Search in sources :

Example 31 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedNonPersistentStore_RestoreToCheckpointedOffset.

@Test
public void testGetStoreActionsForLoggedNonPersistentStore_RestoreToCheckpointedOffset() {
    TaskModel mockTaskModel = mock(TaskModel.class);
    TaskName taskName = new TaskName("Partition 0");
    when(mockTaskModel.getTaskName()).thenReturn(taskName);
    Partition taskChangelogPartition = new Partition(0);
    when(mockTaskModel.getChangelogPartition()).thenReturn(taskChangelogPartition);
    String store1Name = "store1";
    StorageEngine store1Engine = mock(StorageEngine.class);
    StoreProperties mockStore1Properties = mock(StoreProperties.class);
    when(store1Engine.getStoreProperties()).thenReturn(mockStore1Properties);
    when(mockStore1Properties.isLoggedStore()).thenReturn(true);
    // non-persistent (in memory) store
    when(mockStore1Properties.isPersistedToDisk()).thenReturn(false);
    Map<String, StorageEngine> mockStoreEngines = ImmutableMap.of(store1Name, store1Engine);
    String changelog1SystemName = "system1";
    String changelog1StreamName = "store1Changelog";
    SystemStream changelog1SystemStream = new SystemStream(changelog1SystemName, changelog1StreamName);
    SystemStreamPartition changelog1SSP = new SystemStreamPartition(changelog1SystemStream, taskChangelogPartition);
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata("0", "10", "11");
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = "5";
    CheckpointId checkpointId = CheckpointId.create();
    KafkaStateCheckpointMarker kafkaStateCheckpointMarker = new KafkaStateCheckpointMarker(changelog1SSP, changelog1CheckpointedOffset);
    ImmutableMap<String, KafkaStateCheckpointMarker> mockCheckpointedChangelogOffset = ImmutableMap.of(store1Name, kafkaStateCheckpointMarker);
    Map<SystemStreamPartition, SystemStreamPartitionMetadata> mockCurrentChangelogOffsets = ImmutableMap.of(changelog1SSP, changelog1SSPMetadata);
    SystemAdmins mockSystemAdmins = mock(SystemAdmins.class);
    SystemAdmin mockSystemAdmin = mock(SystemAdmin.class);
    when(mockSystemAdmins.getSystemAdmin(changelog1SSP.getSystem())).thenReturn(mockSystemAdmin);
    StorageManagerUtil mockStorageManagerUtil = mock(StorageManagerUtil.class);
    File mockLoggedStoreBaseDir = mock(File.class);
    File mockNonLoggedStoreBaseDir = mock(File.class);
    Config mockConfig = mock(Config.class);
    Clock mockClock = mock(Clock.class);
    Mockito.when(mockSystemAdmin.offsetComparator(anyString(), anyString())).thenAnswer((Answer<Integer>) invocation -> {
        String offset1 = (String) invocation.getArguments()[0];
        String offset2 = (String) invocation.getArguments()[1];
        return Long.valueOf(offset1).compareTo(Long.valueOf(offset2));
    });
    StoreActions storeActions = TransactionalStateTaskRestoreManager.getStoreActions(mockTaskModel, mockStoreEngines, mockStoreChangelogs, mockCheckpointedChangelogOffset, checkpointId, mockCurrentChangelogOffsets, mockSystemAdmins, mockStorageManagerUtil, mockLoggedStoreBaseDir, mockNonLoggedStoreBaseDir, mockConfig, mockClock);
    // ensure that there is nothing to delete or retain
    assertEquals(0, storeActions.storeDirsToDelete.size());
    // ensure that we restore from the oldest changelog offset to checkpointed changelog offset
    assertEquals("0", storeActions.storesToRestore.get(store1Name).startingOffset);
    assertEquals(changelog1CheckpointedOffset, storeActions.storesToRestore.get(store1Name).endingOffset);
}
Also used : ArrayListMultimap(com.google.common.collect.ArrayListMultimap) ListMultimap(com.google.common.collect.ListMultimap) SSPMetadataCache(org.apache.samza.system.SSPMetadataCache) HashMap(java.util.HashMap) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Matchers.anyString(org.mockito.Matchers.anyString) FileUtil(org.apache.samza.util.FileUtil) Answer(org.mockito.stubbing.Answer) ImmutableList(com.google.common.collect.ImmutableList) SystemConsumer(org.apache.samza.system.SystemConsumer) Mockito.verifyNoMoreInteractions(org.mockito.Mockito.verifyNoMoreInteractions) SystemStream(org.apache.samza.system.SystemStream) Matchers.eq(org.mockito.Matchers.eq) Map(java.util.Map) Assert.fail(org.junit.Assert.fail) Path(java.nio.file.Path) MapConfig(org.apache.samza.config.MapConfig) ImmutableSet(com.google.common.collect.ImmutableSet) TaskName(org.apache.samza.container.TaskName) ImmutableMap(com.google.common.collect.ImmutableMap) TaskConfig(org.apache.samza.config.TaskConfig) Assert.assertNotNull(org.junit.Assert.assertNotNull) Partition(org.apache.samza.Partition) Set(java.util.Set) Assert.assertTrue(org.junit.Assert.assertTrue) Clock(org.apache.samza.util.Clock) Test(org.junit.Test) Mockito.times(org.mockito.Mockito.times) Mockito.when(org.mockito.Mockito.when) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) File(java.io.File) CheckpointId(org.apache.samza.checkpoint.CheckpointId) Mockito.verify(org.mockito.Mockito.verify) Matchers.any(org.mockito.Matchers.any) Mockito(org.mockito.Mockito) TaskMode(org.apache.samza.job.model.TaskMode) Mockito.never(org.mockito.Mockito.never) Assert.assertNull(org.junit.Assert.assertNull) RestoreOffsets(org.apache.samza.storage.TransactionalStateTaskRestoreManager.RestoreOffsets) SystemAdmin(org.apache.samza.system.SystemAdmin) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Config(org.apache.samza.config.Config) Collections(java.util.Collections) Assert.assertEquals(org.junit.Assert.assertEquals) SystemAdmins(org.apache.samza.system.SystemAdmins) Mockito.mock(org.mockito.Mockito.mock) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Partition(org.apache.samza.Partition) SystemStream(org.apache.samza.system.SystemStream) MapConfig(org.apache.samza.config.MapConfig) TaskConfig(org.apache.samza.config.TaskConfig) Config(org.apache.samza.config.Config) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) Matchers.anyString(org.mockito.Matchers.anyString) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) Clock(org.apache.samza.util.Clock) TaskName(org.apache.samza.container.TaskName) CheckpointId(org.apache.samza.checkpoint.CheckpointId) SystemAdmin(org.apache.samza.system.SystemAdmin) SystemAdmins(org.apache.samza.system.SystemAdmins) File(java.io.File) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Test(org.junit.Test)

Example 32 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedNonPersistentStore_FullRestoreIfCheckpointedOffsetInRangeButMaybeCompacted.

/**
 * This can happen if the changelog offset is valid but the checkpoint is older than min compaction lag ms. E.g., when
 * the job/container shut down and restarted after a long time.
 */
@Test
public void testGetStoreActionsForLoggedNonPersistentStore_FullRestoreIfCheckpointedOffsetInRangeButMaybeCompacted() {
    TaskModel mockTaskModel = mock(TaskModel.class);
    TaskName taskName = new TaskName("Partition 0");
    when(mockTaskModel.getTaskName()).thenReturn(taskName);
    Partition taskChangelogPartition = new Partition(0);
    when(mockTaskModel.getChangelogPartition()).thenReturn(taskChangelogPartition);
    String store1Name = "store1";
    StorageEngine store1Engine = mock(StorageEngine.class);
    StoreProperties mockStore1Properties = mock(StoreProperties.class);
    when(store1Engine.getStoreProperties()).thenReturn(mockStore1Properties);
    when(mockStore1Properties.isLoggedStore()).thenReturn(true);
    // non-persistent store
    when(mockStore1Properties.isPersistedToDisk()).thenReturn(false);
    Map<String, StorageEngine> mockStoreEngines = ImmutableMap.of(store1Name, store1Engine);
    String changelog1SystemName = "system1";
    String changelog1StreamName = "store1Changelog";
    SystemStream changelog1SystemStream = new SystemStream(changelog1SystemName, changelog1StreamName);
    SystemStreamPartition changelog1SSP = new SystemStreamPartition(changelog1SystemStream, taskChangelogPartition);
    // checkpointed changelog offset > newest offset (e.g. changelog topic got changed)
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata("10", "20", "21");
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = "5";
    // checkpoint id older than default min.compaction.lag.ms
    CheckpointId checkpointId = CheckpointId.deserialize("0-0");
    KafkaStateCheckpointMarker kafkaStateCheckpointMarker = new KafkaStateCheckpointMarker(changelog1SSP, changelog1CheckpointedOffset);
    Map<String, KafkaStateCheckpointMarker> mockCheckpointedChangelogOffset = new HashMap<String, KafkaStateCheckpointMarker>() {

        {
            put(store1Name, kafkaStateCheckpointMarker);
        }
    };
    Map<SystemStreamPartition, SystemStreamPartitionMetadata> mockCurrentChangelogOffsets = ImmutableMap.of(changelog1SSP, changelog1SSPMetadata);
    SystemAdmins mockSystemAdmins = mock(SystemAdmins.class);
    SystemAdmin mockSystemAdmin = mock(SystemAdmin.class);
    when(mockSystemAdmins.getSystemAdmin(changelog1SSP.getSystem())).thenReturn(mockSystemAdmin);
    StorageManagerUtil mockStorageManagerUtil = mock(StorageManagerUtil.class);
    File mockLoggedStoreBaseDir = mock(File.class);
    File mockNonLoggedStoreBaseDir = mock(File.class);
    Config mockConfig = mock(Config.class);
    Clock mockClock = mock(Clock.class);
    Mockito.when(mockSystemAdmin.offsetComparator(anyString(), anyString())).thenAnswer((Answer<Integer>) invocation -> {
        String offset1 = (String) invocation.getArguments()[0];
        String offset2 = (String) invocation.getArguments()[1];
        return Long.valueOf(offset1).compareTo(Long.valueOf(offset2));
    });
    StoreActions storeActions = TransactionalStateTaskRestoreManager.getStoreActions(mockTaskModel, mockStoreEngines, mockStoreChangelogs, mockCheckpointedChangelogOffset, checkpointId, mockCurrentChangelogOffsets, mockSystemAdmins, mockStorageManagerUtil, mockLoggedStoreBaseDir, mockNonLoggedStoreBaseDir, mockConfig, mockClock);
    // ensure that there is nothing to retain or delete
    assertEquals(0, storeActions.storeDirsToDelete.size());
    assertEquals(0, storeActions.storeDirsToRetain.size());
    // ensure that we mark the store for full restore (from current oldest to current newest)
    assertEquals("10", storeActions.storesToRestore.get(store1Name).startingOffset);
    assertEquals("20", storeActions.storesToRestore.get(store1Name).endingOffset);
}
Also used : ArrayListMultimap(com.google.common.collect.ArrayListMultimap) ListMultimap(com.google.common.collect.ListMultimap) SSPMetadataCache(org.apache.samza.system.SSPMetadataCache) HashMap(java.util.HashMap) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Matchers.anyString(org.mockito.Matchers.anyString) FileUtil(org.apache.samza.util.FileUtil) Answer(org.mockito.stubbing.Answer) ImmutableList(com.google.common.collect.ImmutableList) SystemConsumer(org.apache.samza.system.SystemConsumer) Mockito.verifyNoMoreInteractions(org.mockito.Mockito.verifyNoMoreInteractions) SystemStream(org.apache.samza.system.SystemStream) Matchers.eq(org.mockito.Matchers.eq) Map(java.util.Map) Assert.fail(org.junit.Assert.fail) Path(java.nio.file.Path) MapConfig(org.apache.samza.config.MapConfig) ImmutableSet(com.google.common.collect.ImmutableSet) TaskName(org.apache.samza.container.TaskName) ImmutableMap(com.google.common.collect.ImmutableMap) TaskConfig(org.apache.samza.config.TaskConfig) Assert.assertNotNull(org.junit.Assert.assertNotNull) Partition(org.apache.samza.Partition) Set(java.util.Set) Assert.assertTrue(org.junit.Assert.assertTrue) Clock(org.apache.samza.util.Clock) Test(org.junit.Test) Mockito.times(org.mockito.Mockito.times) Mockito.when(org.mockito.Mockito.when) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) File(java.io.File) CheckpointId(org.apache.samza.checkpoint.CheckpointId) Mockito.verify(org.mockito.Mockito.verify) Matchers.any(org.mockito.Matchers.any) Mockito(org.mockito.Mockito) TaskMode(org.apache.samza.job.model.TaskMode) Mockito.never(org.mockito.Mockito.never) Assert.assertNull(org.junit.Assert.assertNull) RestoreOffsets(org.apache.samza.storage.TransactionalStateTaskRestoreManager.RestoreOffsets) SystemAdmin(org.apache.samza.system.SystemAdmin) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Config(org.apache.samza.config.Config) Collections(java.util.Collections) Assert.assertEquals(org.junit.Assert.assertEquals) SystemAdmins(org.apache.samza.system.SystemAdmins) Mockito.mock(org.mockito.Mockito.mock) HashMap(java.util.HashMap) MapConfig(org.apache.samza.config.MapConfig) TaskConfig(org.apache.samza.config.TaskConfig) Config(org.apache.samza.config.Config) Matchers.anyString(org.mockito.Matchers.anyString) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) Clock(org.apache.samza.util.Clock) SystemAdmins(org.apache.samza.system.SystemAdmins) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Partition(org.apache.samza.Partition) SystemStream(org.apache.samza.system.SystemStream) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) TaskName(org.apache.samza.container.TaskName) CheckpointId(org.apache.samza.checkpoint.CheckpointId) SystemAdmin(org.apache.samza.system.SystemAdmin) File(java.io.File) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Test(org.junit.Test)

Example 33 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedNonPersistentStore_FullRestoreIfNullCheckpointedOffsetAndRetainExistingChangelogState.

@Test
public void testGetStoreActionsForLoggedNonPersistentStore_FullRestoreIfNullCheckpointedOffsetAndRetainExistingChangelogState() {
    TaskModel mockTaskModel = mock(TaskModel.class);
    TaskName taskName = new TaskName("Partition 0");
    when(mockTaskModel.getTaskName()).thenReturn(taskName);
    Partition taskChangelogPartition = new Partition(0);
    when(mockTaskModel.getChangelogPartition()).thenReturn(taskChangelogPartition);
    String store1Name = "store1";
    StorageEngine store1Engine = mock(StorageEngine.class);
    StoreProperties mockStore1Properties = mock(StoreProperties.class);
    when(store1Engine.getStoreProperties()).thenReturn(mockStore1Properties);
    when(mockStore1Properties.isLoggedStore()).thenReturn(true);
    // non-persistent store
    when(mockStore1Properties.isPersistedToDisk()).thenReturn(false);
    Map<String, StorageEngine> mockStoreEngines = ImmutableMap.of(store1Name, store1Engine);
    String changelog1SystemName = "system1";
    String changelog1StreamName = "store1Changelog";
    SystemStream changelog1SystemStream = new SystemStream(changelog1SystemName, changelog1StreamName);
    SystemStreamPartition changelog1SSP = new SystemStreamPartition(changelog1SystemStream, taskChangelogPartition);
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata("0", "10", "11");
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = null;
    CheckpointId checkpointId = CheckpointId.create();
    KafkaStateCheckpointMarker kafkaStateCheckpointMarker = new KafkaStateCheckpointMarker(changelog1SSP, changelog1CheckpointedOffset);
    Map<String, KafkaStateCheckpointMarker> mockCheckpointedChangelogOffset = new HashMap<String, KafkaStateCheckpointMarker>() {

        {
            put(store1Name, kafkaStateCheckpointMarker);
        }
    };
    Map<SystemStreamPartition, SystemStreamPartitionMetadata> mockCurrentChangelogOffsets = ImmutableMap.of(changelog1SSP, changelog1SSPMetadata);
    SystemAdmins mockSystemAdmins = mock(SystemAdmins.class);
    SystemAdmin mockSystemAdmin = mock(SystemAdmin.class);
    when(mockSystemAdmins.getSystemAdmin(changelog1SSP.getSystem())).thenReturn(mockSystemAdmin);
    StorageManagerUtil mockStorageManagerUtil = mock(StorageManagerUtil.class);
    File mockLoggedStoreBaseDir = mock(File.class);
    File mockNonLoggedStoreBaseDir = mock(File.class);
    HashMap<String, String> configMap = new HashMap<>();
    configMap.put(TaskConfig.TRANSACTIONAL_STATE_RETAIN_EXISTING_STATE, "true");
    Config mockConfig = new MapConfig(configMap);
    Clock mockClock = mock(Clock.class);
    Mockito.when(mockSystemAdmin.offsetComparator(anyString(), anyString())).thenAnswer((Answer<Integer>) invocation -> {
        String offset1 = (String) invocation.getArguments()[0];
        String offset2 = (String) invocation.getArguments()[1];
        return Long.valueOf(offset1).compareTo(Long.valueOf(offset2));
    });
    StoreActions storeActions = TransactionalStateTaskRestoreManager.getStoreActions(mockTaskModel, mockStoreEngines, mockStoreChangelogs, mockCheckpointedChangelogOffset, checkpointId, mockCurrentChangelogOffsets, mockSystemAdmins, mockStorageManagerUtil, mockLoggedStoreBaseDir, mockNonLoggedStoreBaseDir, mockConfig, mockClock);
    // ensure that there is nothing to delete or retain
    assertEquals(0, storeActions.storeDirsToDelete.size());
    assertEquals(0, storeActions.storeDirsToRetain.size());
    // ensure that we mark the store for restore (full trim == restore from oldest to null)
    assertEquals("0", storeActions.storesToRestore.get(store1Name).startingOffset);
    assertEquals("10", storeActions.storesToRestore.get(store1Name).endingOffset);
}
Also used : ArrayListMultimap(com.google.common.collect.ArrayListMultimap) ListMultimap(com.google.common.collect.ListMultimap) SSPMetadataCache(org.apache.samza.system.SSPMetadataCache) HashMap(java.util.HashMap) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Matchers.anyString(org.mockito.Matchers.anyString) FileUtil(org.apache.samza.util.FileUtil) Answer(org.mockito.stubbing.Answer) ImmutableList(com.google.common.collect.ImmutableList) SystemConsumer(org.apache.samza.system.SystemConsumer) Mockito.verifyNoMoreInteractions(org.mockito.Mockito.verifyNoMoreInteractions) SystemStream(org.apache.samza.system.SystemStream) Matchers.eq(org.mockito.Matchers.eq) Map(java.util.Map) Assert.fail(org.junit.Assert.fail) Path(java.nio.file.Path) MapConfig(org.apache.samza.config.MapConfig) ImmutableSet(com.google.common.collect.ImmutableSet) TaskName(org.apache.samza.container.TaskName) ImmutableMap(com.google.common.collect.ImmutableMap) TaskConfig(org.apache.samza.config.TaskConfig) Assert.assertNotNull(org.junit.Assert.assertNotNull) Partition(org.apache.samza.Partition) Set(java.util.Set) Assert.assertTrue(org.junit.Assert.assertTrue) Clock(org.apache.samza.util.Clock) Test(org.junit.Test) Mockito.times(org.mockito.Mockito.times) Mockito.when(org.mockito.Mockito.when) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) File(java.io.File) CheckpointId(org.apache.samza.checkpoint.CheckpointId) Mockito.verify(org.mockito.Mockito.verify) Matchers.any(org.mockito.Matchers.any) Mockito(org.mockito.Mockito) TaskMode(org.apache.samza.job.model.TaskMode) Mockito.never(org.mockito.Mockito.never) Assert.assertNull(org.junit.Assert.assertNull) RestoreOffsets(org.apache.samza.storage.TransactionalStateTaskRestoreManager.RestoreOffsets) SystemAdmin(org.apache.samza.system.SystemAdmin) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Config(org.apache.samza.config.Config) Collections(java.util.Collections) Assert.assertEquals(org.junit.Assert.assertEquals) SystemAdmins(org.apache.samza.system.SystemAdmins) Mockito.mock(org.mockito.Mockito.mock) HashMap(java.util.HashMap) MapConfig(org.apache.samza.config.MapConfig) TaskConfig(org.apache.samza.config.TaskConfig) Config(org.apache.samza.config.Config) Matchers.anyString(org.mockito.Matchers.anyString) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) Clock(org.apache.samza.util.Clock) MapConfig(org.apache.samza.config.MapConfig) SystemAdmins(org.apache.samza.system.SystemAdmins) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Partition(org.apache.samza.Partition) SystemStream(org.apache.samza.system.SystemStream) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) TaskName(org.apache.samza.container.TaskName) CheckpointId(org.apache.samza.checkpoint.CheckpointId) SystemAdmin(org.apache.samza.system.SystemAdmin) File(java.io.File) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Test(org.junit.Test)

Example 34 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedPersistentStore_RestoreDeltaIfStaleStoreCheckpoint.

@Test
public void testGetStoreActionsForLoggedPersistentStore_RestoreDeltaIfStaleStoreCheckpoint() {
    TaskModel mockTaskModel = mock(TaskModel.class);
    TaskName taskName = new TaskName("Partition 0");
    when(mockTaskModel.getTaskName()).thenReturn(taskName);
    Partition taskChangelogPartition = new Partition(0);
    when(mockTaskModel.getChangelogPartition()).thenReturn(taskChangelogPartition);
    String store1Name = "store1";
    StorageEngine store1Engine = mock(StorageEngine.class);
    StoreProperties mockStore1Properties = mock(StoreProperties.class);
    when(store1Engine.getStoreProperties()).thenReturn(mockStore1Properties);
    when(mockStore1Properties.isLoggedStore()).thenReturn(true);
    when(mockStore1Properties.isPersistedToDisk()).thenReturn(true);
    Map<String, StorageEngine> mockStoreEngines = ImmutableMap.of(store1Name, store1Engine);
    String changelog1SystemName = "system1";
    String changelog1StreamName = "store1Changelog";
    SystemStream changelog1SystemStream = new SystemStream(changelog1SystemName, changelog1StreamName);
    SystemStreamPartition changelog1SSP = new SystemStreamPartition(changelog1SystemStream, taskChangelogPartition);
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata("0", "10", "11");
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = "5";
    CheckpointId checkpointId = CheckpointId.create();
    KafkaStateCheckpointMarker kafkaStateCheckpointMarker = new KafkaStateCheckpointMarker(changelog1SSP, changelog1CheckpointedOffset);
    ImmutableMap<String, KafkaStateCheckpointMarker> mockCheckpointedChangelogOffset = ImmutableMap.of(store1Name, kafkaStateCheckpointMarker);
    Map<SystemStreamPartition, SystemStreamPartitionMetadata> mockCurrentChangelogOffsets = ImmutableMap.of(changelog1SSP, changelog1SSPMetadata);
    SystemAdmins mockSystemAdmins = mock(SystemAdmins.class);
    SystemAdmin mockSystemAdmin = mock(SystemAdmin.class);
    when(mockSystemAdmins.getSystemAdmin(changelog1SSP.getSystem())).thenReturn(mockSystemAdmin);
    StorageManagerUtil mockStorageManagerUtil = mock(StorageManagerUtil.class);
    File mockLoggedStoreBaseDir = mock(File.class);
    File mockNonLoggedStoreBaseDir = mock(File.class);
    Config mockConfig = mock(Config.class);
    Clock mockClock = mock(Clock.class);
    File mockCurrentStoreDir = mock(File.class);
    File mockStoreNewerCheckpointDir = mock(File.class);
    String newerCheckpointDirLocalOffset = "4";
    File mockStoreOlderCheckpointDir = mock(File.class);
    String olderCheckpointDirLocalOffset = "3";
    when(mockStorageManagerUtil.getTaskStoreDir(eq(mockLoggedStoreBaseDir), eq(store1Name), eq(taskName), any())).thenReturn(mockCurrentStoreDir);
    when(mockStorageManagerUtil.getTaskStoreCheckpointDirs(eq(mockLoggedStoreBaseDir), eq(store1Name), eq(taskName), any())).thenReturn(ImmutableList.of(mockStoreNewerCheckpointDir, mockStoreOlderCheckpointDir));
    when(mockStorageManagerUtil.isLoggedStoreValid(eq(store1Name), eq(mockStoreNewerCheckpointDir), any(), eq(mockStoreChangelogs), eq(mockTaskModel), any(), eq(mockStoreEngines))).thenReturn(true);
    when(mockStorageManagerUtil.isLoggedStoreValid(eq(store1Name), eq(mockStoreOlderCheckpointDir), any(), eq(mockStoreChangelogs), eq(mockTaskModel), any(), eq(mockStoreEngines))).thenReturn(true);
    Set<SystemStreamPartition> mockChangelogSSPs = ImmutableSet.of(changelog1SSP);
    when(mockStorageManagerUtil.readOffsetFile(eq(mockStoreNewerCheckpointDir), eq(mockChangelogSSPs), eq(false))).thenReturn(ImmutableMap.of(changelog1SSP, newerCheckpointDirLocalOffset));
    when(mockStorageManagerUtil.readOffsetFile(eq(mockStoreOlderCheckpointDir), eq(mockChangelogSSPs), eq(false))).thenReturn(// less than checkpointed offset (5)
    ImmutableMap.of(changelog1SSP, olderCheckpointDirLocalOffset));
    Mockito.when(mockSystemAdmin.offsetComparator(anyString(), anyString())).thenAnswer((Answer<Integer>) invocation -> {
        String offset1 = (String) invocation.getArguments()[0];
        String offset2 = (String) invocation.getArguments()[1];
        return Long.valueOf(offset1).compareTo(Long.valueOf(offset2));
    });
    StoreActions storeActions = TransactionalStateTaskRestoreManager.getStoreActions(mockTaskModel, mockStoreEngines, mockStoreChangelogs, mockCheckpointedChangelogOffset, checkpointId, mockCurrentChangelogOffsets, mockSystemAdmins, mockStorageManagerUtil, mockLoggedStoreBaseDir, mockNonLoggedStoreBaseDir, mockConfig, mockClock);
    // ensure that both the current dir and older stale checkpoint dir are marked for deletion
    assertEquals(2, storeActions.storeDirsToDelete.get(store1Name).size());
    assertTrue(storeActions.storeDirsToDelete.get(store1Name).contains(mockCurrentStoreDir));
    assertTrue(storeActions.storeDirsToDelete.get(store1Name).contains(mockStoreOlderCheckpointDir));
    // ensure that the newer (but still stale) store checkpoint is marked for retention
    assertEquals(mockStoreNewerCheckpointDir, storeActions.storeDirsToRetain.get(store1Name));
    // ensure that we mark the store for restore from newer (but still stale) local offset to checkpointed offset
    assertEquals(newerCheckpointDirLocalOffset, storeActions.storesToRestore.get(store1Name).startingOffset);
    assertEquals(changelog1CheckpointedOffset, storeActions.storesToRestore.get(store1Name).endingOffset);
}
Also used : ArrayListMultimap(com.google.common.collect.ArrayListMultimap) ListMultimap(com.google.common.collect.ListMultimap) SSPMetadataCache(org.apache.samza.system.SSPMetadataCache) HashMap(java.util.HashMap) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Matchers.anyString(org.mockito.Matchers.anyString) FileUtil(org.apache.samza.util.FileUtil) Answer(org.mockito.stubbing.Answer) ImmutableList(com.google.common.collect.ImmutableList) SystemConsumer(org.apache.samza.system.SystemConsumer) Mockito.verifyNoMoreInteractions(org.mockito.Mockito.verifyNoMoreInteractions) SystemStream(org.apache.samza.system.SystemStream) Matchers.eq(org.mockito.Matchers.eq) Map(java.util.Map) Assert.fail(org.junit.Assert.fail) Path(java.nio.file.Path) MapConfig(org.apache.samza.config.MapConfig) ImmutableSet(com.google.common.collect.ImmutableSet) TaskName(org.apache.samza.container.TaskName) ImmutableMap(com.google.common.collect.ImmutableMap) TaskConfig(org.apache.samza.config.TaskConfig) Assert.assertNotNull(org.junit.Assert.assertNotNull) Partition(org.apache.samza.Partition) Set(java.util.Set) Assert.assertTrue(org.junit.Assert.assertTrue) Clock(org.apache.samza.util.Clock) Test(org.junit.Test) Mockito.times(org.mockito.Mockito.times) Mockito.when(org.mockito.Mockito.when) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) File(java.io.File) CheckpointId(org.apache.samza.checkpoint.CheckpointId) Mockito.verify(org.mockito.Mockito.verify) Matchers.any(org.mockito.Matchers.any) Mockito(org.mockito.Mockito) TaskMode(org.apache.samza.job.model.TaskMode) Mockito.never(org.mockito.Mockito.never) Assert.assertNull(org.junit.Assert.assertNull) RestoreOffsets(org.apache.samza.storage.TransactionalStateTaskRestoreManager.RestoreOffsets) SystemAdmin(org.apache.samza.system.SystemAdmin) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Config(org.apache.samza.config.Config) Collections(java.util.Collections) Assert.assertEquals(org.junit.Assert.assertEquals) SystemAdmins(org.apache.samza.system.SystemAdmins) Mockito.mock(org.mockito.Mockito.mock) MapConfig(org.apache.samza.config.MapConfig) TaskConfig(org.apache.samza.config.TaskConfig) Config(org.apache.samza.config.Config) Matchers.anyString(org.mockito.Matchers.anyString) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) Clock(org.apache.samza.util.Clock) SystemAdmins(org.apache.samza.system.SystemAdmins) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Partition(org.apache.samza.Partition) SystemStream(org.apache.samza.system.SystemStream) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) TaskName(org.apache.samza.container.TaskName) CheckpointId(org.apache.samza.checkpoint.CheckpointId) SystemAdmin(org.apache.samza.system.SystemAdmin) File(java.io.File) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Test(org.junit.Test)

Example 35 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedPersistentStore_FullRestoreIfNullCheckpointedAndOldestOffset.

/**
 * This can be the case if the changelog topic is empty (although KafkaSystemAdmin returns 0 as the oldest offset
 * instead of null for empty topics).
 */
@Test
public void testGetStoreActionsForLoggedPersistentStore_FullRestoreIfNullCheckpointedAndOldestOffset() {
    // full restore == clear existing state
    TaskModel mockTaskModel = mock(TaskModel.class);
    TaskName taskName = new TaskName("Partition 0");
    when(mockTaskModel.getTaskName()).thenReturn(taskName);
    Partition taskChangelogPartition = new Partition(0);
    when(mockTaskModel.getChangelogPartition()).thenReturn(taskChangelogPartition);
    String store1Name = "store1";
    StorageEngine store1Engine = mock(StorageEngine.class);
    StoreProperties mockStore1Properties = mock(StoreProperties.class);
    when(store1Engine.getStoreProperties()).thenReturn(mockStore1Properties);
    when(mockStore1Properties.isLoggedStore()).thenReturn(true);
    when(mockStore1Properties.isPersistedToDisk()).thenReturn(true);
    Map<String, StorageEngine> mockStoreEngines = ImmutableMap.of(store1Name, store1Engine);
    String changelog1SystemName = "system1";
    String changelog1StreamName = "store1Changelog";
    SystemStream changelog1SystemStream = new SystemStream(changelog1SystemName, changelog1StreamName);
    SystemStreamPartition changelog1SSP = new SystemStreamPartition(changelog1SystemStream, taskChangelogPartition);
    // SSPMetadata contract allows for (and recommends) null as the oldest offset for empty streams.
    // KafkaSystemAdmin does not follow this convention and returns 0 instead, but we should test for this
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata(null, null, null);
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = null;
    CheckpointId checkpointId = CheckpointId.create();
    KafkaStateCheckpointMarker kafkaStateCheckpointMarker = new KafkaStateCheckpointMarker(changelog1SSP, changelog1CheckpointedOffset);
    Map<String, KafkaStateCheckpointMarker> mockCheckpointedChangelogOffset = new HashMap<String, KafkaStateCheckpointMarker>() {

        {
            put(store1Name, kafkaStateCheckpointMarker);
        }
    };
    Map<SystemStreamPartition, SystemStreamPartitionMetadata> mockCurrentChangelogOffsets = ImmutableMap.of(changelog1SSP, changelog1SSPMetadata);
    SystemAdmins mockSystemAdmins = mock(SystemAdmins.class);
    SystemAdmin mockSystemAdmin = mock(SystemAdmin.class);
    when(mockSystemAdmins.getSystemAdmin(changelog1SSP.getSystem())).thenReturn(mockSystemAdmin);
    StorageManagerUtil mockStorageManagerUtil = mock(StorageManagerUtil.class);
    File mockLoggedStoreBaseDir = mock(File.class);
    File mockNonLoggedStoreBaseDir = mock(File.class);
    HashMap<String, String> configMap = new HashMap<>();
    // should not matter
    configMap.put(TaskConfig.TRANSACTIONAL_STATE_RETAIN_EXISTING_STATE, "true");
    Config mockConfig = new MapConfig(configMap);
    Clock mockClock = mock(Clock.class);
    File mockCurrentStoreDir = mock(File.class);
    File mockStoreNewerCheckpointDir = mock(File.class);
    File mockStoreOlderCheckpointDir = mock(File.class);
    String olderCheckpointDirLocalOffset = "3";
    String newerCheckpointDirLocalOffset = "5";
    when(mockStorageManagerUtil.getTaskStoreDir(eq(mockLoggedStoreBaseDir), eq(store1Name), eq(taskName), any())).thenReturn(mockCurrentStoreDir);
    when(mockStorageManagerUtil.getTaskStoreCheckpointDirs(eq(mockLoggedStoreBaseDir), eq(store1Name), eq(taskName), any())).thenReturn(ImmutableList.of(mockStoreNewerCheckpointDir, mockStoreOlderCheckpointDir));
    when(mockStorageManagerUtil.isLoggedStoreValid(eq(store1Name), eq(mockStoreNewerCheckpointDir), any(), eq(mockStoreChangelogs), eq(mockTaskModel), any(), eq(mockStoreEngines))).thenReturn(true);
    when(mockStorageManagerUtil.isLoggedStoreValid(eq(store1Name), eq(mockStoreOlderCheckpointDir), any(), eq(mockStoreChangelogs), eq(mockTaskModel), any(), eq(mockStoreEngines))).thenReturn(true);
    Set<SystemStreamPartition> mockChangelogSSPs = ImmutableSet.of(changelog1SSP);
    when(mockStorageManagerUtil.readOffsetFile(eq(mockStoreNewerCheckpointDir), eq(mockChangelogSSPs), eq(false))).thenReturn(ImmutableMap.of(changelog1SSP, newerCheckpointDirLocalOffset));
    when(mockStorageManagerUtil.readOffsetFile(eq(mockStoreOlderCheckpointDir), eq(mockChangelogSSPs), eq(false))).thenReturn(// less than checkpointed offset (5)
    ImmutableMap.of(changelog1SSP, olderCheckpointDirLocalOffset));
    Mockito.when(mockSystemAdmin.offsetComparator(anyString(), anyString())).thenAnswer((Answer<Integer>) invocation -> {
        String offset1 = (String) invocation.getArguments()[0];
        String offset2 = (String) invocation.getArguments()[1];
        if (offset1 == null || offset2 == null) {
            return -1;
        }
        return Long.valueOf(offset1).compareTo(Long.valueOf(offset2));
    });
    StoreActions storeActions = TransactionalStateTaskRestoreManager.getStoreActions(mockTaskModel, mockStoreEngines, mockStoreChangelogs, mockCheckpointedChangelogOffset, checkpointId, mockCurrentChangelogOffsets, mockSystemAdmins, mockStorageManagerUtil, mockLoggedStoreBaseDir, mockNonLoggedStoreBaseDir, mockConfig, mockClock);
    // ensure that all the store dirs (current or checkpoint) are marked for deletion
    assertEquals(3, storeActions.storeDirsToDelete.get(store1Name).size());
    assertTrue(storeActions.storeDirsToDelete.get(store1Name).contains(mockCurrentStoreDir));
    assertTrue(storeActions.storeDirsToDelete.get(store1Name).contains(mockStoreOlderCheckpointDir));
    assertTrue(storeActions.storeDirsToDelete.get(store1Name).contains(mockStoreNewerCheckpointDir));
    // ensure that no directories are retained
    assertEquals(0, storeActions.storeDirsToRetain.size());
    // ensure that we do a full restore (on the empty topic)
    assertNull(storeActions.storesToRestore.get(store1Name).startingOffset);
    assertNull(storeActions.storesToRestore.get(store1Name).endingOffset);
}
Also used : ArrayListMultimap(com.google.common.collect.ArrayListMultimap) ListMultimap(com.google.common.collect.ListMultimap) SSPMetadataCache(org.apache.samza.system.SSPMetadataCache) HashMap(java.util.HashMap) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Matchers.anyString(org.mockito.Matchers.anyString) FileUtil(org.apache.samza.util.FileUtil) Answer(org.mockito.stubbing.Answer) ImmutableList(com.google.common.collect.ImmutableList) SystemConsumer(org.apache.samza.system.SystemConsumer) Mockito.verifyNoMoreInteractions(org.mockito.Mockito.verifyNoMoreInteractions) SystemStream(org.apache.samza.system.SystemStream) Matchers.eq(org.mockito.Matchers.eq) Map(java.util.Map) Assert.fail(org.junit.Assert.fail) Path(java.nio.file.Path) MapConfig(org.apache.samza.config.MapConfig) ImmutableSet(com.google.common.collect.ImmutableSet) TaskName(org.apache.samza.container.TaskName) ImmutableMap(com.google.common.collect.ImmutableMap) TaskConfig(org.apache.samza.config.TaskConfig) Assert.assertNotNull(org.junit.Assert.assertNotNull) Partition(org.apache.samza.Partition) Set(java.util.Set) Assert.assertTrue(org.junit.Assert.assertTrue) Clock(org.apache.samza.util.Clock) Test(org.junit.Test) Mockito.times(org.mockito.Mockito.times) Mockito.when(org.mockito.Mockito.when) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) File(java.io.File) CheckpointId(org.apache.samza.checkpoint.CheckpointId) Mockito.verify(org.mockito.Mockito.verify) Matchers.any(org.mockito.Matchers.any) Mockito(org.mockito.Mockito) TaskMode(org.apache.samza.job.model.TaskMode) Mockito.never(org.mockito.Mockito.never) Assert.assertNull(org.junit.Assert.assertNull) RestoreOffsets(org.apache.samza.storage.TransactionalStateTaskRestoreManager.RestoreOffsets) SystemAdmin(org.apache.samza.system.SystemAdmin) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Config(org.apache.samza.config.Config) Collections(java.util.Collections) Assert.assertEquals(org.junit.Assert.assertEquals) SystemAdmins(org.apache.samza.system.SystemAdmins) Mockito.mock(org.mockito.Mockito.mock) HashMap(java.util.HashMap) MapConfig(org.apache.samza.config.MapConfig) TaskConfig(org.apache.samza.config.TaskConfig) Config(org.apache.samza.config.Config) Matchers.anyString(org.mockito.Matchers.anyString) SystemStreamPartitionMetadata(org.apache.samza.system.SystemStreamMetadata.SystemStreamPartitionMetadata) Clock(org.apache.samza.util.Clock) MapConfig(org.apache.samza.config.MapConfig) SystemAdmins(org.apache.samza.system.SystemAdmins) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Partition(org.apache.samza.Partition) SystemStream(org.apache.samza.system.SystemStream) StoreActions(org.apache.samza.storage.TransactionalStateTaskRestoreManager.StoreActions) TaskName(org.apache.samza.container.TaskName) CheckpointId(org.apache.samza.checkpoint.CheckpointId) SystemAdmin(org.apache.samza.system.SystemAdmin) File(java.io.File) TaskModel(org.apache.samza.job.model.TaskModel) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) KafkaStateCheckpointMarker(org.apache.samza.checkpoint.kafka.KafkaStateCheckpointMarker) Test(org.junit.Test)

Aggregations

CheckpointId (org.apache.samza.checkpoint.CheckpointId)53 Test (org.junit.Test)48 File (java.io.File)44 HashMap (java.util.HashMap)43 Map (java.util.Map)42 Path (java.nio.file.Path)41 ImmutableMap (com.google.common.collect.ImmutableMap)40 TaskName (org.apache.samza.container.TaskName)38 Collections (java.util.Collections)36 Partition (org.apache.samza.Partition)34 SystemStreamPartition (org.apache.samza.system.SystemStreamPartition)34 Answer (org.mockito.stubbing.Answer)34 MapConfig (org.apache.samza.config.MapConfig)31 TaskMode (org.apache.samza.job.model.TaskMode)30 FileUtil (org.apache.samza.util.FileUtil)30 Mockito (org.mockito.Mockito)30 Set (java.util.Set)29 ImmutableList (com.google.common.collect.ImmutableList)28 ImmutableSet (com.google.common.collect.ImmutableSet)28 SystemStream (org.apache.samza.system.SystemStream)27