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

Example 1 with CheckpointManager

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

the class TestTaskStorageCommitManager method testCommitManagerStart.

@Test
public void testCommitManagerStart() {
    CheckpointManager checkpointManager = mock(CheckpointManager.class);
    TaskBackupManager taskBackupManager1 = mock(TaskBackupManager.class);
    TaskBackupManager taskBackupManager2 = mock(TaskBackupManager.class);
    ContainerStorageManager containerStorageManager = mock(ContainerStorageManager.class);
    Checkpoint checkpoint = mock(Checkpoint.class);
    TaskName taskName = new TaskName("task1");
    Map<String, TaskBackupManager> backupManagers = ImmutableMap.of("factory1", taskBackupManager1, "factory2", taskBackupManager2);
    TaskStorageCommitManager cm = new TaskStorageCommitManager(taskName, backupManagers, containerStorageManager, Collections.emptyMap(), new Partition(1), checkpointManager, new MapConfig(), ForkJoinPool.commonPool(), new StorageManagerUtil(), null, null);
    when(checkpointManager.readLastCheckpoint(taskName)).thenReturn(checkpoint);
    cm.init();
    verify(taskBackupManager1).init(eq(checkpoint));
    verify(taskBackupManager2).init(eq(checkpoint));
}

Example 2 with CheckpointManager

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

the class StreamManager method clearStreamsFromPreviousRun.

/**
 * This is a best-effort approach to clear the internal streams from previous run, including intermediate streams,
 * checkpoint stream and changelog streams.
 * For batch processing, we always clean up the previous internal streams and create a new set for each run.
 * @param prevConfig config of the previous run
 */
public void clearStreamsFromPreviousRun(Config prevConfig) {
    try {
        ApplicationConfig appConfig = new ApplicationConfig(prevConfig);
        LOGGER.info("run.id from previous run is {}", appConfig.getRunId());
        StreamConfig streamConfig = new StreamConfig(prevConfig);
        // Find all intermediate streams and clean up
        Set<StreamSpec> intStreams = streamConfig.getStreamIds().stream().filter(streamConfig::getIsIntermediateStream).map(id -> new StreamSpec(id, streamConfig.getPhysicalName(id), streamConfig.getSystem(id))).collect(Collectors.toSet());
        intStreams.forEach(stream -> {
            LOGGER.info("Clear intermediate stream {} in system {}", stream.getPhysicalName(), stream.getSystemName());
            systemAdmins.getSystemAdmin(stream.getSystemName()).clearStream(stream);
        });
        // Find checkpoint stream and clean up
        TaskConfig taskConfig = new TaskConfig(prevConfig);
        taskConfig.getCheckpointManager(new MetricsRegistryMap()).ifPresent(CheckpointManager::clearCheckpoints);
        // Find changelog streams and remove them
        StorageConfig storageConfig = new StorageConfig(prevConfig);
        for (String store : storageConfig.getStoreNames()) {
            String changelog = storageConfig.getChangelogStream(store).orElse(null);
            if (changelog != null) {
                LOGGER.info("Clear store {} changelog {}", store, changelog);
                SystemStream systemStream = StreamUtil.getSystemStreamFromNames(changelog);
                StreamSpec spec = StreamSpec.createChangeLogStreamSpec(systemStream.getStream(), systemStream.getSystem(), 1);
                systemAdmins.getSystemAdmin(spec.getSystemName()).clearStream(spec);
            }
        }
    } catch (Exception e) {
        // For batch, we always create a new set of internal streams (checkpoint, changelog and intermediate) with unique
        // id. So if clearStream doesn't work, it won't affect the correctness of the results.
        // We log a warning here and rely on retention to clean up the streams later.
        LOGGER.warn("Fail to clear internal streams from previous run. Please clean up manually.", e);
    }
}

Example 3 with CheckpointManager

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

the class StorageRecovery method getContainerStorageManagers.

/**
 * create one TaskStorageManager for each task. Add all of them to the
 * List<TaskStorageManager>
 */
@SuppressWarnings("rawtypes")
private void getContainerStorageManagers() {
    Set<String> factoryClasses = new StorageConfig(jobConfig).getRestoreFactories();
    Map<String, StateBackendFactory> stateBackendFactories = factoryClasses.stream().collect(Collectors.toMap(factoryClass -> factoryClass, factoryClass -> ReflectionUtil.getObj(factoryClass, StateBackendFactory.class)));
    Clock clock = SystemClock.instance();
    StreamMetadataCache streamMetadataCache = new StreamMetadataCache(systemAdmins, 5000, clock);
    // don't worry about prefetching for this; looks like the tool doesn't flush to offset files anyways
    Map<String, SystemFactory> systemFactories = new SystemConfig(jobConfig).getSystemFactories();
    CheckpointManager checkpointManager = new TaskConfig(jobConfig).getCheckpointManager(new MetricsRegistryMap()).orElse(null);
    for (ContainerModel containerModel : containers.values()) {
        ContainerContext containerContext = new ContainerContextImpl(containerModel, new MetricsRegistryMap());
        ContainerStorageManager containerStorageManager = new ContainerStorageManager(checkpointManager, containerModel, streamMetadataCache, systemAdmins, changeLogSystemStreams, new HashMap<>(), storageEngineFactories, systemFactories, this.getSerdes(), jobConfig, new HashMap<>(), new SamzaContainerMetrics(containerModel.getId(), new MetricsRegistryMap(), ""), JobContextImpl.fromConfigWithDefaults(jobConfig, jobModel), containerContext, stateBackendFactories, new HashMap<>(), storeBaseDir, storeBaseDir, null, new SystemClock());
        this.containerStorageManagers.put(containerModel.getId(), containerStorageManager);
    }
}

Example 4 with CheckpointManager

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

the class TestContainerStorageManager method testNoConfiguredDurableStores.

@Test
public void testNoConfiguredDurableStores() throws InterruptedException {
    taskRestoreMetricGauges = new HashMap<>();
    this.tasks = new HashMap<>();
    this.taskInstanceMetrics = new HashMap<>();
    // Add two mocked tasks
    addMockedTask("task 0", 0);
    addMockedTask("task 1", 1);
    // Mock container metrics
    samzaContainerMetrics = mock(SamzaContainerMetrics.class);
    when(samzaContainerMetrics.taskStoreRestorationMetrics()).thenReturn(taskRestoreMetricGauges);
    // Create mocked configs for specifying serdes
    Map<String, String> configMap = new HashMap<>();
    configMap.put("serializers.registry.stringserde.class", StringSerdeFactory.class.getName());
    configMap.put(TaskConfig.TRANSACTIONAL_STATE_RETAIN_EXISTING_STATE, "true");
    Config config = new MapConfig(configMap);
    Map<String, Serde<Object>> serdes = new HashMap<>();
    serdes.put("stringserde", mock(Serde.class));
    CheckpointManager checkpointManager = mock(CheckpointManager.class);
    when(checkpointManager.readLastCheckpoint(any(TaskName.class))).thenReturn(new CheckpointV1(new HashMap<>()));
    ContainerContext mockContainerContext = mock(ContainerContext.class);
    ContainerModel mockContainerModel = new ContainerModel("samza-container-test", tasks);
    when(mockContainerContext.getContainerModel()).thenReturn(mockContainerModel);
    // Reset the expected number of sysConsumer create, start and stop calls, and store.restore() calls
    this.systemConsumerCreationCount = 0;
    this.systemConsumerStartCount = 0;
    this.systemConsumerStopCount = 0;
    this.storeRestoreCallCount = 0;
    StateBackendFactory backendFactory = mock(StateBackendFactory.class);
    TaskRestoreManager restoreManager = mock(TaskRestoreManager.class);
    when(backendFactory.getRestoreManager(any(), any(), any(), any(), any(), any(), any(), any(), any(), any(), any())).thenReturn(restoreManager);
    doAnswer(invocation -> {
        storeRestoreCallCount++;
        return CompletableFuture.completedFuture(null);
    }).when(restoreManager).restore();
    // Create the container storage manager
    ContainerStorageManager containerStorageManager = new ContainerStorageManager(checkpointManager, mockContainerModel, mock(StreamMetadataCache.class), mock(SystemAdmins.class), new HashMap<>(), new HashMap<>(), new HashMap<>(), new HashMap<>(), serdes, config, taskInstanceMetrics, samzaContainerMetrics, mock(JobContext.class), mockContainerContext, new HashMap<>(), mock(Map.class), DEFAULT_LOGGED_STORE_BASE_DIR, DEFAULT_STORE_BASE_DIR, null, new SystemClock());
    containerStorageManager.start();
    containerStorageManager.shutdown();
    for (Gauge gauge : taskRestoreMetricGauges.values()) {
        Assert.assertTrue("Restoration time gauge value should never be invoked", mockingDetails(gauge).getInvocations().size() == 0);
    }
    Assert.assertEquals("Store restore count should be 2 because there are 0 stores", 0, this.storeRestoreCallCount);
    Assert.assertEquals(0, this.systemConsumerCreationCount);
    Assert.assertEquals(0, this.systemConsumerStopCount);
    Assert.assertEquals(0, this.systemConsumerStartCount);
}

Example 5 with CheckpointManager

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

the class TestContainerStorageManager method setUp.

/**
 * Method to create a containerStorageManager with mocked dependencies
 */
@Before
public void setUp() throws InterruptedException {
    taskRestoreMetricGauges = new HashMap<>();
    this.tasks = new HashMap<>();
    this.taskInstanceMetrics = new HashMap<>();
    // Add two mocked tasks
    addMockedTask("task 0", 0);
    addMockedTask("task 1", 1);
    // Mock container metrics
    samzaContainerMetrics = mock(SamzaContainerMetrics.class);
    when(samzaContainerMetrics.taskStoreRestorationMetrics()).thenReturn(taskRestoreMetricGauges);
    // Create a map of test changeLogSSPs
    Map<String, SystemStream> changelogSystemStreams = new HashMap<>();
    changelogSystemStreams.put(STORE_NAME, new SystemStream(SYSTEM_NAME, STREAM_NAME));
    // Create mocked storage engine factories
    Map<String, StorageEngineFactory<Object, Object>> storageEngineFactories = new HashMap<>();
    StorageEngineFactory mockStorageEngineFactory = (StorageEngineFactory<Object, Object>) mock(StorageEngineFactory.class);
    StorageEngine mockStorageEngine = mock(StorageEngine.class);
    when(mockStorageEngine.getStoreProperties()).thenReturn(new StoreProperties.StorePropertiesBuilder().setLoggedStore(true).setPersistedToDisk(true).build());
    doAnswer(invocation -> {
        return mockStorageEngine;
    }).when(mockStorageEngineFactory).getStorageEngine(anyString(), any(), any(), any(), any(), any(), any(), any(), any(), any());
    storageEngineFactories.put(STORE_NAME, mockStorageEngineFactory);
    // Add instrumentation to mocked storage engine, to record the number of store.restore() calls
    doAnswer(invocation -> {
        storeRestoreCallCount++;
        return CompletableFuture.completedFuture(null);
    }).when(mockStorageEngine).restore(any());
    // Set the mocked stores' properties to be persistent
    doAnswer(invocation -> {
        return new StoreProperties.StorePropertiesBuilder().setLoggedStore(true).build();
    }).when(mockStorageEngine).getStoreProperties();
    // Mock and setup sysconsumers
    SystemConsumer mockSystemConsumer = mock(SystemConsumer.class);
    doAnswer(invocation -> {
        systemConsumerStartCount++;
        return null;
    }).when(mockSystemConsumer).start();
    doAnswer(invocation -> {
        systemConsumerStopCount++;
        return null;
    }).when(mockSystemConsumer).stop();
    // Create mocked system factories
    Map<String, SystemFactory> systemFactories = new HashMap<>();
    // Count the number of sysConsumers created
    SystemFactory mockSystemFactory = mock(SystemFactory.class);
    doAnswer(invocation -> {
        this.systemConsumerCreationCount++;
        return mockSystemConsumer;
    }).when(mockSystemFactory).getConsumer(anyString(), any(), any());
    systemFactories.put(SYSTEM_NAME, mockSystemFactory);
    // Create mocked configs for specifying serdes
    Map<String, String> configMap = new HashMap<>();
    configMap.put("stores." + STORE_NAME + ".key.serde", "stringserde");
    configMap.put("stores." + STORE_NAME + ".msg.serde", "stringserde");
    configMap.put("stores." + STORE_NAME + ".factory", mockStorageEngineFactory.getClass().getName());
    configMap.put("stores." + STORE_NAME + ".changelog", SYSTEM_NAME + "." + STREAM_NAME);
    configMap.put("serializers.registry.stringserde.class", StringSerdeFactory.class.getName());
    configMap.put(TaskConfig.TRANSACTIONAL_STATE_RETAIN_EXISTING_STATE, "true");
    Config config = new MapConfig(configMap);
    Map<String, Serde<Object>> serdes = new HashMap<>();
    serdes.put("stringserde", mock(Serde.class));
    // Create mocked system admins
    SystemAdmin mockSystemAdmin = mock(SystemAdmin.class);
    doAnswer(new Answer<Void>() {

        public Void answer(InvocationOnMock invocation) {
            Object[] args = invocation.getArguments();
            System.out.println("called with arguments: " + Arrays.toString(args));
            return null;
        }
    }).when(mockSystemAdmin).validateStream(any());
    SystemAdmins mockSystemAdmins = mock(SystemAdmins.class);
    when(mockSystemAdmins.getSystemAdmin("kafka")).thenReturn(mockSystemAdmin);
    // Create a mocked mockStreamMetadataCache
    SystemStreamMetadata.SystemStreamPartitionMetadata sspMetadata = new SystemStreamMetadata.SystemStreamPartitionMetadata("0", "50", "51");
    Map<Partition, SystemStreamMetadata.SystemStreamPartitionMetadata> partitionMetadata = new HashMap<>();
    partitionMetadata.put(new Partition(0), sspMetadata);
    partitionMetadata.put(new Partition(1), sspMetadata);
    SystemStreamMetadata systemStreamMetadata = new SystemStreamMetadata(STREAM_NAME, partitionMetadata);
    StreamMetadataCache mockStreamMetadataCache = mock(StreamMetadataCache.class);
    when(mockStreamMetadataCache.getStreamMetadata(JavaConverters.asScalaSetConverter(new HashSet<SystemStream>(changelogSystemStreams.values())).asScala().toSet(), false)).thenReturn(new scala.collection.immutable.Map.Map1(new SystemStream(SYSTEM_NAME, STREAM_NAME), systemStreamMetadata));
    CheckpointManager checkpointManager = mock(CheckpointManager.class);
    when(checkpointManager.readLastCheckpoint(any(TaskName.class))).thenReturn(new CheckpointV1(new HashMap<>()));
    SSPMetadataCache mockSSPMetadataCache = mock(SSPMetadataCache.class);
    when(mockSSPMetadataCache.getMetadata(any(SystemStreamPartition.class))).thenReturn(new SystemStreamMetadata.SystemStreamPartitionMetadata("0", "10", "11"));
    ContainerContext mockContainerContext = mock(ContainerContext.class);
    ContainerModel mockContainerModel = new ContainerModel("samza-container-test", tasks);
    when(mockContainerContext.getContainerModel()).thenReturn(mockContainerModel);
    // Reset the expected number of sysConsumer create, start and stop calls, and store.restore() calls
    this.systemConsumerCreationCount = 0;
    this.systemConsumerStartCount = 0;
    this.systemConsumerStopCount = 0;
    this.storeRestoreCallCount = 0;
    StateBackendFactory backendFactory = mock(StateBackendFactory.class);
    TaskRestoreManager restoreManager = mock(TaskRestoreManager.class);
    ArgumentCaptor<ExecutorService> restoreExecutorCaptor = ArgumentCaptor.forClass(ExecutorService.class);
    when(backendFactory.getRestoreManager(any(), any(), any(), restoreExecutorCaptor.capture(), any(), any(), any(), any(), any(), any(), any())).thenReturn(restoreManager);
    doAnswer(invocation -> {
        storeRestoreCallCount++;
        return CompletableFuture.completedFuture(null);
    }).when(restoreManager).restore();
    // Create the container storage manager
    this.containerStorageManager = new ContainerStorageManager(checkpointManager, mockContainerModel, mockStreamMetadataCache, mockSystemAdmins, changelogSystemStreams, new HashMap<>(), storageEngineFactories, systemFactories, serdes, config, taskInstanceMetrics, samzaContainerMetrics, mock(JobContext.class), mockContainerContext, ImmutableMap.of(StorageConfig.KAFKA_STATE_BACKEND_FACTORY, backendFactory), mock(Map.class), DEFAULT_LOGGED_STORE_BASE_DIR, DEFAULT_STORE_BASE_DIR, null, new SystemClock());
}