional<String> factoryNameOpt = storeFactories.stream().filter(factoryName -> stateCheckpointMarkers.containsKey(factoryName) && stateCheckpointMarkers.get(factoryName).containsKey(storeName)).findFirst();
                String factoryName;
                if (factoryNameOpt.isPresent()) {
                    factoryName = factoryNameOpt.get();
                } else {
                    // Restore factories configured but no checkpoints found
                    // Use first configured restore factory
                    factoryName = storeFactories.get(0);
                    LOG.warn("No matching checkpoints found for configured factories: {}, " + "defaulting to using the first configured factory with no checkpoints", storeFactories);
                }
                if (!backendFactoryStoreNames.containsKey(factoryName)) {
                    backendFactoryStoreNames.put(factoryName, new HashSet<>());
                }
                backendFactoryStoreNames.get(factoryName).add(storeName);
            }
        });
    } else {
        throw new SamzaException(String.format("Unsupported checkpoint version %s", checkpoint.getVersion()));
    }
    return backendFactoryStoreNames;
}