Examples with KafkaIndexTask org.apache.druid.indexing.kafka.KafkaIndexTask used on opensource projects

Example 16 with KafkaIndexTask

use of org.apache.druid.indexing.kafka.KafkaIndexTask in project druid by druid-io.

the class KafkaSupervisorTest method testFailedInitializationAndRecovery.

@Test
public void testFailedInitializationAndRecovery() throws Exception {
    // Block the supervisor initialization with a bad hostname config, make sure this doesn't block the lifecycle
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null, false, StringUtils.format("badhostname:%d", kafkaServer.getPort()));
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    addSomeEvents(1);
    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 KafkaDataSourceMetadata(null)).anyTimes();
    replayAll();
    supervisor.start();
    Assert.assertTrue(supervisor.isLifecycleStarted());
    Assert.assertFalse(supervisor.isStarted());
    verifyAll();
    while (supervisor.getInitRetryCounter() < 3) {
        Thread.sleep(1000);
    }
    // Portion below is the same test as testNoInitialState(), testing the supervisor after the initialiation is fixed
    resetAll();
    Capture<KafkaIndexTask> 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 KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true);
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    // Fix the bad hostname during the initialization retries and finish the supervisor start.
    // This is equivalent to supervisor.start() in testNoInitialState().
    // The test supervisor has a P1D period, so we need to manually trigger the initialization retry.
    supervisor.getIoConfig().getConsumerProperties().put("bootstrap.servers", kafkaHost);
    supervisor.tryInit();
    Assert.assertTrue(supervisor.isLifecycleStarted());
    Assert.assertTrue(supervisor.isStarted());
    supervisor.runInternal();
    verifyAll();
    KafkaIndexTask task = captured.getValue();
    Assert.assertEquals(dataSchema, task.getDataSchema());
    Assert.assertEquals(tuningConfig.convertToTaskTuningConfig(), task.getTuningConfig());
    KafkaIndexTaskIOConfig taskConfig = task.getIOConfig();
    Assert.assertEquals(kafkaHost, taskConfig.getConsumerProperties().get("bootstrap.servers"));
    Assert.assertEquals("myCustomValue", taskConfig.getConsumerProperties().get("myCustomKey"));
    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(topic, taskConfig.getStartSequenceNumbers().getStream());
    Assert.assertEquals(0L, taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(0).longValue());
    Assert.assertEquals(0L, taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(1).longValue());
    Assert.assertEquals(0L, taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(2).longValue());
    Assert.assertEquals(topic, taskConfig.getEndSequenceNumbers().getStream());
    Assert.assertEquals(Long.MAX_VALUE, taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(0).longValue());
    Assert.assertEquals(Long.MAX_VALUE, taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(1).longValue());
    Assert.assertEquals(Long.MAX_VALUE, taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(2).longValue());
}

Example 17 with KafkaIndexTask

use of org.apache.druid.indexing.kafka.KafkaIndexTask in project druid by druid-io.

the class KafkaSupervisorTest method testNoInitialState.

@Test
public void testNoInitialState() throws Exception {
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    addSomeEvents(1);
    Capture<KafkaIndexTask> 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 KafkaDataSourceMetadata(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();
    KafkaIndexTask task = captured.getValue();
    Assert.assertEquals(dataSchema, task.getDataSchema());
    Assert.assertEquals(tuningConfig.convertToTaskTuningConfig(), task.getTuningConfig());
    KafkaIndexTaskIOConfig taskConfig = task.getIOConfig();
    Assert.assertEquals(kafkaHost, taskConfig.getConsumerProperties().get("bootstrap.servers"));
    Assert.assertEquals("myCustomValue", taskConfig.getConsumerProperties().get("myCustomKey"));
    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(topic, taskConfig.getStartSequenceNumbers().getStream());
    Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(0));
    Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(1));
    Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(2));
    Assert.assertEquals(topic, taskConfig.getEndSequenceNumbers().getStream());
    Assert.assertEquals(Long.MAX_VALUE, (long) taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(0));
    Assert.assertEquals(Long.MAX_VALUE, (long) taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(1));
    Assert.assertEquals(Long.MAX_VALUE, (long) taskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(2));
}

Example 18 with KafkaIndexTask

use of org.apache.druid.indexing.kafka.KafkaIndexTask in project druid by druid-io.

the class KafkaSupervisorTest method testCheckpointForInactiveTaskGroup.

@Test(timeout = 60_000L)
public void testCheckpointForInactiveTaskGroup() throws InterruptedException {
    supervisor = getTestableSupervisor(2, 1, true, "PT1S", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    supervisor.getStateManager().markRunFinished();
    // not adding any events
    final KafkaIndexTask id1 = createKafkaIndexTask("id1", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 0L, 1, 0L, 2, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    final Task id2 = createKafkaIndexTask("id2", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 10L, 1, 20L, 2, 30L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    final Task id3 = createKafkaIndexTask("id3", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 10L, 1, 20L, 2, 30L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    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));
    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 KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id3")).andReturn(Futures.immediateFuture(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<Integer, Long>> checkpoints = new TreeMap<>();
    checkpoints.put(0, ImmutableMap.of(0, 10L, 1, 20L, 2, 30L));
    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 KafkaDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(topic, checkpoints.get(0), ImmutableSet.of())));
    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 19 with KafkaIndexTask

use of org.apache.druid.indexing.kafka.KafkaIndexTask in project druid by druid-io.

the class KafkaSupervisorTest method testIsTaskCurrent.

@Test
public void testIsTaskCurrent() {
    DateTime minMessageTime = DateTimes.nowUtc();
    DateTime maxMessageTime = DateTimes.nowUtc().plus(10000);
    KafkaSupervisor supervisor = getSupervisor(2, 1, true, "PT1H", new Period("P1D"), new Period("P1D"), false, kafkaHost, dataSchema, new KafkaSupervisorTuningConfig(null, 1000, null, null, 50000, null, new Period("P1Y"), new File("/test"), null, null, null, false, null, false, null, numThreads, TEST_CHAT_THREADS, TEST_CHAT_RETRIES, TEST_HTTP_TIMEOUT, TEST_SHUTDOWN_TIMEOUT, null, null, null, null, null));
    supervisor.addTaskGroupToActivelyReadingTaskGroup(42, ImmutableMap.of(0, 0L, 2, 0L), Optional.of(minMessageTime), Optional.of(maxMessageTime), ImmutableSet.of("id1", "id2", "id3", "id4"), ImmutableSet.of());
    DataSchema modifiedDataSchema = getDataSchema("some other datasource");
    KafkaSupervisorTuningConfig modifiedTuningConfig = new KafkaSupervisorTuningConfig(null, // This is different
    42, null, null, 50000, null, new Period("P1Y"), new File("/test"), null, null, null, false, null, null, null, numThreads, TEST_CHAT_THREADS, TEST_CHAT_RETRIES, TEST_HTTP_TIMEOUT, TEST_SHUTDOWN_TIMEOUT, null, null, null, null, null);
    KafkaIndexTask taskFromStorage = createKafkaIndexTask("id1", 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(0, 0L, 2, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(0, Long.MAX_VALUE, 2, Long.MAX_VALUE)), minMessageTime, maxMessageTime, dataSchema, supervisor.getTuningConfig());
    KafkaIndexTask taskFromStorageMismatchedDataSchema = createKafkaIndexTask("id2", 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(0, 0L, 2, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(0, Long.MAX_VALUE, 2, Long.MAX_VALUE)), minMessageTime, maxMessageTime, modifiedDataSchema, supervisor.getTuningConfig());
    KafkaIndexTask taskFromStorageMismatchedTuningConfig = createKafkaIndexTask("id3", 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(0, 0L, 2, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(0, Long.MAX_VALUE, 2, Long.MAX_VALUE)), minMessageTime, maxMessageTime, dataSchema, modifiedTuningConfig);
    KafkaIndexTask taskFromStorageMismatchedPartitionsWithTaskGroup = createKafkaIndexTask("id4", 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(0, 0L, 2, 6L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(0, Long.MAX_VALUE, 2, Long.MAX_VALUE)), minMessageTime, maxMessageTime, dataSchema, supervisor.getTuningConfig());
    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();
}

Example 20 with KafkaIndexTask

use of org.apache.druid.indexing.kafka.KafkaIndexTask in project druid by druid-io.

the class KafkaSupervisorTest method testCheckpointForUnknownTaskGroup.

@Test(timeout = 60_000L)
public void testCheckpointForUnknownTaskGroup() throws InterruptedException {
    supervisor = getTestableSupervisor(2, 1, true, "PT1S", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    // not adding any events
    final KafkaIndexTask id1 = createKafkaIndexTask("id1", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 0L, 1, 0L, 2, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    final Task id2 = createKafkaIndexTask("id2", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 10L, 1, 20L, 2, 30L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    final Task id3 = createKafkaIndexTask("id3", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 10L, 1, 20L, 2, 30L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    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 KafkaDataSourceMetadata(null)).anyTimes();
    replayAll();
    supervisor.start();
    supervisor.checkpoint(0, new KafkaDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(topic, Collections.emptyMap(), ImmutableSet.of())));
    while (supervisor.getNoticesQueueSize() > 0) {
        Thread.sleep(100);
    }
    verifyAll();
    while (serviceEmitter.getStackTrace() == null) {
        Thread.sleep(100);
    }
    Assert.assertTrue(serviceEmitter.getStackTrace().startsWith("org.apache.druid.java.util.common.ISE: Cannot find"));
    Assert.assertEquals("Cannot find taskGroup [0] among all activelyReadingTaskGroups [{}]", serviceEmitter.getExceptionMessage());
    Assert.assertEquals(ISE.class, serviceEmitter.getExceptionClass());
}