Search in sources :

Example 26 with CheckpointId

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

the class TestBlobStoreUtil method testRemoveTTL.

@Test
public void testRemoveTTL() throws IOException, ExecutionException, InterruptedException {
    BlobStoreManager blobStoreManager = mock(BlobStoreManager.class);
    // File, dir and recursive dir added, retained and removed in local
    // Using unique file names since test setup returns it as the blob id
    String local = "[a, c, z/1, y/2, p/m/3, q/n/4]";
    String remote = "[a, b, z/1, x/5, p/m/3, r/o/6]";
    String expectedAdded = "[c, y/2, q/n/4]";
    String expectedRetained = "[a, z/1, p/m/3]";
    SortedSet<String> expectedAddedFiles = BlobStoreTestUtil.getExpected(expectedAdded);
    SortedSet<String> expectedRetainedFiles = BlobStoreTestUtil.getExpected(expectedRetained);
    SortedSet<String> expectedPresentFiles = new TreeSet<>(expectedAddedFiles);
    expectedPresentFiles.addAll(expectedRetainedFiles);
    // Set up environment
    Path localSnapshotDir = BlobStoreTestUtil.createLocalDir(local);
    String basePath = localSnapshotDir.toAbsolutePath().toString();
    DirIndex remoteSnapshotDir = BlobStoreTestUtil.createDirIndex(remote);
    SnapshotMetadata snapshotMetadata = new SnapshotMetadata(checkpointId, jobName, jobId, taskName, storeName);
    DirDiff dirDiff = DirDiffUtil.getDirDiff(localSnapshotDir.toFile(), remoteSnapshotDir, (localFile, remoteFile) -> localFile.getName().equals(remoteFile.getFileName()));
    when(blobStoreManager.put(any(InputStream.class), any(Metadata.class))).thenAnswer((Answer<CompletableFuture<String>>) invocation -> {
        Metadata metadata = invocation.getArgumentAt(1, Metadata.class);
        String path = metadata.getPayloadPath();
        String fileName = path.substring(path.length() - 1);
        return CompletableFuture.completedFuture(fileName);
    });
    BlobStoreUtil blobStoreUtil = new BlobStoreUtil(blobStoreManager, EXECUTOR, null, null);
    CompletionStage<DirIndex> dirIndexFuture = blobStoreUtil.putDir(dirDiff, snapshotMetadata);
    DirIndex dirIndex = null;
    try {
        // should be already complete. if not, future composition in putDir is broken.
        dirIndex = dirIndexFuture.toCompletableFuture().get(0, TimeUnit.MILLISECONDS);
    } catch (TimeoutException e) {
        fail("Future returned from putDir should be already complete.");
    }
    SnapshotIndex mockSnapshotIndex = mock(SnapshotIndex.class);
    when(mockSnapshotIndex.getSnapshotMetadata()).thenReturn(snapshotMetadata);
    when(mockSnapshotIndex.getDirIndex()).thenReturn(dirIndex);
    SortedSet<String> allTTLRemoved = new TreeSet<>();
    when(blobStoreManager.removeTTL(anyString(), any(Metadata.class))).thenAnswer((Answer<CompletableFuture<String>>) invocation -> {
        String blobId = invocation.getArgumentAt(0, String.class);
        allTTLRemoved.add(blobId);
        return CompletableFuture.completedFuture(null);
    });
    // Execute
    blobStoreUtil.removeTTL("snapshotIndexBlobId", mockSnapshotIndex, metadata);
    // Assert
    SortedSet<String> expectedBlobIds = new TreeSet<>();
    // test uses unique file name (last char) as the blob ID.
    expectedPresentFiles.forEach(f -> expectedBlobIds.add(f.substring(f.length() - 1)));
    expectedBlobIds.add("snapshotIndexBlobId");
    assertEquals(expectedBlobIds, allTTLRemoved);
}
Also used : Path(java.nio.file.Path) SortedSet(java.util.SortedSet) FileMetadata(org.apache.samza.storage.blobstore.index.FileMetadata) FileTime(java.nio.file.attribute.FileTime) TimeoutException(java.util.concurrent.TimeoutException) Random(java.util.Random) RetriableException(org.apache.samza.storage.blobstore.exceptions.RetriableException) FileUtil(org.apache.samza.util.FileUtil) Pair(org.apache.commons.lang3.tuple.Pair) Map(java.util.Map) Path(java.nio.file.Path) FutureUtil(org.apache.samza.util.FutureUtil) ImmutableSet(com.google.common.collect.ImmutableSet) PosixFileAttributes(java.nio.file.attribute.PosixFileAttributes) ImmutableMap(com.google.common.collect.ImmutableMap) Set(java.util.Set) CompletionException(java.util.concurrent.CompletionException) Checkpoint(org.apache.samza.checkpoint.Checkpoint) DirDiff(org.apache.samza.storage.blobstore.diff.DirDiff) CheckpointId(org.apache.samza.checkpoint.CheckpointId) IOUtils(org.apache.commons.io.IOUtils) List(java.util.List) CompletionStage(java.util.concurrent.CompletionStage) SnapshotIndex(org.apache.samza.storage.blobstore.index.SnapshotIndex) Optional(java.util.Optional) RandomStringUtils(org.apache.commons.lang3.RandomStringUtils) SnapshotMetadata(org.apache.samza.storage.blobstore.index.SnapshotMetadata) MoreExecutors(com.google.common.util.concurrent.MoreExecutors) DirIndex(org.apache.samza.storage.blobstore.index.DirIndex) FileBlob(org.apache.samza.storage.blobstore.index.FileBlob) Matchers(org.mockito.Matchers) CheckpointV2(org.apache.samza.checkpoint.CheckpointV2) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) TreeSet(java.util.TreeSet) ArrayList(java.util.ArrayList) HashSet(java.util.HashSet) Answer(org.mockito.stubbing.Answer) PosixFilePermissions(java.nio.file.attribute.PosixFilePermissions) ArgumentCaptor(org.mockito.ArgumentCaptor) ImmutableList(com.google.common.collect.ImmutableList) BlobStoreManager(org.apache.samza.storage.blobstore.BlobStoreManager) BlobStoreStateBackendFactory(org.apache.samza.storage.blobstore.BlobStoreStateBackendFactory) ExecutorService(java.util.concurrent.ExecutorService) OutputStream(java.io.OutputStream) FileIndex(org.apache.samza.storage.blobstore.index.FileIndex) Files(java.nio.file.Files) FileOutputStream(java.io.FileOutputStream) IOException(java.io.IOException) FileUtils(org.apache.commons.io.FileUtils) Test(org.junit.Test) Metadata(org.apache.samza.storage.blobstore.Metadata) File(java.io.File) SamzaException(org.apache.samza.SamzaException) ExecutionException(java.util.concurrent.ExecutionException) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) Ignore(org.junit.Ignore) Paths(java.nio.file.Paths) NullOutputStream(org.apache.commons.io.output.NullOutputStream) CRC32(java.util.zip.CRC32) Assert(org.junit.Assert) Collections(java.util.Collections) InputStream(java.io.InputStream) DeletedException(org.apache.samza.storage.blobstore.exceptions.DeletedException) SnapshotIndex(org.apache.samza.storage.blobstore.index.SnapshotIndex) InputStream(java.io.InputStream) FileMetadata(org.apache.samza.storage.blobstore.index.FileMetadata) SnapshotMetadata(org.apache.samza.storage.blobstore.index.SnapshotMetadata) Metadata(org.apache.samza.storage.blobstore.Metadata) DirDiff(org.apache.samza.storage.blobstore.diff.DirDiff) BlobStoreManager(org.apache.samza.storage.blobstore.BlobStoreManager) CompletableFuture(java.util.concurrent.CompletableFuture) SnapshotMetadata(org.apache.samza.storage.blobstore.index.SnapshotMetadata) TreeSet(java.util.TreeSet) DirIndex(org.apache.samza.storage.blobstore.index.DirIndex) TimeoutException(java.util.concurrent.TimeoutException) Test(org.junit.Test)

Example 27 with CheckpointId

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

the class TestBlobStoreUtil method testCleanUpFailsIfAnySubDirFileDeleteFails.

@Test
public void testCleanUpFailsIfAnySubDirFileDeleteFails() throws IOException, TimeoutException, InterruptedException, ExecutionException {
    BlobStoreManager blobStoreManager = mock(BlobStoreManager.class);
    // File, dir and recursive dir added, retained and removed in local
    // Using unique file names since test util uses only the file name (leaf node)
    // as the mock blob id, not the full file path.
    String local = "[a/1, b/2]";
    String remote = "[c/3, d/4]";
    // Set up environment
    Path localSnapshotDir = BlobStoreTestUtil.createLocalDir(local);
    String basePath = localSnapshotDir.toAbsolutePath().toString();
    DirIndex remoteSnapshotDir = BlobStoreTestUtil.createDirIndex(remote);
    SnapshotMetadata snapshotMetadata = new SnapshotMetadata(checkpointId, jobName, jobId, taskName, storeName);
    DirDiff dirDiff = DirDiffUtil.getDirDiff(localSnapshotDir.toFile(), remoteSnapshotDir, (localFile, remoteFile) -> localFile.getName().equals(remoteFile.getFileName()));
    BlobStoreUtil blobStoreUtil = new BlobStoreUtil(blobStoreManager, EXECUTOR, null, null);
    when(blobStoreManager.put(any(InputStream.class), any(Metadata.class))).thenReturn(CompletableFuture.completedFuture("blobId"));
    CompletionStage<DirIndex> dirIndexFuture = blobStoreUtil.putDir(dirDiff, snapshotMetadata);
    DirIndex dirIndex = null;
    try {
        // should be already complete. if not, future composition in putDir is broken.
        dirIndex = dirIndexFuture.toCompletableFuture().get(0, TimeUnit.MILLISECONDS);
    } catch (TimeoutException e) {
        fail("Future returned from putDir should be already complete.");
    }
    // Set up mocks
    SamzaException exception = new SamzaException("Error deleting file");
    CompletableFuture<Void> failedFuture = new CompletableFuture<>();
    failedFuture.completeExceptionally(exception);
    when(blobStoreManager.delete(anyString(), any(Metadata.class))).thenAnswer((Answer<CompletableFuture<Void>>) invocation -> {
        String blobId = invocation.getArgumentAt(0, String.class);
        if (blobId.equals("3")) {
            return CompletableFuture.completedFuture(null);
        } else {
            return failedFuture;
        }
    });
    // Execute
    CompletionStage<Void> cleanUpFuture = blobStoreUtil.cleanUpDir(dirIndex, metadata);
    try {
        // should be already complete. if not, future composition in putDir is broken.
        cleanUpFuture.toCompletableFuture().get(0, TimeUnit.MILLISECONDS);
    } catch (ExecutionException e) {
        Throwable cause = e.getCause();
        // Assert that the result future fails and that the cause is propagated correctly
        assertEquals(exception, cause);
        return;
    }
    fail("Clean up future should have been completed with an exception");
}
Also used : Path(java.nio.file.Path) SortedSet(java.util.SortedSet) FileMetadata(org.apache.samza.storage.blobstore.index.FileMetadata) FileTime(java.nio.file.attribute.FileTime) TimeoutException(java.util.concurrent.TimeoutException) Random(java.util.Random) RetriableException(org.apache.samza.storage.blobstore.exceptions.RetriableException) FileUtil(org.apache.samza.util.FileUtil) Pair(org.apache.commons.lang3.tuple.Pair) Map(java.util.Map) Path(java.nio.file.Path) FutureUtil(org.apache.samza.util.FutureUtil) ImmutableSet(com.google.common.collect.ImmutableSet) PosixFileAttributes(java.nio.file.attribute.PosixFileAttributes) ImmutableMap(com.google.common.collect.ImmutableMap) Set(java.util.Set) CompletionException(java.util.concurrent.CompletionException) Checkpoint(org.apache.samza.checkpoint.Checkpoint) DirDiff(org.apache.samza.storage.blobstore.diff.DirDiff) CheckpointId(org.apache.samza.checkpoint.CheckpointId) IOUtils(org.apache.commons.io.IOUtils) List(java.util.List) CompletionStage(java.util.concurrent.CompletionStage) SnapshotIndex(org.apache.samza.storage.blobstore.index.SnapshotIndex) Optional(java.util.Optional) RandomStringUtils(org.apache.commons.lang3.RandomStringUtils) SnapshotMetadata(org.apache.samza.storage.blobstore.index.SnapshotMetadata) MoreExecutors(com.google.common.util.concurrent.MoreExecutors) DirIndex(org.apache.samza.storage.blobstore.index.DirIndex) FileBlob(org.apache.samza.storage.blobstore.index.FileBlob) Matchers(org.mockito.Matchers) CheckpointV2(org.apache.samza.checkpoint.CheckpointV2) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) TreeSet(java.util.TreeSet) ArrayList(java.util.ArrayList) HashSet(java.util.HashSet) Answer(org.mockito.stubbing.Answer) PosixFilePermissions(java.nio.file.attribute.PosixFilePermissions) ArgumentCaptor(org.mockito.ArgumentCaptor) ImmutableList(com.google.common.collect.ImmutableList) BlobStoreManager(org.apache.samza.storage.blobstore.BlobStoreManager) BlobStoreStateBackendFactory(org.apache.samza.storage.blobstore.BlobStoreStateBackendFactory) ExecutorService(java.util.concurrent.ExecutorService) OutputStream(java.io.OutputStream) FileIndex(org.apache.samza.storage.blobstore.index.FileIndex) Files(java.nio.file.Files) FileOutputStream(java.io.FileOutputStream) IOException(java.io.IOException) FileUtils(org.apache.commons.io.FileUtils) Test(org.junit.Test) Metadata(org.apache.samza.storage.blobstore.Metadata) File(java.io.File) SamzaException(org.apache.samza.SamzaException) ExecutionException(java.util.concurrent.ExecutionException) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) Ignore(org.junit.Ignore) Paths(java.nio.file.Paths) NullOutputStream(org.apache.commons.io.output.NullOutputStream) CRC32(java.util.zip.CRC32) Assert(org.junit.Assert) Collections(java.util.Collections) InputStream(java.io.InputStream) DeletedException(org.apache.samza.storage.blobstore.exceptions.DeletedException) InputStream(java.io.InputStream) FileMetadata(org.apache.samza.storage.blobstore.index.FileMetadata) SnapshotMetadata(org.apache.samza.storage.blobstore.index.SnapshotMetadata) Metadata(org.apache.samza.storage.blobstore.Metadata) DirDiff(org.apache.samza.storage.blobstore.diff.DirDiff) BlobStoreManager(org.apache.samza.storage.blobstore.BlobStoreManager) SamzaException(org.apache.samza.SamzaException) CompletableFuture(java.util.concurrent.CompletableFuture) SnapshotMetadata(org.apache.samza.storage.blobstore.index.SnapshotMetadata) DirIndex(org.apache.samza.storage.blobstore.index.DirIndex) ExecutionException(java.util.concurrent.ExecutionException) TimeoutException(java.util.concurrent.TimeoutException) Test(org.junit.Test)

Example 28 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedPersistentStore_FullRestoreIfCheckpointedOffsetOlderThanOldest.

/**
 * This can happen if the changelog topic gets compacted and the local store offset was written prior to the
 * compaction. If so, we do a full restore.
 */
@Test
public void testGetStoreActionsForLoggedPersistentStore_FullRestoreIfCheckpointedOffsetOlderThanOldest() {
    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);
    // oldest offset > checkpointed changelog offset
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata("11", "20", "21");
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = "5";
    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);
    Config mockConfig = mock(Config.class);
    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];
        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 mark the store for full restore (from current oldest to current newest)
    assertEquals("11", 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 29 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedPersistentStore_FullRestoreIfCheckpointedOffsetNewerThanNewest.

/**
 * This can happen if the changelog topic was manually deleted and recreated, and the checkpointed/local changelog
 * offset is not valid anymore.
 */
@Test
public void testGetStoreActionsForLoggedPersistentStore_FullRestoreIfCheckpointedOffsetNewerThanNewest() {
    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);
    // checkpointed changelog offset > newest offset (e.g. changelog topic got changed)
    SystemStreamPartitionMetadata changelog1SSPMetadata = new SystemStreamPartitionMetadata("0", "10", "11");
    Map<String, SystemStream> mockStoreChangelogs = ImmutableMap.of(store1Name, changelog1SystemStream);
    String changelog1CheckpointedOffset = "21";
    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);
    Config mockConfig = mock(Config.class);
    Clock mockClock = mock(Clock.class);
    File mockCurrentStoreDir = mock(File.class);
    File mockStoreNewerCheckpointDir = mock(File.class);
    File mockStoreOlderCheckpointDir = mock(File.class);
    String olderCheckpointDirLocalOffset = "5";
    String newerCheckpointDirLocalOffset = "15";
    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 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 mark the store for full restore (from current oldest to current newest)
    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) 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 30 with CheckpointId

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

the class TestTransactionalStateTaskRestoreManager method testGetStoreActionsForLoggedPersistentStore_FullTrimIfNullCheckpointedOffsetAndNotRetainExistingState.

/**
 * We need to trim the changelog topic to handle the scenario where container wrote some messages to store and
 * changelog, but died before the first commit (leaving checkpointed changelog offset as null).
 *
 * Retain existing state flag exists to support cases when user is turning on transactional support for the first
 * time and does not have an existing checkpointed changelog offset. Retain existing state flag allows them to
 * carry over the existing changelog state after a full bootstrap. Flag should be turned off after the first deploy.
 */
@Test
public void testGetStoreActionsForLoggedPersistentStore_FullTrimIfNullCheckpointedOffsetAndNotRetainExistingState() {
    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 = 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, "false");
    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];
        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 mark the store for restore (full trim == restore from oldest to null)
    assertEquals("0", 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