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

Example 1 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testKillUnresponsiveTasksWhileSettingEndOffsets.

@Test
public void testKillUnresponsiveTasksWhileSettingEndOffsets() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234, -1);
    supervisor = getTestableSupervisor(2, 2, true, "PT1M", null, null);
    addSomeEvents(100);
    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 KafkaDataSourceMetadata(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<Integer, Long>> checkpoints1 = new TreeMap<>();
    checkpoints1.put(0, ImmutableMap.of(0, 0L, 2, 0L));
    TreeMap<Integer, Map<Integer, Long>> checkpoints2 = new TreeMap<>();
    checkpoints2.put(0, ImmutableMap.of(1, 0L));
    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(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.immediateFuture(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L))).andReturn(Futures.immediateFuture(ImmutableMap.of(0, 10L, 1, 15L, 2, 35L)));
    EasyMock.expect(taskClient.setEndOffsetsAsync(EasyMock.contains("sequenceName-0"), EasyMock.eq(ImmutableMap.of(0, 10L, 1, 20L, 2, 35L)), EasyMock.eq(true))).andReturn(Futures.immediateFailedFuture(new RuntimeException())).times(2);
    taskQueue.shutdown(EasyMock.contains("sequenceName-0"), EasyMock.eq("Task [%s] failed to respond to [set end offsets] in a timely manner, killing task"), EasyMock.contains("sequenceName-0"));
    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()) {
        KafkaIndexTaskIOConfig taskConfig = ((KafkaIndexTask) task).getIOConfig();
        Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(0));
        Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(2));
    }
}

Example 2 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testLatestOffset.

/**
 * Test generating the starting offsets from the partition high water marks in Kafka.
 */
@Test
public void testLatestOffset() throws Exception {
    supervisor = getTestableSupervisor(1, 1, false, "PT1H", null, null);
    addSomeEvents(1100);
    Capture<KafkaIndexTask> captured = Capture.newInstance();
    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 KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    KafkaIndexTask task = captured.getValue();
    Assert.assertEquals(1101L, task.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().get(0).longValue());
    Assert.assertEquals(1101L, task.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().get(1).longValue());
    Assert.assertEquals(1101L, task.getIOConfig().getStartSequenceNumbers().getPartitionSequenceNumberMap().get(2).longValue());
}

Example 3 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testRequeueTaskWhenFailed.

@Test
public void testRequeueTaskWhenFailed() throws Exception {
    supervisor = getTestableSupervisor(2, 2, true, "PT1H", null, null);
    addSomeEvents(1);
    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(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(Status.NOT_STARTED)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true).times(4);
    TreeMap<Integer, Map<Integer, Long>> checkpoints1 = new TreeMap<>();
    checkpoints1.put(0, ImmutableMap.of(0, 0L, 2, 0L));
    TreeMap<Integer, Map<Integer, Long>> checkpoints2 = new TreeMap<>();
    checkpoints2.put(0, ImmutableMap.of(1, 0L));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("sequenceName-0"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints1)).anyTimes();
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("sequenceName-1"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints2)).anyTimes();
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    List<Task> tasks = captured.getValues();
    // test that running the main loop again checks the status of the tasks that were created and does nothing if they
    // are all still running
    EasyMock.reset(taskStorage);
    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.replay(taskStorage);
    supervisor.runInternal();
    verifyAll();
    // test that a task failing causes a new task to be re-queued with the same parameters
    Capture<Task> aNewTaskCapture = Capture.newInstance();
    List<Task> imStillAlive = tasks.subList(0, 3);
    KafkaIndexTask iHaveFailed = (KafkaIndexTask) tasks.get(3);
    EasyMock.reset(taskStorage);
    EasyMock.reset(taskQueue);
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(imStillAlive).anyTimes();
    for (Task task : imStillAlive) {
        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(taskStorage.getStatus(iHaveFailed.getId())).andReturn(Optional.of(TaskStatus.failure(iHaveFailed.getId(), "Dummy task status failure err message")));
    EasyMock.expect(taskStorage.getTask(iHaveFailed.getId())).andReturn(Optional.of(iHaveFailed)).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(aNewTaskCapture))).andReturn(true);
    EasyMock.replay(taskStorage);
    EasyMock.replay(taskQueue);
    supervisor.runInternal();
    verifyAll();
    Assert.assertNotEquals(iHaveFailed.getId(), aNewTaskCapture.getValue().getId());
    Assert.assertEquals(iHaveFailed.getIOConfig().getBaseSequenceName(), ((KafkaIndexTask) aNewTaskCapture.getValue()).getIOConfig().getBaseSequenceName());
}

Example 4 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testResetDataSourceMetadata.

@Test
public void testResetDataSourceMetadata() throws Exception {
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null);
    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();
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    Capture<String> captureDataSource = EasyMock.newCapture();
    Capture<DataSourceMetadata> captureDataSourceMetadata = EasyMock.newCapture();
    KafkaDataSourceMetadata kafkaDataSourceMetadata = new KafkaDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 1000L, 1, 1000L, 2, 1000L), ImmutableSet.of()));
    KafkaDataSourceMetadata resetMetadata = new KafkaDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(1, 1000L, 2, 1000L), ImmutableSet.of()));
    KafkaDataSourceMetadata expectedMetadata = new KafkaDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(topic, ImmutableMap.of(0, 1000L), ImmutableSet.of()));
    EasyMock.reset(indexerMetadataStorageCoordinator);
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(kafkaDataSourceMetadata);
    EasyMock.expect(indexerMetadataStorageCoordinator.resetDataSourceMetadata(EasyMock.capture(captureDataSource), EasyMock.capture(captureDataSourceMetadata))).andReturn(true);
    EasyMock.replay(indexerMetadataStorageCoordinator);
    try {
        supervisor.resetInternal(resetMetadata);
    } catch (NullPointerException npe) {
        // Expected as there will be an attempt to EasyMock.reset partitionGroups offsets to NOT_SET
        // however there would be no entries in the map as we have not put nay data in kafka
        Assert.assertNull(npe.getCause());
    }
    verifyAll();
    Assert.assertEquals(DATASOURCE, captureDataSource.getValue());
    Assert.assertEquals(expectedMetadata, captureDataSourceMetadata.getValue());
}

Example 5 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testKillBadPartitionAssignment.

@Test
public void testKillBadPartitionAssignment() throws Exception {
    supervisor = getTestableSupervisor(1, 2, true, "PT1H", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    addSomeEvents(1);
    Task id1 = createKafkaIndexTask("id1", DATASOURCE, 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)), null, null, tuningConfig);
    Task id2 = createKafkaIndexTask("id2", DATASOURCE, 1, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(1, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(1, Long.MAX_VALUE)), null, null, tuningConfig);
    Task id3 = createKafkaIndexTask("id3", 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);
    Task id4 = createKafkaIndexTask("id4", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(0, 0L, 1, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE)), null, null, tuningConfig);
    Task id5 = createKafkaIndexTask("id5", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(1, 0L, 2, 0L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    List<Task> existingTasks = ImmutableList.of(id1, id2, id3, id4, id5);
    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(existingTasks).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.getStatus("id4")).andReturn(Optional.of(TaskStatus.running("id4"))).anyTimes();
    EasyMock.expect(taskStorage.getStatus("id5")).andReturn(Optional.of(TaskStatus.running("id5"))).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(taskStorage.getTask("id4")).andReturn(Optional.of(id4)).anyTimes();
    EasyMock.expect(taskStorage.getTask("id5")).andReturn(Optional.of(id5)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(Status.NOT_STARTED)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.stopAsync("id3", false)).andReturn(Futures.immediateFuture(true));
    EasyMock.expect(taskClient.stopAsync("id4", false)).andReturn(Futures.immediateFuture(false));
    EasyMock.expect(taskClient.stopAsync("id5", false)).andReturn(Futures.immediateFuture(null));
    TreeMap<Integer, Map<Integer, Long>> checkpoints1 = new TreeMap<>();
    checkpoints1.put(0, ImmutableMap.of(0, 0L, 2, 0L));
    TreeMap<Integer, Map<Integer, Long>> checkpoints2 = new TreeMap<>();
    checkpoints2.put(0, ImmutableMap.of(1, 0L));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id1"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints1)).times(1);
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id2"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints2)).times(1);
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    taskQueue.shutdown("id4", "Task [%s] failed to stop in a timely manner, killing task", "id4");
    taskQueue.shutdown("id5", "Task [%s] failed to stop in a timely manner, killing task", "id5");
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
}