use of io.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);
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());
}
use of io.druid.indexing.kafka.KafkaDataSourceMetadata in project druid by druid-io.
the class KafkaSupervisorTest method testReplicas.
@Test
public void testReplicas() throws Exception {
supervisor = getSupervisor(2, 1, true, "PT1H", null);
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);
Assert.assertEquals(3, task1.getIOConfig().getStartPartitions().getPartitionOffsetMap().size());
Assert.assertEquals(3, task1.getIOConfig().getEndPartitions().getPartitionOffsetMap().size());
Assert.assertEquals(0L, (long) task1.getIOConfig().getStartPartitions().getPartitionOffsetMap().get(0));
Assert.assertEquals(0L, (long) task1.getIOConfig().getStartPartitions().getPartitionOffsetMap().get(1));
Assert.assertEquals(0L, (long) task1.getIOConfig().getStartPartitions().getPartitionOffsetMap().get(2));
KafkaIndexTask task2 = captured.getValues().get(1);
Assert.assertEquals(3, task2.getIOConfig().getStartPartitions().getPartitionOffsetMap().size());
Assert.assertEquals(3, task2.getIOConfig().getEndPartitions().getPartitionOffsetMap().size());
Assert.assertEquals(0L, (long) task2.getIOConfig().getStartPartitions().getPartitionOffsetMap().get(0));
Assert.assertEquals(0L, (long) task2.getIOConfig().getStartPartitions().getPartitionOffsetMap().get(1));
Assert.assertEquals(0L, (long) task2.getIOConfig().getStartPartitions().getPartitionOffsetMap().get(2));
}
use of io.druid.indexing.kafka.KafkaDataSourceMetadata in project druid by druid-io.
the class KafkaSupervisorTest method testKillUnresponsiveTasksWhileSettingEndOffsets.
@Test
public void testKillUnresponsiveTasksWhileSettingEndOffsets() throws Exception {
final TaskLocation location = new TaskLocation("testHost", 1234);
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.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 10L, 1, 20L, 2, 30L))).andReturn(Futures.immediateFuture((Map<Integer, Long>) ImmutableMap.of(0, 10L, 1, 15L, 2, 35L)));
expect(taskClient.setEndOffsetsAsync(EasyMock.contains("sequenceName-0"), EasyMock.eq(ImmutableMap.of(0, 10L, 1, 20L, 2, 35L)), EasyMock.eq(true))).andReturn(Futures.<Boolean>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));
}
}
use of io.druid.indexing.kafka.KafkaDataSourceMetadata in project druid by druid-io.
the class KafkaSupervisorTest method testKillUnresponsiveTasksWhileGettingStartTime.
@Test
public void testKillUnresponsiveTasksWhileGettingStartTime() throws Exception {
supervisor = getSupervisor(2, 2, true, "PT1H", null);
addSomeEvents(1);
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();
reset(taskStorage, taskClient, taskQueue);
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(taskClient.getStatusAsync(task.getId())).andReturn(Futures.immediateFuture(KafkaIndexTask.Status.NOT_STARTED));
expect(taskClient.getStartTimeAsync(task.getId())).andReturn(Futures.<DateTime>immediateFailedFuture(new RuntimeException()));
taskQueue.shutdown(task.getId());
}
replay(taskStorage, taskClient, taskQueue);
supervisor.runInternal();
verifyAll();
}
use of io.druid.indexing.kafka.KafkaDataSourceMetadata in project druid by druid-io.
the class KafkaSupervisorTest method testNoInitialState.
@Test
public void testNoInitialState() throws Exception {
supervisor = getSupervisor(1, 1, true, "PT1H", null);
addSomeEvents(1);
Capture<KafkaIndexTask> captured = Capture.newInstance();
expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).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);
taskRunner.registerListener(anyObject(TaskRunnerListener.class), anyObject(Executor.class));
replayAll();
supervisor.start();
supervisor.runInternal();
verifyAll();
KafkaIndexTask task = captured.getValue();
Assert.assertEquals(dataSchema, task.getDataSchema());
Assert.assertEquals(KafkaTuningConfig.copyOf(tuningConfig), task.getTuningConfig());
KafkaIOConfig taskConfig = task.getIOConfig();
Assert.assertEquals(kafkaHost, taskConfig.getConsumerProperties().get("bootstrap.servers"));
Assert.assertEquals("myCustomValue", taskConfig.getConsumerProperties().get("myCustomKey"));
Assert.assertEquals("sequenceName-0", taskConfig.getBaseSequenceName());
Assert.assertTrue("isUseTransaction", taskConfig.isUseTransaction());
Assert.assertFalse("pauseAfterRead", taskConfig.isPauseAfterRead());
Assert.assertFalse("minimumMessageTime", taskConfig.getMinimumMessageTime().isPresent());
Assert.assertEquals(KAFKA_TOPIC, taskConfig.getStartPartitions().getTopic());
Assert.assertEquals(0L, (long) taskConfig.getStartPartitions().getPartitionOffsetMap().get(0));
Assert.assertEquals(0L, (long) taskConfig.getStartPartitions().getPartitionOffsetMap().get(1));
Assert.assertEquals(0L, (long) taskConfig.getStartPartitions().getPartitionOffsetMap().get(2));
Assert.assertEquals(KAFKA_TOPIC, taskConfig.getEndPartitions().getTopic());
Assert.assertEquals(Long.MAX_VALUE, (long) taskConfig.getEndPartitions().getPartitionOffsetMap().get(0));
Assert.assertEquals(Long.MAX_VALUE, (long) taskConfig.getEndPartitions().getPartitionOffsetMap().get(1));
Assert.assertEquals(Long.MAX_VALUE, (long) taskConfig.getEndPartitions().getPartitionOffsetMap().get(2));
}
Aggregations