Search in sources :

Example 11 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testBadMetadataOffsets.

@Test(expected = ISE.class)
public void testBadMetadataOffsets() throws Exception {
    supervisor = getSupervisor(1, 1, true, "PT1H", null);
    addSomeEvents(1);
    expect(taskMaster.getTaskRunner()).andReturn(Optional.<TaskRunner>absent()).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.<Task>of()).anyTimes();
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(new KafkaPartitions(KAFKA_TOPIC, ImmutableMap.of(0, 10L, 1, 20L, 2, 30L)))).anyTimes();
    replayAll();
    supervisor.start();
    supervisor.runInternal();
}
Also used : 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) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) TaskRunner(io.druid.indexing.overlord.TaskRunner) Test(org.junit.Test)

Example 12 with KafkaDataSourceMetadata

use of io.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);
    supervisor = getSupervisor(2, 2, true, "PT1M", null);
    addSomeEvents(100);
    Capture<Task> captured = Capture.newInstance(CaptureType.ALL);
    expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    expect(taskRunner.getRunningTasks()).andReturn(Collections.EMPTY_LIST).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.<Task>of()).anyTimes();
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    expect(taskQueue.add(capture(captured))).andReturn(true).times(4);
    taskRunner.registerListener(anyObject(TaskRunnerListener.class), 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.getId(), null, location));
    }
    reset(taskStorage, taskRunner, taskClient, taskQueue);
    captured = Capture.newInstance(CaptureType.ALL);
    expect(taskStorage.getActiveTasks()).andReturn(tasks).anyTimes();
    for (Task task : tasks) {
        expect(taskStorage.getStatus(task.getId())).andReturn(Optional.of(TaskStatus.running(task.getId()))).anyTimes();
        expect(taskStorage.getTask(task.getId())).andReturn(Optional.of(task)).anyTimes();
    }
    expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    expect(taskClient.getStatusAsync(anyString())).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.READING)).anyTimes();
    expect(taskClient.getStartTimeAsync(EasyMock.contains("sequenceName-0"))).andReturn(Futures.immediateFuture(DateTime.now().minusMinutes(2))).andReturn(Futures.immediateFuture(DateTime.now()));
    expect(taskClient.getStartTimeAsync(EasyMock.contains("sequenceName-1"))).andReturn(Futures.immediateFuture(DateTime.now())).times(2);
    expect(taskClient.pauseAsync(EasyMock.contains("sequenceName-0"))).andReturn(Futures.<Map<Integer, Long>>immediateFailedFuture(new RuntimeException())).times(2);
    taskQueue.shutdown(EasyMock.contains("sequenceName-0"));
    expectLastCall().times(2);
    expect(taskQueue.add(capture(captured))).andReturn(true).times(2);
    replay(taskStorage, taskRunner, taskClient, taskQueue);
    supervisor.runInternal();
    verifyAll();
    for (Task task : captured.getValues()) {
        KafkaIOConfig taskConfig = ((KafkaIndexTask) task).getIOConfig();
        Assert.assertEquals(0L, (long) taskConfig.getStartPartitions().getPartitionOffsetMap().get(0));
        Assert.assertEquals(0L, (long) taskConfig.getStartPartitions().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) ArrayList(java.util.ArrayList) TaskLocation(io.druid.indexing.common.TaskLocation) Executor(java.util.concurrent.Executor) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) 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 13 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testLateMessageRejectionPeriod.

@Test
public void testLateMessageRejectionPeriod() throws Exception {
    supervisor = getSupervisor(2, 1, true, "PT1H", new Period("PT1H"));
    addSomeEvents(1);
    Capture<KafkaIndexTask> captured = Capture.newInstance(CaptureType.ALL);
    expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    expect(taskMaster.getTaskRunner()).andReturn(Optional.<TaskRunner>absent()).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.<Task>of()).anyTimes();
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(null)).anyTimes();
    expect(taskQueue.add(capture(captured))).andReturn(true).times(2);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    KafkaIndexTask task1 = captured.getValues().get(0);
    KafkaIndexTask task2 = captured.getValues().get(1);
    Assert.assertTrue("minimumMessageTime", task1.getIOConfig().getMinimumMessageTime().get().plusMinutes(59).isBeforeNow());
    Assert.assertTrue("minimumMessageTime", task1.getIOConfig().getMinimumMessageTime().get().plusMinutes(61).isAfterNow());
    Assert.assertEquals(task1.getIOConfig().getMinimumMessageTime().get(), task2.getIOConfig().getMinimumMessageTime().get());
}
Also used : RealtimeIndexTask(io.druid.indexing.common.task.RealtimeIndexTask) Task(io.druid.indexing.common.task.Task) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) KafkaIndexTask(io.druid.indexing.kafka.KafkaIndexTask) Period(org.joda.time.Period) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) TaskRunner(io.druid.indexing.overlord.TaskRunner) Test(org.junit.Test)

Example 14 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testStopGracefully.

@Test
public void testStopGracefully() throws Exception {
    final TaskLocation location1 = new TaskLocation("testHost", 1234);
    final TaskLocation location2 = new TaskLocation("testHost2", 145);
    final DateTime startTime = new DateTime();
    supervisor = getSupervisor(2, 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);
    Task id3 = createKafkaIndexTask("id3", 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, id3)).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.getStatus("id3")).andReturn(Optional.of(TaskStatus.running("id3"))).anyTimes();
    expect(taskStorage.getTask("id1")).andReturn(Optional.of(id1)).anyTimes();
    expect(taskStorage.getTask("id2")).andReturn(Optional.of(id2)).anyTimes();
    expect(taskStorage.getTask("id3")).andReturn(Optional.of(id3)).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.getStatusAsync("id3")).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.READING));
    expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    expect(taskClient.getStartTimeAsync("id3")).andReturn(Futures.immediateFuture(startTime));
    expect(taskClient.getCurrentOffsets("id1", true)).andReturn(ImmutableMap.of(0, 10L, 1, 20L, 2, 30L));
    taskRunner.registerListener(anyObject(TaskRunnerListener.class), anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    reset(taskRunner, taskClient, taskQueue);
    expect(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    expect(taskClient.pauseAsync("id2")).andReturn(Futures.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 15L, 1, 25L, 2, 30L)));
    expect(taskClient.setEndOffsetsAsync("id2", ImmutableMap.of(0, 15L, 1, 25L, 2, 30L), true)).andReturn(Futures.immediateFuture(true));
    taskQueue.shutdown("id3");
    expectLastCall().times(2);
    replay(taskRunner, taskClient, taskQueue);
    supervisor.gracefulShutdownInternal();
    verifyAll();
}
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) Executor(java.util.concurrent.Executor) KafkaPartitions(io.druid.indexing.kafka.KafkaPartitions) ArrayList(java.util.ArrayList) Collection(java.util.Collection) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) TaskLocation(io.druid.indexing.common.TaskLocation) DateTime(org.joda.time.DateTime) Test(org.junit.Test)

Example 15 with KafkaDataSourceMetadata

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

the class KafkaSupervisorTest method testResetNoDataSourceMetadata.

@Test
public void testResetNoDataSourceMetadata() throws Exception {
    supervisor = getSupervisor(1, 1, true, "PT1H", null);
    expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).anyTimes();
    expect(taskRunner.getRunningTasks()).andReturn(Collections.EMPTY_LIST).anyTimes();
    expect(taskStorage.getActiveTasks()).andReturn(ImmutableList.<Task>of()).anyTimes();
    taskRunner.registerListener(anyObject(TaskRunnerListener.class), anyObject(Executor.class));
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
    KafkaDataSourceMetadata resetMetadata = new KafkaDataSourceMetadata(new KafkaPartitions(KAFKA_TOPIC, ImmutableMap.of(1, 1000L, 2, 1000L)));
    reset(indexerMetadataStorageCoordinator);
    // no DataSourceMetadata in metadata store
    expect(indexerMetadataStorageCoordinator.getDataSourceMetadata(DATASOURCE)).andReturn(null);
    replay(indexerMetadataStorageCoordinator);
    supervisor.resetInternal(resetMetadata);
    verifyAll();
}
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) Executor(java.util.concurrent.Executor) KafkaPartitions(io.druid.indexing.kafka.KafkaPartitions) KafkaDataSourceMetadata(io.druid.indexing.kafka.KafkaDataSourceMetadata) Test(org.junit.Test)

Aggregations

KafkaDataSourceMetadata (io.druid.indexing.kafka.KafkaDataSourceMetadata)24 RealtimeIndexTask (io.druid.indexing.common.task.RealtimeIndexTask)23 Task (io.druid.indexing.common.task.Task)23 KafkaIndexTask (io.druid.indexing.kafka.KafkaIndexTask)23 Test (org.junit.Test)23 TaskRunnerListener (io.druid.indexing.overlord.TaskRunnerListener)17 Executor (java.util.concurrent.Executor)17 KafkaPartitions (io.druid.indexing.kafka.KafkaPartitions)12 TaskLocation (io.druid.indexing.common.TaskLocation)8 ArrayList (java.util.ArrayList)8 Collection (java.util.Collection)8 ImmutableMap (com.google.common.collect.ImmutableMap)7 KafkaIOConfig (io.druid.indexing.kafka.KafkaIOConfig)7 Map (java.util.Map)7 TaskRunner (io.druid.indexing.overlord.TaskRunner)6 DateTime (org.joda.time.DateTime)4 SupervisorReport (io.druid.indexing.overlord.supervisor.SupervisorReport)3 EasyMock.anyString (org.easymock.EasyMock.anyString)3 DataSourceMetadata (io.druid.indexing.overlord.DataSourceMetadata)2 VisibleForTesting (com.google.common.annotations.VisibleForTesting)1