Search in sources :

Example 1 with SupervisorReport

use of io.druid.indexing.overlord.supervisor.SupervisorReport in project druid by druid-io.

the class KafkaSupervisorTest method testDiscoverExistingPublishingTask.

@Test
public void testDiscoverExistingPublishingTask() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234);
    supervisor = getSupervisor(1, 1, true, "PT1H", null);
    addSomeEvents(1);
    Task task = createKafkaIndexTask("id1", DATASOURCE, "sequenceName-0", new KafkaPartitions("topic", ImmutableMap.of(0, 0L, 1, 0L, 2, 0L)), new KafkaPartitions("topic", ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(task.getId(), null, location));
    Capture<KafkaIndexTask> captured = Capture.newInstance();
    expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.of(task)).anyTimes();
    expect(taskStorage.getStatus("id1")).andReturn(Optional.of(TaskStatus.running("id1"))).anyTimes();
    expect(taskStorage.getTask("id1")).andReturn(Optional.of(task)).anyTimes();
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.PUBLISHING));
    expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 10L, 1, 20L, 2, 30L)));
    expect(taskClient.getCurrentOffsets("id1", true)).andReturn(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L));
    expect(taskQueue.add(capture(captured))).andReturn(true);
    taskRunner.registerListener(anyObject(TaskRunnerListener.class), anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    SupervisorReport report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    Assert.assertTrue(report.getPayload() instanceof KafkaSupervisorReport.KafkaSupervisorReportPayload);
    KafkaSupervisorReport.KafkaSupervisorReportPayload payload = (KafkaSupervisorReport.KafkaSupervisorReportPayload) report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, (long) payload.getDurationSeconds());
    Assert.assertEquals(NUM_PARTITIONS, (int) payload.getPartitions());
    Assert.assertEquals(1, (int) payload.getReplicas());
    Assert.assertEquals(KAFKA_TOPIC, payload.getTopic());
    Assert.assertEquals(0, payload.getActiveTasks().size());
    Assert.assertEquals(1, payload.getPublishingTasks().size());
    TaskReportData publishingReport = payload.getPublishingTasks().get(0);
    Assert.assertEquals("id1", publishingReport.getId());
    Assert.assertEquals(ImmutableMap.of(0, 0L, 1, 0L, 2, 0L), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L), publishingReport.getCurrentOffsets());
    KafkaIndexTask capturedTask = captured.getValue();
    Assert.assertEquals(dataSchema, capturedTask.getDataSchema());
    Assert.assertEquals(KafkaTuningConfig.copyOf(tuningConfig), capturedTask.getTuningConfig());
    KafkaIOConfig 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());
    Assert.assertFalse("pauseAfterRead", capturedTaskConfig.isPauseAfterRead());
    // check that the new task was created with starting offsets matching where the publishing task finished
    Assert.assertEquals(KAFKA_TOPIC, capturedTaskConfig.getStartPartitions().getTopic());
    Assert.assertEquals(10L, (long) capturedTaskConfig.getStartPartitions().getPartitionOffsetMap().get(0));
    Assert.assertEquals(20L, (long) capturedTaskConfig.getStartPartitions().getPartitionOffsetMap().get(1));
    Assert.assertEquals(30L, (long) capturedTaskConfig.getStartPartitions().getPartitionOffsetMap().get(2));
    Assert.assertEquals(KAFKA_TOPIC, capturedTaskConfig.getEndPartitions().getTopic());
    Assert.assertEquals(Long.MAX_VALUE, (long) capturedTaskConfig.getEndPartitions().getPartitionOffsetMap().get(0));
    Assert.assertEquals(Long.MAX_VALUE, (long) capturedTaskConfig.getEndPartitions().getPartitionOffsetMap().get(1));
    Assert.assertEquals(Long.MAX_VALUE, (long) capturedTaskConfig.getEndPartitions().getPartitionOffsetMap().get(2));
}
Also used : TaskRunnerListener(io.druid.indexing.overlord.TaskRunnerListener) RealtimeIndexTask(io.druid.indexing.common.task.RealtimeIndexTask) Task(io.druid.indexing.common.task.Task) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) KafkaPartitions(io.druid.indexing.kafka.KafkaPartitions) ArrayList(java.util.ArrayList) TaskLocation(io.druid.indexing.common.TaskLocation) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) Executor(java.util.concurrent.Executor) SupervisorReport(io.druid.indexing.overlord.supervisor.SupervisorReport) KafkaIOConfig(io.druid.indexing.kafka.KafkaIOConfig) Collection(java.util.Collection) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) Test(org.junit.Test)

Example 2 with SupervisorReport

use of io.druid.indexing.overlord.supervisor.SupervisorReport in project druid by druid-io.

the class KafkaSupervisorTest method testDiscoverExistingPublishingTaskWithDifferentPartitionAllocation.

@Test
public void testDiscoverExistingPublishingTaskWithDifferentPartitionAllocation() throws Exception {
    final TaskLocation location = new TaskLocation("testHost", 1234);
    supervisor = getSupervisor(1, 1, true, "PT1H", null);
    addSomeEvents(1);
    Task task = createKafkaIndexTask("id1", DATASOURCE, "sequenceName-0", new KafkaPartitions("topic", ImmutableMap.of(0, 0L, 2, 0L)), new KafkaPartitions("topic", ImmutableMap.of(0, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(task.getId(), null, location));
    Capture<KafkaIndexTask> captured = Capture.newInstance();
    expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.of(task)).anyTimes();
    expect(taskStorage.getStatus("id1")).andReturn(Optional.of(TaskStatus.running("id1"))).anyTimes();
    expect(taskStorage.getTask("id1")).andReturn(Optional.of(task)).anyTimes();
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.PUBLISHING));
    expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 10L, 2, 30L)));
    expect(taskClient.getCurrentOffsets("id1", true)).andReturn(ImmutableMap.of(0, 10L, 2, 30L));
    expect(taskQueue.add(capture(captured))).andReturn(true);
    taskRunner.registerListener(anyObject(TaskRunnerListener.class), anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    SupervisorReport report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    Assert.assertTrue(report.getPayload() instanceof KafkaSupervisorReport.KafkaSupervisorReportPayload);
    KafkaSupervisorReport.KafkaSupervisorReportPayload payload = (KafkaSupervisorReport.KafkaSupervisorReportPayload) report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, (long) payload.getDurationSeconds());
    Assert.assertEquals(NUM_PARTITIONS, (int) payload.getPartitions());
    Assert.assertEquals(1, (int) payload.getReplicas());
    Assert.assertEquals(KAFKA_TOPIC, payload.getTopic());
    Assert.assertEquals(0, payload.getActiveTasks().size());
    Assert.assertEquals(1, payload.getPublishingTasks().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(KafkaTuningConfig.copyOf(tuningConfig), capturedTask.getTuningConfig());
    KafkaIOConfig 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());
    Assert.assertFalse("pauseAfterRead", capturedTaskConfig.isPauseAfterRead());
    // check that the new task was created with starting offsets matching where the publishing task finished
    Assert.assertEquals(KAFKA_TOPIC, capturedTaskConfig.getStartPartitions().getTopic());
    Assert.assertEquals(10L, (long) capturedTaskConfig.getStartPartitions().getPartitionOffsetMap().get(0));
    Assert.assertEquals(0L, (long) capturedTaskConfig.getStartPartitions().getPartitionOffsetMap().get(1));
    Assert.assertEquals(30L, (long) capturedTaskConfig.getStartPartitions().getPartitionOffsetMap().get(2));
    Assert.assertEquals(KAFKA_TOPIC, capturedTaskConfig.getEndPartitions().getTopic());
    Assert.assertEquals(Long.MAX_VALUE, (long) capturedTaskConfig.getEndPartitions().getPartitionOffsetMap().get(0));
    Assert.assertEquals(Long.MAX_VALUE, (long) capturedTaskConfig.getEndPartitions().getPartitionOffsetMap().get(1));
    Assert.assertEquals(Long.MAX_VALUE, (long) capturedTaskConfig.getEndPartitions().getPartitionOffsetMap().get(2));
}
Also used : TaskRunnerListener(io.druid.indexing.overlord.TaskRunnerListener) RealtimeIndexTask(io.druid.indexing.common.task.RealtimeIndexTask) Task(io.druid.indexing.common.task.Task) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) KafkaPartitions(io.druid.indexing.kafka.KafkaPartitions) ArrayList(java.util.ArrayList) TaskLocation(io.druid.indexing.common.TaskLocation) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) Executor(java.util.concurrent.Executor) SupervisorReport(io.druid.indexing.overlord.supervisor.SupervisorReport) KafkaIOConfig(io.druid.indexing.kafka.KafkaIOConfig) Collection(java.util.Collection) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) Test(org.junit.Test)

Example 3 with SupervisorReport

use of io.druid.indexing.overlord.supervisor.SupervisorReport in project druid by druid-io.

the class KafkaSupervisorTest method testDiscoverExistingPublishingAndReadingTask.

@Test
public void testDiscoverExistingPublishingAndReadingTask() throws Exception {
    final TaskLocation location1 = new TaskLocation("testHost", 1234);
    final TaskLocation location2 = new TaskLocation("testHost2", 145);
    final DateTime startTime = new DateTime();
    supervisor = getSupervisor(1, 1, true, "PT1H", null);
    addSomeEvents(1);
    Task id1 = createKafkaIndexTask("id1", DATASOURCE, "sequenceName-0", new KafkaPartitions("topic", ImmutableMap.of(0, 0L, 1, 0L, 2, 0L)), new KafkaPartitions("topic", ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null);
    Task id2 = createKafkaIndexTask("id2", DATASOURCE, "sequenceName-0", new KafkaPartitions("topic", ImmutableMap.of(0, 10L, 1, 20L, 2, 30L)), new KafkaPartitions("topic", ImmutableMap.of(0, Long.MAX_VALUE, 1, Long.MAX_VALUE, 2, Long.MAX_VALUE)), null);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(id1.getId(), null, location1));
    workItems.add(new TestTaskRunnerWorkItem(id2.getId(), null, location2));
    expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.of(id1, id2)).anyTimes();
    expect(taskStorage.getStatus("id1")).andReturn(Optional.of(TaskStatus.running("id1"))).anyTimes();
    expect(taskStorage.getStatus("id2")).andReturn(Optional.of(TaskStatus.running("id2"))).anyTimes();
    expect(taskStorage.getTask("id1")).andReturn(Optional.of(id1)).anyTimes();
    expect(taskStorage.getTask("id2")).andReturn(Optional.of(id2)).anyTimes();
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.PUBLISHING));
    expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.READING));
    expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    expect(taskClient.getCurrentOffsetsAsync("id1", false)).andReturn(Futures.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 10L, 1, 20L, 2, 30L)));
    expect(taskClient.getCurrentOffsets("id1", true)).andReturn(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L));
    expect(taskClient.getCurrentOffsetsAsync("id2", false)).andReturn(Futures.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 40L, 1, 50L, 2, 60L)));
    taskRunner.registerListener(anyObject(TaskRunnerListener.class), anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    SupervisorReport report = supervisor.getStatus();
    verifyAll();
    Assert.assertEquals(DATASOURCE, report.getId());
    Assert.assertTrue(report.getPayload() instanceof KafkaSupervisorReport.KafkaSupervisorReportPayload);
    KafkaSupervisorReport.KafkaSupervisorReportPayload payload = (KafkaSupervisorReport.KafkaSupervisorReportPayload) report.getPayload();
    Assert.assertEquals(DATASOURCE, payload.getDataSource());
    Assert.assertEquals(3600L, (long) payload.getDurationSeconds());
    Assert.assertEquals(NUM_PARTITIONS, (int) payload.getPartitions());
    Assert.assertEquals(1, (int) payload.getReplicas());
    Assert.assertEquals(KAFKA_TOPIC, payload.getTopic());
    Assert.assertEquals(1, payload.getActiveTasks().size());
    Assert.assertEquals(1, payload.getPublishingTasks().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, 10L, 1, 20L, 2, 30L), activeReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 40L, 1, 50L, 2, 60L), activeReport.getCurrentOffsets());
    Assert.assertEquals("id1", publishingReport.getId());
    Assert.assertEquals(ImmutableMap.of(0, 0L, 1, 0L, 2, 0L), publishingReport.getStartingOffsets());
    Assert.assertEquals(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L), publishingReport.getCurrentOffsets());
}
Also used : TaskRunnerListener(io.druid.indexing.overlord.TaskRunnerListener) RealtimeIndexTask(io.druid.indexing.common.task.RealtimeIndexTask) Task(io.druid.indexing.common.task.Task) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) KafkaPartitions(io.druid.indexing.kafka.KafkaPartitions) ArrayList(java.util.ArrayList) TaskLocation(io.druid.indexing.common.TaskLocation) DateTime(org.joda.time.DateTime) Executor(java.util.concurrent.Executor) SupervisorReport(io.druid.indexing.overlord.supervisor.SupervisorReport) Collection(java.util.Collection) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) Test(org.junit.Test)

Aggregations

ImmutableMap (com.google.common.collect.ImmutableMap)3 TaskLocation (io.druid.indexing.common.TaskLocation)3 RealtimeIndexTask (io.druid.indexing.common.task.RealtimeIndexTask)3 Task (io.druid.indexing.common.task.Task)3 KafkaDataSourceMetadata (io.druid.indexing.kafka.KafkaDataSourceMetadata)3 KafkaIndexTask (io.druid.indexing.kafka.KafkaIndexTask)3 KafkaPartitions (io.druid.indexing.kafka.KafkaPartitions)3 TaskRunnerListener (io.druid.indexing.overlord.TaskRunnerListener)3 SupervisorReport (io.druid.indexing.overlord.supervisor.SupervisorReport)3 ArrayList (java.util.ArrayList)3 Collection (java.util.Collection)3 Map (java.util.Map)3 Executor (java.util.concurrent.Executor)3 Test (org.junit.Test)3 KafkaIOConfig (io.druid.indexing.kafka.KafkaIOConfig)2 DateTime (org.joda.time.DateTime)1