Search in sources :

Example 1 with TaskReportData

use of org.apache.druid.indexing.seekablestream.supervisor.TaskReportData 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 2 with TaskReportData

use of org.apache.druid.indexing.seekablestream.supervisor.TaskReportData 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 3 with TaskReportData

use of org.apache.druid.indexing.seekablestream.supervisor.TaskReportData in project druid by druid-io.

the class KinesisSupervisorTest 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();
    final Map<String, Long> timeLag = ImmutableMap.of(SHARD_ID0, 100L, SHARD_ID1, 200L);
    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();
    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(supervisorRecordSupplier.getPartitionsTimeLag(EasyMock.anyString(), EasyMock.anyObject())).andReturn(timeLag).atLeastOnce();
    Task 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);
    Task id2 = createKinesisIndexTask("id2", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"), 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);
    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 KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.PUBLISHING));
    EasyMock.expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    EasyMock.expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1")));
    EasyMock.expect(taskClient.getEndOffsets("id1")).andReturn(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id2", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(SHARD_ID1, "12", SHARD_ID0, "1")));
    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<String, String>> checkpoints = new TreeMap<>();
    checkpoints.put(0, ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id2"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints)).times(1);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    supervisor.updateCurrentAndLatestOffsets();
    SupervisorReport<KinesisSupervisorReportPayload> report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    KinesisSupervisorReportPayload payload = report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, payload.getDurationSeconds());
    Assert.assertEquals(2, payload.getPartitions());
    Assert.assertEquals(1, payload.getReplicas());
    Assert.assertEquals(STREAM, 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(SHARD_ID1, "2", SHARD_ID0, "1"), activeReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(SHARD_ID1, "12", SHARD_ID0, "1"), activeReport.getCurrentOffsets());
    Assert.assertEquals(timeLag, activeReport.getLagMillis());
    Assert.assertEquals("id1", publishingReport.getId());
    Assert.assertEquals(ImmutableMap.of(SHARD_ID1, "0", SHARD_ID0, "0"), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"), publishingReport.getCurrentOffsets());
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) KinesisIndexTask(org.apache.druid.indexing.kinesis.KinesisIndexTask) 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) KinesisDataSourceMetadata(org.apache.druid.indexing.kinesis.KinesisDataSourceMetadata) TaskReportData(org.apache.druid.indexing.seekablestream.supervisor.TaskReportData) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap) TreeMap(java.util.TreeMap) Test(org.junit.Test)

Example 4 with TaskReportData

use of org.apache.druid.indexing.seekablestream.supervisor.TaskReportData in project druid by druid-io.

the class KinesisSupervisorTest method testDiscoverExistingPublishingTaskWithDifferentPartitionAllocation.

@Test
public void testDiscoverExistingPublishingTaskWithDifferentPartitionAllocation() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234, -1);
    final Map<String, Long> timeLag = ImmutableMap.of(SHARD_ID1, 9000L, SHARD_ID0, 1234L);
    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();
    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(supervisorRecordSupplier.getPartitionsTimeLag(EasyMock.anyString(), EasyMock.anyObject())).andReturn(timeLag).atLeastOnce();
    Task task = 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);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(task, null, location));
    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(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 KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.PUBLISHING));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1")));
    EasyMock.expect(taskClient.getEndOffsets("id1")).andReturn(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"));
    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<KinesisSupervisorReportPayload> report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    KinesisSupervisorReportPayload payload = report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, payload.getDurationSeconds());
    Assert.assertEquals(2, payload.getPartitions());
    Assert.assertEquals(1, payload.getReplicas());
    Assert.assertEquals(STREAM, 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(SHARD_ID1, "0", SHARD_ID0, "0"), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"), publishingReport.getCurrentOffsets());
    KinesisIndexTask capturedTask = captured.getValue();
    Assert.assertEquals(dataSchema, capturedTask.getDataSchema());
    Assert.assertEquals(tuningConfig.convertToTaskTuningConfig(), capturedTask.getTuningConfig());
    KinesisIndexTaskIOConfig capturedTaskConfig = capturedTask.getIOConfig();
    Assert.assertEquals("awsEndpoint", capturedTaskConfig.getEndpoint());
    Assert.assertEquals("sequenceName-0", capturedTaskConfig.getBaseSequenceName());
    Assert.assertTrue("isUseTransaction", capturedTaskConfig.isUseTransaction());
    // check that the new task was created with starting sequences matching where the publishing task finished
    Assert.assertEquals(STREAM, capturedTaskConfig.getStartSequenceNumbers().getStream());
    Assert.assertEquals("2", capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals("1", capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
    Assert.assertEquals(STREAM, capturedTaskConfig.getEndSequenceNumbers().getStream());
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) KinesisIndexTask(org.apache.druid.indexing.kinesis.KinesisIndexTask) ArrayList(java.util.ArrayList) TaskLocation(org.apache.druid.indexer.TaskLocation) KinesisIndexTask(org.apache.druid.indexing.kinesis.KinesisIndexTask) Executor(java.util.concurrent.Executor) Collection(java.util.Collection) KinesisDataSourceMetadata(org.apache.druid.indexing.kinesis.KinesisDataSourceMetadata) TaskReportData(org.apache.druid.indexing.seekablestream.supervisor.TaskReportData) KinesisIndexTaskIOConfig(org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig) Test(org.junit.Test)

Example 5 with TaskReportData

use of org.apache.druid.indexing.seekablestream.supervisor.TaskReportData in project druid by druid-io.

the class KinesisSupervisorTest method testDiscoverExistingPublishingTask.

@Test
public void testDiscoverExistingPublishingTask() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234, -1);
    final Map<String, Long> timeLag = ImmutableMap.of(SHARD_ID1, 0L, SHARD_ID0, 20000000L);
    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();
    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(supervisorRecordSupplier.getPartitionsTimeLag(EasyMock.anyString(), EasyMock.anyObject())).andReturn(timeLag).atLeastOnce();
    Task task = 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);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(task, null, location));
    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(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 KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.PUBLISHING));
    EasyMock.expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1")));
    EasyMock.expect(taskClient.getEndOffsets("id1")).andReturn(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"));
    EasyMock.expect(taskQueue.add(EasyMock.capture(captured))).andReturn(true);
    TreeMap<Integer, Map<String, String>> checkpoints = new TreeMap<>();
    checkpoints.put(0, ImmutableMap.of(SHARD_ID1, "0", SHARD_ID0, "0"));
    EasyMock.expect(taskClient.getCheckpoints(EasyMock.anyString(), EasyMock.anyBoolean())).andReturn(checkpoints).anyTimes();
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    supervisor.updateCurrentAndLatestOffsets();
    SupervisorReport<KinesisSupervisorReportPayload> report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    KinesisSupervisorReportPayload payload = report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, payload.getDurationSeconds());
    Assert.assertEquals(2, payload.getPartitions());
    Assert.assertEquals(1, payload.getReplicas());
    Assert.assertEquals(STREAM, 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(SHARD_ID1, "0", SHARD_ID0, "0"), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(SHARD_ID1, "2", SHARD_ID0, "1"), publishingReport.getCurrentOffsets());
    KinesisIndexTask capturedTask = captured.getValue();
    Assert.assertEquals(dataSchema, capturedTask.getDataSchema());
    Assert.assertEquals(tuningConfig.convertToTaskTuningConfig(), capturedTask.getTuningConfig());
    KinesisIndexTaskIOConfig capturedTaskConfig = capturedTask.getIOConfig();
    Assert.assertEquals("awsEndpoint", capturedTaskConfig.getEndpoint());
    Assert.assertEquals("sequenceName-0", capturedTaskConfig.getBaseSequenceName());
    Assert.assertTrue("isUseTransaction", capturedTaskConfig.isUseTransaction());
    // check that the new task was created with starting sequences matching where the publishing task finished
    Assert.assertEquals(STREAM, capturedTaskConfig.getStartSequenceNumbers().getStream());
    Assert.assertEquals("2", capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals("1", capturedTaskConfig.getStartSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
    Assert.assertEquals(STREAM, capturedTaskConfig.getEndSequenceNumbers().getStream());
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID1));
    Assert.assertEquals(KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER, capturedTaskConfig.getEndSequenceNumbers().getPartitionSequenceNumberMap().get(SHARD_ID0));
}
Also used : TaskRunnerListener(org.apache.druid.indexing.overlord.TaskRunnerListener) Task(org.apache.druid.indexing.common.task.Task) RealtimeIndexTask(org.apache.druid.indexing.common.task.RealtimeIndexTask) KinesisIndexTask(org.apache.druid.indexing.kinesis.KinesisIndexTask) ArrayList(java.util.ArrayList) TreeMap(java.util.TreeMap) TaskLocation(org.apache.druid.indexer.TaskLocation) KinesisIndexTask(org.apache.druid.indexing.kinesis.KinesisIndexTask) Executor(java.util.concurrent.Executor) Collection(java.util.Collection) KinesisDataSourceMetadata(org.apache.druid.indexing.kinesis.KinesisDataSourceMetadata) TaskReportData(org.apache.druid.indexing.seekablestream.supervisor.TaskReportData) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap) TreeMap(java.util.TreeMap) KinesisIndexTaskIOConfig(org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig) Test(org.junit.Test)

Aggregations

ArrayList (java.util.ArrayList)6 Collection (java.util.Collection)6 Executor (java.util.concurrent.Executor)6 TaskLocation (org.apache.druid.indexer.TaskLocation)6 RealtimeIndexTask (org.apache.druid.indexing.common.task.RealtimeIndexTask)6 Task (org.apache.druid.indexing.common.task.Task)6 TaskRunnerListener (org.apache.druid.indexing.overlord.TaskRunnerListener)6 TaskReportData (org.apache.druid.indexing.seekablestream.supervisor.TaskReportData)6 Test (org.junit.Test)6 ImmutableMap (com.google.common.collect.ImmutableMap)4 HashMap (java.util.HashMap)4 Map (java.util.Map)4 TreeMap (java.util.TreeMap)4 KafkaDataSourceMetadata (org.apache.druid.indexing.kafka.KafkaDataSourceMetadata)3 KafkaIndexTask (org.apache.druid.indexing.kafka.KafkaIndexTask)3 KinesisDataSourceMetadata (org.apache.druid.indexing.kinesis.KinesisDataSourceMetadata)3 KinesisIndexTask (org.apache.druid.indexing.kinesis.KinesisIndexTask)3 ConcurrentHashMap (java.util.concurrent.ConcurrentHashMap)2 KafkaIndexTaskIOConfig (org.apache.druid.indexing.kafka.KafkaIndexTaskIOConfig)2 KinesisIndexTaskIOConfig (org.apache.druid.indexing.kinesis.KinesisIndexTaskIOConfig)2