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

Example 1 with KinesisIndexTask

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

the class KinesisSupervisorTest method testNoInitialStateWithAutoScaleIn.

@Test
public void testNoInitialStateWithAutoScaleIn() throws Exception {
    HashMap<String, Object> autoScalerConfigMap = new HashMap<>();
    autoScalerConfigMap.put("enableTaskAutoScaler", true);
    autoScalerConfigMap.put("lagCollectionIntervalMillis", 500);
    autoScalerConfigMap.put("lagCollectionRangeMillis", 500);
    autoScalerConfigMap.put("scaleOutThreshold", 1000000);
    autoScalerConfigMap.put("triggerScaleOutFractionThreshold", 0.8);
    autoScalerConfigMap.put("scaleInThreshold", 0);
    autoScalerConfigMap.put("triggerScaleInFractionThreshold", 0.0);
    autoScalerConfigMap.put("scaleActionStartDelayMillis", 0);
    autoScalerConfigMap.put("scaleActionPeriodMillis", 100);
    autoScalerConfigMap.put("taskCountMax", 2);
    autoScalerConfigMap.put("taskCountMin", 1);
    autoScalerConfigMap.put("scaleInStep", 1);
    autoScalerConfigMap.put("scaleOutStep", 2);
    autoScalerConfigMap.put("minTriggerScaleActionFrequencyMillis", 1200000);
    AutoScalerConfig autoScalerConfig = OBJECT_MAPPER.convertValue(autoScalerConfigMap, AutoScalerConfig.class);
    supervisor = getTestableSupervisor(1, 2, true, "PT1H", null, null, false, null, null, autoScalerConfig);
    KinesisSupervisorSpec kinesisSupervisorSpec = supervisor.getKinesisSupervisorSpec();
    SupervisorTaskAutoScaler autoscaler = kinesisSupervisorSpec.createAutoscaler(supervisor);
    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(CaptureType.ALL);
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.absent()).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(2);
    replayAll();
    int taskCountInit = supervisor.getIoConfig().getTaskCount();
    // when enable autoScaler the init taskCount will be equal to taskCountMin
    Assert.assertEquals(1, taskCountInit);
    supervisor.getIoConfig().setTaskCount(2);
    supervisor.start();
    int taskCountBeforeScale = supervisor.getIoConfig().getTaskCount();
    Assert.assertEquals(2, taskCountBeforeScale);
    autoscaler.start();
    supervisor.runInternal();
    verifyAll();
    Thread.sleep(1 * 1000);
    int taskCountAfterScale = supervisor.getIoConfig().getTaskCount();
    Assert.assertEquals(1, taskCountAfterScale);
}

Example 2 with KinesisIndexTask

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

the class KinesisSupervisorTest method testMultiTask.

@Test
public void testMultiTask() throws Exception {
    supervisor = getTestableSupervisor(1, 2, 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(CaptureType.ALL);
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.absent()).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(2);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    KinesisIndexTask task1 = captured.getValues().get(0);
    Assert.assertEquals(1, task1.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().size());
    Assert.assertEquals(1, task1.getIOConfig().getEndSequenceNumbers().getPartitionSequenceNumberMap().size());
    Assert.assertEquals("0", task1.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, task1.getIOConfig().getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    KinesisIndexTask task2 = captured.getValues().get(1);
    Assert.assertEquals(1, task2.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().size());
    Assert.assertEquals(1, task2.getIOConfig().getEndSequenceNumbers().getPartitionSequenceNumberMap().size());
    Assert.assertEquals("0", task2.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, task2.getIOConfig().getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
}

Example 3 with KinesisIndexTask

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

the class KinesisSupervisorTest method testLateMessageRejectionPeriod.

@Test
public void testLateMessageRejectionPeriod() throws Exception {
    supervisor = getTestableSupervisor(2, 1, true, "PT1H", new Period("PT1H"), 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(CaptureType.ALL);
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.absent()).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(2);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    KinesisIndexTask task1 = captured.getValues().get(0);
    KinesisIndexTask task2 = captured.getValues().get(1);
    Assert.assertTrue("minimumMessageTime", task1.getIOConfig().getMinimumMessageTime().get().plusMinutes(59).isBeforeNow());
    Assert.assertTrue("minimumMessageTime", task1.getIOConfig().getMinimumMessageTime().get().plusMinutes(61).isAfterNow());
    Assert.assertEquals(task1.getIOConfig().getMinimumMessageTime().get(), task2.getIOConfig().getMinimumMessageTime().get());
}

Example 4 with KinesisIndexTask

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

the class KinesisSupervisorTest method testCheckpointForInactiveTaskGroup.

@Test(timeout = 60_000L)
public void testCheckpointForInactiveTaskGroup() throws InterruptedException {
    supervisor = getTestableSupervisor(2, 1, true, "PT1S", null, null, false);
    // not adding any events
    final KinesisIndexTask id1;
    id1 = createKinesisIndexTask("id1", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "0", SHARD_ID0, "0"), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, SHARD_ID0, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), null, null);
    final Task id2 = createKinesisIndexTask("id2", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "10", SHARD_ID0, "20"), ImmutableSet.of(SHARD_ID0, SHARD_ID1)), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, SHARD_ID0, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), null, null);
    final Task id3 = createKinesisIndexTask("id3", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "10", SHARD_ID0, "20"), ImmutableSet.of(SHARD_ID0, SHARD_ID1)), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, SHARD_ID0, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), null, null);
    final TaskLocation location1 = new TaskLocation("testHost", 1234, -1);
    final TaskLocation location2 = new TaskLocation("testHost2", 145, -1);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(id1, null, location1));
    workItems.add(new TestTaskRunnerWorkItem(id2, null, location2));
    workItems.add(new TestTaskRunnerWorkItem(id2, null, location2));
    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();
    EasyMock.expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    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(id1, id2, id3)).anyTimes();
    EasyMock.expect(taskStorage.getStatus("id1")).andReturn(Optional.of(TaskStatus.running("id1"))).anyTimes();
    EasyMock.expect(taskStorage.getStatus("id2")).andReturn(Optional.of(TaskStatus.running("id2"))).anyTimes();
    EasyMock.expect(taskStorage.getStatus("id3")).andReturn(Optional.of(TaskStatus.running("id3"))).anyTimes();
    EasyMock.expect(taskStorage.getTask("id1")).andReturn(Optional.of(id1)).anyTimes();
    EasyMock.expect(taskStorage.getTask("id2")).andReturn(Optional.of(id2)).anyTimes();
    EasyMock.expect(taskStorage.getTask("id3")).andReturn(Optional.of(id3)).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id3")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    final DateTime startTime = DateTimes.nowUtc();
    EasyMock.expect(taskClient.getStartTimeAsync("id1")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getStartTimeAsync("id3")).andReturn(Futures.immediateFuture(startTime));
    final TreeMap<Integer, Map<String, String>> checkpoints = new TreeMap<>();
    checkpoints.put(0, ImmutableMap.of(SHARD_ID1, "10", SHARD_ID0, "20"));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id1"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints)).times(1);
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id2"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints)).times(1);
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id3"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints)).times(1);
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    supervisor.moveTaskGroupToPendingCompletion(0);
    supervisor.checkpoint(0, new KinesisDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(STREAM, checkpoints.get(0), checkpoints.get(0).keySet())));
    while (supervisor.getNoticesQueueSize() > 0) {
        Thread.sleep(100);
    }
    verifyAll();
    Assert.assertNull(serviceEmitter.getStackTrace(), serviceEmitter.getStackTrace());
    Assert.assertNull(serviceEmitter.getExceptionMessage(), serviceEmitter.getExceptionMessage());
    Assert.assertNull(serviceEmitter.getExceptionClass());
}

Example 5 with KinesisIndexTask

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

the class KinesisSupervisorTest method testIsTaskCurrent.

@Test
public void testIsTaskCurrent() {
    DateTime minMessageTime = DateTimes.nowUtc();
    DateTime maxMessageTime = DateTimes.nowUtc().plus(10000);
    KinesisSupervisor supervisor = getSupervisor(1, 1, true, "PT1H", new Period("P1D"), new Period("P1D"), false, 42, 42, dataSchema, tuningConfig);
    supervisor.addTaskGroupToActivelyReadingTaskGroup(42, ImmutableMap.of(SHARD_ID1, "3"), Optional.of(minMessageTime), Optional.of(maxMessageTime), ImmutableSet.of("id1", "id2", "id3", "id4"), ImmutableSet.of());
    DataSchema modifiedDataSchema = getDataSchema("some other datasource");
    KinesisSupervisorTuningConfig modifiedTuningConfig = new KinesisSupervisorTuningConfig(null, 1000, null, null, 50000, null, new Period("P1Y"), new File("/test"), null, null, null, false, null, null, null, null, numThreads, TEST_CHAT_THREADS, TEST_CHAT_RETRIES, TEST_HTTP_TIMEOUT, TEST_SHUTDOWN_TIMEOUT, null, null, null, 5000, null, null, null, null, // This property is different from tuningConfig
    42, null, null, null, null);
    KinesisIndexTask taskFromStorage = createKinesisIndexTask("id1", 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "3"), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), minMessageTime, maxMessageTime, dataSchema);
    KinesisIndexTask taskFromStorageMismatchedDataSchema = createKinesisIndexTask("id2", 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "3"), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), minMessageTime, maxMessageTime, modifiedDataSchema);
    KinesisIndexTask taskFromStorageMismatchedTuningConfig = createKinesisIndexTask("id3", 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "3"), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), minMessageTime, maxMessageTime, dataSchema, modifiedTuningConfig);
    KinesisIndexTask taskFromStorageMismatchedPartitionsWithTaskGroup = createKinesisIndexTask("id4", 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, // this is the mismatch
    "4"), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER)), minMessageTime, maxMessageTime, dataSchema);
    EasyMock.expect(taskStorage.getTask("id1")).andReturn(Optional.of(taskFromStorage)).once();
    EasyMock.expect(taskStorage.getTask("id2")).andReturn(Optional.of(taskFromStorageMismatchedDataSchema)).once();
    EasyMock.expect(taskStorage.getTask("id3")).andReturn(Optional.of(taskFromStorageMismatchedTuningConfig)).once();
    EasyMock.expect(taskStorage.getTask("id4")).andReturn(Optional.of(taskFromStorageMismatchedPartitionsWithTaskGroup)).once();
    replayAll();
    Assert.assertTrue(supervisor.isTaskCurrent(42, "id1"));
    Assert.assertFalse(supervisor.isTaskCurrent(42, "id2"));
    Assert.assertFalse(supervisor.isTaskCurrent(42, "id3"));
    Assert.assertFalse(supervisor.isTaskCurrent(42, "id4"));
    verifyAll();
}