Examples with KinesisIndexTaskIOConfig org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig used on opensource projects

Example 16 with KinesisIndexTaskIOConfig

use of org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig in project druid by druid-io.

the class KinesisSupervisorTest method testShardMergePhaseTwo.

/**
 * Test task creation after a shard split with a closed shard
 *
 * @param phaseOneTasks List of tasks from the initial phase where only one shard was present
 */
private List<Task> testShardMergePhaseTwo(List<Task> phaseOneTasks) throws Exception {
    EasyMock.reset(indexerMetadataStorageCoordinator);
    EasyMock.reset(taskStorage);
    EasyMock.reset(taskQueue);
    EasyMock.reset(taskClient);
    EasyMock.reset(taskMaster);
    EasyMock.reset(taskRunner);
    EasyMock.reset(supervisorRecordSupplier);
    // first tasks ran, both shard 0 and shard 1 have reached EOS, merged into shard 2
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KinesisDataSourceMetadata(new SeekableStreamEndSequenceNumbers<String, String>(STREAM, ImmutableMap.of(SHARD_ID0, KinesisSequenceNumber.END_OF_SHARD_MARKER, SHARD_ID1, KinesisSequenceNumber.END_OF_SHARD_MARKER)))).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getPartitionIds(STREAM)).andReturn(ImmutableSet.of(SHARD_ID0, SHARD_ID1, SHARD_ID2)).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getAssignment()).andReturn(ImmutableSet.of(SHARD0_PARTITION, SHARD1_PARTITION, SHARD2_PARTITION)).anyTimes();
    supervisorRecordSupplier.seekToLatest(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getEarliestSequenceNumber(EasyMock.anyObject())).andReturn("0").anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getLatestSequenceNumber(new StreamPartition<>(STREAM, SHARD_ID0))).andReturn(KinesisSequenceNumber.END_OF_SHARD_MARKER).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getLatestSequenceNumber(new StreamPartition<>(STREAM, SHARD_ID1))).andReturn(KinesisSequenceNumber.END_OF_SHARD_MARKER).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getLatestSequenceNumber(new StreamPartition<>(STREAM, SHARD_ID2))).andReturn("100").anyTimes();
    supervisorRecordSupplier.seek(EasyMock.anyObject(), EasyMock.anyString());
    EasyMock.expectLastCall().anyTimes();
    Capture<Task> postMergeCaptured = Capture.newInstance(CaptureType.ALL);
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    EasyMock.expect(taskRunner.getRunningTasks()).andReturn(Collections.emptyList()).anyTimes();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(ImmutableList.of()).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.NOT_STARTED)).anyTimes();
    Task successfulTask0 = phaseOneTasks.get(0);
    Task successfulTask1 = phaseOneTasks.get(1);
    EasyMock.expect(taskStorage.getStatus(successfulTask0.getId())).andReturn(Optional.of(TaskStatus.success(successfulTask0.getId())));
    EasyMock.expect(taskStorage.getTask(successfulTask0.getId())).andReturn(Optional.of(successfulTask0)).anyTimes();
    EasyMock.expect(taskStorage.getStatus(successfulTask1.getId())).andReturn(Optional.of(TaskStatus.success(successfulTask1.getId())));
    EasyMock.expect(taskStorage.getTask(successfulTask1.getId())).andReturn(Optional.of(successfulTask1)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(postMergeCaptured))).andReturn(true).times(1);
    replayAll();
    supervisor.runInternal();
    verifyAll();
    EasyMock.reset(taskStorage);
    EasyMock.reset(taskClient);
    EasyMock.expect(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.NOT_STARTED)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).anyTimes();
    TreeMap<Integer, Map<String, String>> checkpointsGroup0 = new TreeMap<>();
    checkpointsGroup0.put(0, ImmutableMap.of(SHARD_ID2, "0", SHARD_ID1, KinesisSequenceNumber.END_OF_SHARD_MARKER));
    // there would be 1 tasks, 1 for each task group, but task group 1 only has closed shards, so no task is created
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("sequenceName-0"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpointsGroup0)).times(1);
    List<Task> postMergeTasks = postMergeCaptured.getValues();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(postMergeTasks).anyTimes();
    for (Task task : postMergeTasks) {
        EasyMock.expect(taskStorage.getStatus(task.getId())).andReturn(Optional.of(TaskStatus.running(task.getId()))).anyTimes();
        EasyMock.expect(taskStorage.getTask(task.getId())).andReturn(Optional.of(task)).anyTimes();
    }
    EasyMock.replay(taskStorage);
    EasyMock.replay(taskClient);
    supervisor.runInternal();
    verifyAll();
    // Check that shardId-000000000000 which has hit EOS is not included in the sequences sent to the task for group 0
    SeekableStreamStartSequenceNumbers<String, String> group0ExpectedStartSequenceNumbers = new SeekableStreamStartSequenceNumbers<>(STREAM, ImmutableMap.of(SHARD_ID2, "0"), ImmutableSet.of());
    SeekableStreamEndSequenceNumbers<String, String> group0ExpectedEndSequenceNumbers = new SeekableStreamEndSequenceNumbers<>(STREAM, ImmutableMap.of(SHARD_ID2, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER));
    Assert.assertEquals(1, postMergeTasks.size());
    KinesisIndexTaskIOConfig group0Config = ((KinesisIndexTask) postMergeTasks.get(0)).getIOConfig();
    Assert.assertEquals((Integer) 0, group0Config.getTaskGroupId());
    Assert.assertEquals(group0ExpectedStartSequenceNumbers, group0Config.getStartSequenceNumbers());
    Assert.assertEquals(group0ExpectedEndSequenceNumbers, group0Config.getEndSequenceNumbers());
    return postMergeTasks;
}

Example 17 with KinesisIndexTaskIOConfig

use of org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig in project druid by druid-io.

the class KinesisSupervisorTest method testNoInitialState.

@Test
public void testNoInitialState() throws Exception {
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null);
    supervisorRecordSupplier.assign(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getPartitionIds(STREAM)).andReturn(ImmutableSet.of(SHARD_ID1, SHARD_ID0)).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getAssignment()).andReturn(ImmutableSet.of(SHARD1_PARTITION, SHARD0_PARTITION)).anyTimes();
    supervisorRecordSupplier.seekToLatest(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getEarliestSequenceNumber(EasyMock.anyObject())).andReturn("0").anyTimes();
    supervisorRecordSupplier.seek(EasyMock.anyObject(), EasyMock.anyString());
    EasyMock.expectLastCall().anyTimes();
    Capture<KinesisIndexTask> captured = Capture.newInstance();
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(ImmutableList.of()).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true);
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    KinesisIndexTask task = captured.getValue();
    Assert.assertEquals(dataSchema, task.getDataSchema());
    Assert.assertEquals(tuningConfig.convertToTaskTuningConfig(), task.getTuningConfig());
    KinesisIndexTaskIOConfig taskConfig = task.getIOConfig();
    Assert.assertEquals("sequenceName-0", taskConfig.getBaseSequenceName());
    Assert.assertTrue("isUseTransaction", taskConfig.isUseTransaction());
    Assert.assertFalse("minimumMessageTime", taskConfig.getMinimumMessageTime().isPresent());
    Assert.assertFalse("maximumMessageTime", taskConfig.getMaximumMessageTime().isPresent());
    Assert.assertEquals(STREAM, taskConfig.getStartSequenceNumbers().getStream());
    Assert.assertEquals("0", taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals("0", taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
    Assert.assertEquals(STREAM, taskConfig.getEndSequenceNumbers().getStream());
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
}

Example 18 with KinesisIndexTaskIOConfig

use of org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig in project druid by druid-io.

the class KinesisSupervisorTest method testKillUnresponsiveTasksWhilePausing.

@Test
public void testKillUnresponsiveTasksWhilePausing() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234, -1);
    supervisor = getTestableSupervisor(2, 2, true, "PT1M", null, null);
    supervisorRecordSupplier.assign(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getPartitionIds(STREAM)).andReturn(ImmutableSet.of(SHARD_ID1, SHARD_ID0)).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getAssignment()).andReturn(ImmutableSet.of(SHARD1_PARTITION, SHARD0_PARTITION)).anyTimes();
    supervisorRecordSupplier.seekToLatest(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getEarliestSequenceNumber(EasyMock.anyObject())).andReturn("0").anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getLatestSequenceNumber(SHARD1_PARTITION)).andReturn("12").anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getLatestSequenceNumber(SHARD0_PARTITION)).andReturn("1").anyTimes();
    supervisorRecordSupplier.seek(EasyMock.anyObject(), EasyMock.anyString());
    EasyMock.expectLastCall().anyTimes();
    Capture<Task> captured = Capture.newInstance(CaptureType.ALL);
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    EasyMock.expect(taskRunner.getRunningTasks()).andReturn(Collections.emptyList()).anyTimes();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(ImmutableList.of()).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true).times(4);
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    List<Task> tasks = captured.getValues();
    Collection workItems = new ArrayList<>();
    for (Task task : tasks) {
        workItems.add(new TestTaskRunnerWorkItem(task, null, location));
    }
    EasyMock.reset(taskStorage, taskRunner, taskClient, taskQueue);
    TreeMap<Integer, Map<String, String>> checkpoints1 = new TreeMap<>();
    checkpoints1.put(0, ImmutableMap.of(SHARD_ID1, "0"));
    TreeMap<Integer, Map<String, String>> checkpoints2 = new TreeMap<>();
    checkpoints2.put(0, ImmutableMap.of(SHARD_ID0, "0"));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("sequenceName-0"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints1)).times(2);
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("sequenceName-1"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints2)).times(2);
    captured = Capture.newInstance(CaptureType.ALL);
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(tasks).anyTimes();
    for (Task task : tasks) {
        EasyMock.expect(taskStorage.getStatus(task.getId())).andReturn(Optional.of(TaskStatus.running(task.getId()))).anyTimes();
        EasyMock.expect(taskStorage.getTask(task.getId())).andReturn(Optional.of(task)).anyTimes();
    }
    EasyMock.expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.contains("sequenceName-0"))).andReturn(Futures.immediateFuture(DateTimes.nowUtc().minusMinutes(2))).andReturn(Futures.immediateFuture(DateTimes.nowUtc()));
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.contains("sequenceName-1"))).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).times(2);
    EasyMock.expect(taskClient.pauseAsync(EasyMock.contains("sequenceName-0"))).andReturn(Futures.immediateFailedFuture(new RuntimeException())).times(2);
    taskQueue.shutdown(EasyMock.contains("sequenceName-0"), EasyMock.eq("An exception occured while waiting for task [%s] to pause: [%s]"), EasyMock.contains("sequenceName-0"), EasyMock.anyString());
    EasyMock.expectLastCall().times(2);
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true).times(2);
    EasyMock.replay(taskStorage, taskRunner, taskClient, taskQueue);
    supervisor.runInternal();
    verifyAll();
    for (Task task : captured.getValues()) {
        KinesisIndexTaskIOConfig taskConfig = ((KinesisIndexTask) task).getIOConfig();
        Assert.assertEquals("0", taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
        Assert.assertNull(taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
    }
}

Example 19 with KinesisIndexTaskIOConfig

use of org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig in project druid by druid-io.

the class KinesisSupervisor method createIndexTasks.

@Override
protected List<SeekableStreamIndexTask<String, String, ByteEntity>> createIndexTasks(int replicas, String baseSequenceName, ObjectMapper sortingMapper, TreeMap<Integer, Map<String, String>> sequenceOffsets, SeekableStreamIndexTaskIOConfig taskIoConfig, SeekableStreamIndexTaskTuningConfig taskTuningConfig, RowIngestionMetersFactory rowIngestionMetersFactory) throws JsonProcessingException {
    final String checkpoints = sortingMapper.writerFor(CHECKPOINTS_TYPE_REF).writeValueAsString(sequenceOffsets);
    final Map<String, Object> context = createBaseTaskContexts();
    context.put(CHECKPOINTS_CTX_KEY, checkpoints);
    List<SeekableStreamIndexTask<String, String, ByteEntity>> taskList = new ArrayList<>();
    for (int i = 0; i < replicas; i++) {
        String taskId = IdUtils.getRandomIdWithPrefix(baseSequenceName);
        taskList.add(new KinesisIndexTask(taskId, new TaskResource(baseSequenceName, 1), spec.getDataSchema(), (KinesisIndexTaskTuningConfig) taskTuningConfig, (KinesisIndexTaskIOConfig) taskIoConfig, context, awsCredentialsConfig));
    }
    return taskList;
}

Example 20 with KinesisIndexTaskIOConfig

use of org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig in project druid by apache.

the class KinesisSupervisor method createIndexTasks.

@Override
protected List<SeekableStreamIndexTask<String, String, ByteEntity>> createIndexTasks(int replicas, String baseSequenceName, ObjectMapper sortingMapper, TreeMap<Integer, Map<String, String>> sequenceOffsets, SeekableStreamIndexTaskIOConfig taskIoConfig, SeekableStreamIndexTaskTuningConfig taskTuningConfig, RowIngestionMetersFactory rowIngestionMetersFactory) throws JsonProcessingException {
    final String checkpoints = sortingMapper.writerFor(CHECKPOINTS_TYPE_REF).writeValueAsString(sequenceOffsets);
    final Map<String, Object> context = createBaseTaskContexts();
    context.put(CHECKPOINTS_CTX_KEY, checkpoints);
    List<SeekableStreamIndexTask<String, String, ByteEntity>> taskList = new ArrayList<>();
    for (int i = 0; i < replicas; i++) {
        String taskId = IdUtils.getRandomIdWithPrefix(baseSequenceName);
        taskList.add(new KinesisIndexTask(taskId, new TaskResource(baseSequenceName, 1), spec.getDataSchema(), (KinesisIndexTaskTuningConfig) taskTuningConfig, (KinesisIndexTaskIOConfig) taskIoConfig, context, awsCredentialsConfig));
    }
    return taskList;
}