Search in sources :

Example 16 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testResetRunningTasks.

@Test
public void testResetRunningTasks() throws Exception {
    final TaskLocation location1 = new TaskLocation("testHost", 1234, -1);
    final TaskLocation location2 = new TaskLocation("testHost2", 145, -1);
    final DateTime startTime = DateTimes.nowUtc();
    supervisor = getTestableSupervisor(2, 1, true, "PT1H", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    addSomeEvents(1);
    Task 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);
    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);
    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);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(id1, null, location1));
    workItems.add(new TestTaskRunnerWorkItem(id2, null, location2));
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    EasyMock.expect(taskRunner.getRunningTasks()).andReturn(workItems).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.PUBLISHING));
    EasyMock.expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id3")).andReturn(Futures.immediateFuture(Status.READING));
    EasyMock.expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getStartTimeAsync("id3")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getEndOffsets("id1")).andReturn(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L));
    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("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();
    verifyAll();
    EasyMock.reset(taskQueue, indexerMetadataStorageCoordinator);
    EasyMock.expect(indexerMetadataStorageCoordinator.deleteDataSourceMetadata(DATASOURCE)).andReturn(true);
    taskQueue.shutdown("id2", "DataSourceMetadata is not found while reset");
    taskQueue.shutdown("id3", "DataSourceMetadata is not found while reset");
    EasyMock.replay(taskQueue, indexerMetadataStorageCoordinator);
    supervisor.resetInternal(null);
    verifyAll();
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) ArrayList(java.util.ArrayList) TreeMap(java.util.TreeMap) TaskLocation(org.apache.druid.indexer.TaskLocation) DateTime(org.joda.time.DateTime) Executor(java.util.concurrent.Executor) Collection(java.util.Collection) KafkaDataSourceMetadata(org.apache.druid.indexing.kafka.KafkaDataSourceMetadata) Map(java.util.Map) TreeMap(java.util.TreeMap) ImmutableMap(com.google.common.collect.ImmutableMap) HashMap(java.util.HashMap) Test(org.junit.Test)

Example 17 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testKillUnresponsiveTasksWhileGettingStartTime.

@Test
public void testKillUnresponsiveTasksWhileGettingStartTime() 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(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();
    EasyMock.reset(taskStorage, 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);
    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(taskClient.getStatusAsync(task.getId())).andReturn(Futures.immediateFuture(Status.NOT_STARTED));
        EasyMock.expect(taskClient.getStartTimeAsync(task.getId())).andReturn(Futures.immediateFailedFuture(new RuntimeException()));
        taskQueue.shutdown(task.getId(), "Task [%s] failed to return start time, killing task", task.getId());
    }
    EasyMock.replay(taskStorage, taskClient, taskQueue);
    supervisor.runInternal();
    verifyAll();
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) Executor(java.util.concurrent.Executor) KafkaDataSourceMetadata(org.apache.druid.indexing.kafka.KafkaDataSourceMetadata) TreeMap(java.util.TreeMap) Map(java.util.Map) TreeMap(java.util.TreeMap) ImmutableMap(com.google.common.collect.ImmutableMap) HashMap(java.util.HashMap) Test(org.junit.Test)

Example 18 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testDiscoverExistingPublishingTaskWithDifferentPartitionAllocation.

@Test
public void testDiscoverExistingPublishingTaskWithDifferentPartitionAllocation() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234, -1);
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    addSomeEvents(1);
    Task task = 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, supervisor.getTuningConfig());
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(task, null, location));
    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(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(ImmutableList.of(task)).anyTimes();
    EasyMock.expect(taskStorage.getStatus("id1")).andReturn(Optional.of(TaskStatus.running("id1"))).anyTimes();
    EasyMock.expect(taskStorage.getTask("id1")).andReturn(Optional.of(task)).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(Status.PUBLISHING));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(0, 10L, 2, 30L)));
    EasyMock.expect(taskClient.getEndOffsets("id1")).andReturn(ImmutableMap.of(0, 10L, 2, 30L));
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true);
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    supervisor.updateCurrentAndLatestOffsets();
    SupervisorReport<KafkaSupervisorReportPayload> report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    KafkaSupervisorReportPayload payload = report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, payload.getDurationSeconds());
    Assert.assertEquals(NUM_PARTITIONS, payload.getPartitions());
    Assert.assertEquals(1, payload.getReplicas());
    Assert.assertEquals(topic, payload.getStream());
    Assert.assertEquals(0, payload.getActiveTasks().size());
    Assert.assertEquals(1, payload.getPublishingTasks().size());
    Assert.assertEquals(SupervisorStateManager.BasicState.RUNNING, payload.getDetailedState());
    Assert.assertEquals(0, payload.getRecentErrors().size());
    TaskReportData publishingReport = payload.getPublishingTasks().get(0);
    Assert.assertEquals("id1", publishingReport.getId());
    Assert.assertEquals(ImmutableMap.of(0, 0L, 2, 0L), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 10L, 2, 30L), publishingReport.getCurrentOffsets());
    KafkaIndexTask capturedTask = captured.getValue();
    Assert.assertEquals(dataSchema, capturedTask.getDataSchema());
    Assert.assertEquals(tuningConfig.convertToTaskTuningConfig(), capturedTask.getTuningConfig());
    KafkaIndexTaskIOConfig capturedTaskConfig = capturedTask.getIOConfig();
    Assert.assertEquals(kafkaHost, capturedTaskConfig.getConsumerProperties().get("bootstrap.servers"));
    Assert.assertEquals("myCustomValue", capturedTaskConfig.getConsumerProperties().get("myCustomKey"));
    Assert.assertEquals("sequenceName-0", capturedTaskConfig.getBaseSequenceName());
    Assert.assertTrue("isUseTransaction", capturedTaskConfig.isUseTransaction());
    // check that the new task was created with starting offsets matching where the publishing task finished
    Assert.assertEquals(topic, capturedTaskConfig.getStartSequenceNumbers().getStream());
    Assert.assertEquals(10L, capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(0).longValue());
    Assert.assertEquals(0L, capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(1).longValue());
    Assert.assertEquals(30L, capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(2).longValue());
    Assert.assertEquals(topic, capturedTaskConfig.getEndSequenceNumbers().getStream());
    Assert.assertEquals(Long.MAX_VALUE, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(0).longValue());
    Assert.assertEquals(Long.MAX_VALUE, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(1).longValue());
    Assert.assertEquals(Long.MAX_VALUE, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(2).longValue());
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) ArrayList(java.util.ArrayList) TaskLocation(org.apache.druid.indexer.TaskLocation) KafkaIndexTaskIOConfig(org.apache.druid.indexing.kafka.KafkaIndexTaskIOConfig) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Executor(java.util.concurrent.Executor) Collection(java.util.Collection) KafkaDataSourceMetadata(org.apache.druid.indexing.kafka.KafkaDataSourceMetadata) TaskReportData(org.apache.druid.indexing.seekablestream.supervisor.TaskReportData) Test(org.junit.Test)

Example 19 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testDiscoverExistingPublishingAndReadingTask.

@Test
public void testDiscoverExistingPublishingAndReadingTask() throws Exception {
    final TaskLocation location1 = new TaskLocation("testHost", 1234, -1);
    final TaskLocation location2 = new TaskLocation("testHost2", 145, -1);
    final DateTime startTime = DateTimes.nowUtc();
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null);
    final KafkaSupervisorTuningConfig tuningConfig = supervisor.getTuningConfig();
    addSomeEvents(6);
    Task 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);
    Task id2 = createKafkaIndexTask("id2", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("topic", ImmutableMap.of(0, 1L, 1, 2L, 2, 3L), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>("topic", ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null, null, tuningConfig);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(id1, null, location1));
    workItems.add(new TestTaskRunnerWorkItem(id2, null, location2));
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    EasyMock.expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(ImmutableList.of(id1, id2)).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.getTask("id1")).andReturn(Optional.of(id1)).anyTimes();
    EasyMock.expect(taskStorage.getTask("id2")).andReturn(Optional.of(id2)).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(Status.PUBLISHING));
    EasyMock.expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(Status.READING));
    EasyMock.expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(0, 1L, 1, 2L, 2, 3L)));
    EasyMock.expect(taskClient.getEndOffsets("id1")).andReturn(ImmutableMap.of(0, 1L, 1, 2L, 2, 3L));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id2", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(0, 4L, 1, 5L, 2, 6L)));
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    // since id1 is publishing, so getCheckpoints wouldn't be called for it
    TreeMap<Integer, Map<Integer, Long>> checkpoints = new TreeMap<>();
    checkpoints.put(0, ImmutableMap.of(0, 1L, 1, 2L, 2, 3L));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id2"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints)).times(1);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    supervisor.updateCurrentAndLatestOffsets();
    SupervisorReport<KafkaSupervisorReportPayload> report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    KafkaSupervisorReportPayload payload = report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, payload.getDurationSeconds());
    Assert.assertEquals(NUM_PARTITIONS, payload.getPartitions());
    Assert.assertEquals(1, payload.getReplicas());
    Assert.assertEquals(topic, payload.getStream());
    Assert.assertEquals(1, payload.getActiveTasks().size());
    Assert.assertEquals(1, payload.getPublishingTasks().size());
    Assert.assertEquals(SupervisorStateManager.BasicState.RUNNING, payload.getDetailedState());
    Assert.assertEquals(0, payload.getRecentErrors().size());
    TaskReportData activeReport = payload.getActiveTasks().get(0);
    TaskReportData publishingReport = payload.getPublishingTasks().get(0);
    Assert.assertEquals("id2", activeReport.getId());
    Assert.assertEquals(startTime, activeReport.getStartTime());
    Assert.assertEquals(ImmutableMap.of(0, 1L, 1, 2L, 2, 3L), activeReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 4L, 1, 5L, 2, 6L), activeReport.getCurrentOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 3L, 1, 2L, 2, 1L), activeReport.getLag());
    Assert.assertEquals("id1", publishingReport.getId());
    Assert.assertEquals(ImmutableMap.of(0, 0L, 1, 0L, 2, 0L), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 1L, 1, 2L, 2, 3L), publishingReport.getCurrentOffsets());
    Assert.assertNull(publishingReport.getLag());
    Assert.assertEquals(ImmutableMap.of(0, 7L, 1, 7L, 2, 7L), payload.getLatestOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 3L, 1, 2L, 2, 1L), payload.getMinimumLag());
    Assert.assertEquals(6L, (long) payload.getAggregateLag());
    Assert.assertTrue(payload.getOffsetsLastUpdated().plusMinutes(1).isAfterNow());
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) ArrayList(java.util.ArrayList) TreeMap(java.util.TreeMap) TaskLocation(org.apache.druid.indexer.TaskLocation) DateTime(org.joda.time.DateTime) Executor(java.util.concurrent.Executor) Collection(java.util.Collection) KafkaDataSourceMetadata(org.apache.druid.indexing.kafka.KafkaDataSourceMetadata) TaskReportData(org.apache.druid.indexing.seekablestream.supervisor.TaskReportData) Map(java.util.Map) TreeMap(java.util.TreeMap) ImmutableMap(com.google.common.collect.ImmutableMap) HashMap(java.util.HashMap) Test(org.junit.Test)

Example 20 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testKillUnresponsiveTasksWhilePausing.

@Test
public void testKillUnresponsiveTasksWhilePausing() 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.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()) {
        KafkaIndexTaskIOConfig taskConfig = ((KafkaIndexTask) task).getIOConfig();
        Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(0));
        Assert.assertEquals(0L, (long) taskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(2));
    }
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) ArrayList(java.util.ArrayList) TreeMap(java.util.TreeMap) TaskLocation(org.apache.druid.indexer.TaskLocation) KafkaIndexTaskIOConfig(org.apache.druid.indexing.kafka.KafkaIndexTaskIOConfig) Executor(java.util.concurrent.Executor) KafkaIndexTask(org.apache.druid.indexing.kafka.KafkaIndexTask) Collection(java.util.Collection) KafkaDataSourceMetadata(org.apache.druid.indexing.kafka.KafkaDataSourceMetadata) Map(java.util.Map) TreeMap(java.util.TreeMap) ImmutableMap(com.google.common.collect.ImmutableMap) HashMap(java.util.HashMap) Test(org.junit.Test)

Aggregations

KafkaDataSourceMetadata (org.apache.druid.indexing.kafka.KafkaDataSourceMetadata)33 Test (org.junit.Test)33 KafkaIndexTask (org.apache.druid.indexing.kafka.KafkaIndexTask)30 Executor (java.util.concurrent.Executor)25 TaskRunnerListener (org.apache.druid.indexing.overlord.TaskRunnerListener)25 RealtimeIndexTask (org.apache.druid.indexing.common.task.RealtimeIndexTask)20 Task (org.apache.druid.indexing.common.task.Task)20 HashMap (java.util.HashMap)17 ImmutableMap (com.google.common.collect.ImmutableMap)16 Map (java.util.Map)16 TreeMap (java.util.TreeMap)16 ArrayList (java.util.ArrayList)10 Collection (java.util.Collection)10 TaskLocation (org.apache.druid.indexer.TaskLocation)10 KafkaIndexTaskIOConfig (org.apache.druid.indexing.kafka.KafkaIndexTaskIOConfig)9 DateTime (org.joda.time.DateTime)7 Period (org.joda.time.Period)5 TaskReportData (org.apache.druid.indexing.seekablestream.supervisor.TaskReportData)3 SeekableStreamStartSequenceNumbers (org.apache.druid.indexing.seekablestream.SeekableStreamStartSequenceNumbers)2 File (java.io.File)1