Examples with SeekableStreamEndSequenceNumbers org.apache.druid.indexing.seekablestream.SeekableStreamEndSequenceNumbers used on opensource projects

Example 6 with SeekableStreamEndSequenceNumbers

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

the class KinesisIndexTaskTest method testMultipleParseExceptionsFailure.

@Test(timeout = 120_000L)
public void testMultipleParseExceptionsFailure() throws Exception {
    reportParseExceptions = false;
    maxParseExceptions = 2;
    maxSavedParseExceptions = 2;
    recordSupplier.assign(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(recordSupplier.getEarliestSequenceNumber(EasyMock.anyObject())).andReturn("0").anyTimes();
    recordSupplier.seek(EasyMock.anyObject(), EasyMock.anyString());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(recordSupplier.poll(EasyMock.anyLong())).andReturn(generateRecords(2, 13)).once();
    recordSupplier.close();
    EasyMock.expectLastCall().once();
    replayAll();
    final KinesisIndexTask task = createTask(null, new KinesisIndexTaskIOConfig(0, "sequence0", new SeekableStreamStartSequenceNumbers<>(STREAM, ImmutableMap.of(SHARD_ID1, "2"), ImmutableSet.of()), new SeekableStreamEndSequenceNumbers<>(STREAM, ImmutableMap.of(SHARD_ID1, "9")), true, null, null, INPUT_FORMAT, "awsEndpoint", null, null, null, null, false));
    final ListenableFuture<TaskStatus> future = runTask(task);
    TaskStatus status = future.get();
    // Wait for task to exit
    Assert.assertEquals(TaskState.FAILED, status.getStatusCode());
    verifyAll();
    IndexTaskTest.checkTaskStatusErrorMsgForParseExceptionsExceeded(status);
    // Check metrics
    Assert.assertEquals(3, task.getRunner().getRowIngestionMeters().getProcessed());
    Assert.assertEquals(0, task.getRunner().getRowIngestionMeters().getProcessedWithError());
    Assert.assertEquals(3, task.getRunner().getRowIngestionMeters().getUnparseable());
    Assert.assertEquals(0, task.getRunner().getRowIngestionMeters().getThrownAway());
    // Check published metadata
    Assert.assertEquals(ImmutableList.of(), publishedDescriptors());
    Assert.assertNull(newDataSchemaMetadata());
    IngestionStatsAndErrorsTaskReportData reportData = getTaskReportData();
    Map<String, Object> expectedMetrics = ImmutableMap.of(RowIngestionMeters.BUILD_SEGMENTS, ImmutableMap.of(RowIngestionMeters.PROCESSED, 3, RowIngestionMeters.PROCESSED_WITH_ERROR, 0, RowIngestionMeters.UNPARSEABLE, 3, RowIngestionMeters.THROWN_AWAY, 0));
    Assert.assertEquals(expectedMetrics, reportData.getRowStats());
    List<LinkedHashMap> parseExceptionReports = (List<LinkedHashMap>) reportData.getUnparseableEvents().get(RowIngestionMeters.BUILD_SEGMENTS);
    List<String> expectedMessages = Arrays.asList("Unable to parse [] as the intermediateRow resulted in empty input row (Record: 1)", "Unable to parse row [unparseable] (Record: 1)");
    List<String> actualMessages = parseExceptionReports.stream().map((r) -> {
        return ((List<String>) r.get("details")).get(0);
    }).collect(Collectors.toList());
    Assert.assertEquals(expectedMessages, actualMessages);
    List<String> expectedInputs = Arrays.asList("", "unparseable");
    List<String> actualInputs = parseExceptionReports.stream().map((r) -> {
        return (String) r.get("input");
    }).collect(Collectors.toList());
    Assert.assertEquals(expectedInputs, actualInputs);
}

Example 7 with SeekableStreamEndSequenceNumbers

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

the class KafkaSupervisorTest method testCreateBaseTaskContexts.

@Test
public void testCreateBaseTaskContexts() throws JsonProcessingException {
    supervisor = getTestableSupervisor(1, 1, true, "PT1H", null, null);
    final Map<String, Object> contexts = supervisor.createIndexTasks(1, "seq", OBJECT_MAPPER, new TreeMap<>(), new KafkaIndexTaskIOConfig(0, "seq", new SeekableStreamStartSequenceNumbers<>("test", Collections.emptyMap(), Collections.emptySet()), new SeekableStreamEndSequenceNumbers<>("test", Collections.emptyMap()), Collections.emptyMap(), null, null, null, null, INPUT_FORMAT), new KafkaIndexTaskTuningConfig(null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null), null).get(0).getContext();
    final Boolean contextValue = (Boolean) contexts.get("IS_INCREMENTAL_HANDOFF_SUPPORTED");
    Assert.assertNotNull(contextValue);
    Assert.assertTrue(contextValue);
}

Example 8 with SeekableStreamEndSequenceNumbers

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

the class KafkaSupervisorTest method testGetOffsetFromStorageForPartitionWithResetOffsetAutomatically.

@Test
public void testGetOffsetFromStorageForPartitionWithResetOffsetAutomatically() throws Exception {
    addSomeEvents(2);
    supervisor = getTestableSupervisor(1, 1, true, true, "PT1H", null, null);
    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();
    taskRunner.registerListener(EasyMock.anyObject(TaskRunnerListener.class), EasyMock.anyObject(Executor.class));
    EasyMock.reset(indexerMetadataStorageCoordinator);
    // unknown DataSourceMetadata in metadata store
    // for simplicity in testing the offset availability check, we use negative stored offsets in metadata here,
    // because the stream's earliest offset is 0, although that would not happen in real usage.
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KafkaDataSourceMetadata(new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(1, -100L, 2, 200L)))).times(4);
    // getOffsetFromStorageForPartition() throws an exception when the offsets are automatically reset.
    // Since getOffsetFromStorageForPartition() is called per partition, all partitions can't be reset at the same time.
    // Instead, subsequent partitions will be reset in the following supervisor runs.
    EasyMock.expect(indexerMetadataStorageCoordinator.resetDataSourceMetadata(DATASOURCE, new KafkaDataSourceMetadata(// Only one partition is reset in a single supervisor run.
    new SeekableStreamEndSequenceNumbers<>(topic, ImmutableMap.of(2, 200L))))).andReturn(true);
    replayAll();
    supervisor.start();
    supervisor.runInternal();
    verifyAll();
}

Example 9 with SeekableStreamEndSequenceNumbers

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

the class KinesisSupervisorTest method testCheckpointForInactiveTaskGroup.

@Test(timeout = 60_000L)
public void testCheckpointForInactiveTaskGroup() throws InterruptedException {
    supervisor = getTestableSupervisor(2, 1, true, "PT1S", null, null, false);
    // not adding any events
    final KinesisIndexTask id1;
    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);
    final Task id2 = createKinesisIndexTask("id2", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "10", SHARD_ID0, "20"), 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);
    final Task id3 = createKinesisIndexTask("id3", DATASOURCE, 0, new SeekableStreamStartSequenceNumbers<>("stream", ImmutableMap.of(SHARD_ID1, "10", SHARD_ID0, "20"), 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);
    final TaskLocation location1 = new TaskLocation("testHost", 1234, -1);
    final TaskLocation location2 = new TaskLocation("testHost2", 145, -1);
    Collection workItems = new ArrayList<>();
    workItems.add(new TestTaskRunnerWorkItem(id1, null, location1));
    workItems.add(new TestTaskRunnerWorkItem(id2, null, location2));
    workItems.add(new TestTaskRunnerWorkItem(id2, null, location2));
    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(taskRunner.getRunningTasks()).andReturn(workItems).anyTimes();
    EasyMock.expect(taskMaster.getTaskQueue()).andReturn(Optional.of(taskQueue)).anyTimes();
    EasyMock.expect(taskMaster.getTaskRunner()).andReturn(Optional.of(taskRunner)).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 KinesisDataSourceMetadata(null)).anyTimes();
    EasyMock.expect(taskClient.getStatusAsync("id1")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id2")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    EasyMock.expect(taskClient.getStatusAsync("id3")).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.READING));
    final DateTime startTime = DateTimes.nowUtc();
    EasyMock.expect(taskClient.getStartTimeAsync("id1")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getStartTimeAsync("id2")).andReturn(Futures.immediateFuture(startTime));
    EasyMock.expect(taskClient.getStartTimeAsync("id3")).andReturn(Futures.immediateFuture(startTime));
    final TreeMap<Integer, Map<String, String>> checkpoints = new TreeMap<>();
    checkpoints.put(0, ImmutableMap.of(SHARD_ID1, "10", SHARD_ID0, "20"));
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("id1"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpoints)).times(1);
    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();
    supervisor.moveTaskGroupToPendingCompletion(0);
    supervisor.checkpoint(0, new KinesisDataSourceMetadata(new SeekableStreamStartSequenceNumbers<>(STREAM, checkpoints.get(0), checkpoints.get(0).keySet())));
    while (supervisor.getNoticesQueueSize() > 0) {
        Thread.sleep(100);
    }
    verifyAll();
    Assert.assertNull(serviceEmitter.getStackTrace(), serviceEmitter.getStackTrace());
    Assert.assertNull(serviceEmitter.getExceptionMessage(), serviceEmitter.getExceptionMessage());
    Assert.assertNull(serviceEmitter.getExceptionClass());
}

Example 10 with SeekableStreamEndSequenceNumbers

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

the class KinesisSupervisorTest method testShardMergePhaseThree.

/**
 * Test task creation after a shard merge with two closed shards and one open shard, with the closed shards
 * expiring and no longer being returned from record supplier.
 *
 * @param phaseTwoTasks List of tasks from the second phase where closed but not expired shards were present.
 */
private void testShardMergePhaseThree(List<Task> phaseTwoTasks) throws Exception {
    EasyMock.reset(indexerMetadataStorageCoordinator);
    EasyMock.reset(taskStorage);
    EasyMock.reset(taskQueue);
    EasyMock.reset(taskClient);
    EasyMock.reset(taskMaster);
    EasyMock.reset(taskRunner);
    EasyMock.reset(supervisorRecordSupplier);
    // second set of tasks ran, shard 0 has expired, but shard 1 and 2 have data
    EasyMock.expect(indexerMetadataStorageCoordinator.retrieveDataSourceMetadata(DATASOURCE)).andReturn(new KinesisDataSourceMetadata(new SeekableStreamEndSequenceNumbers<String, String>(STREAM, ImmutableMap.of(SHARD_ID0, KinesisSequenceNumber.END_OF_SHARD_MARKER, SHARD_ID1, KinesisSequenceNumber.END_OF_SHARD_MARKER, SHARD_ID2, "100")))).anyTimes();
    EasyMock.expect(indexerMetadataStorageCoordinator.resetDataSourceMetadata(DATASOURCE, new KinesisDataSourceMetadata(new SeekableStreamEndSequenceNumbers<String, String>(STREAM, ImmutableMap.of(SHARD_ID0, KinesisSequenceNumber.EXPIRED_MARKER, SHARD_ID1, KinesisSequenceNumber.EXPIRED_MARKER, SHARD_ID2, "100"))))).andReturn(true).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getPartitionIds(STREAM)).andReturn(ImmutableSet.of(SHARD_ID2)).anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getAssignment()).andReturn(ImmutableSet.of(SHARD2_PARTITION)).anyTimes();
    supervisorRecordSupplier.seekToLatest(EasyMock.anyObject());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getEarliestSequenceNumber(EasyMock.anyObject())).andReturn("0").anyTimes();
    EasyMock.expect(supervisorRecordSupplier.getLatestSequenceNumber(new StreamPartition<>(STREAM, SHARD_ID2))).andReturn("200").anyTimes();
    supervisorRecordSupplier.seek(EasyMock.anyObject(), EasyMock.anyString());
    EasyMock.expectLastCall().anyTimes();
    Capture<Task> postSplitCaptured = 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(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.NOT_STARTED)).anyTimes();
    Task successfulTask0 = phaseTwoTasks.get(0);
    EasyMock.expect(taskStorage.getStatus(successfulTask0.getId())).andReturn(Optional.of(TaskStatus.success(successfulTask0.getId())));
    EasyMock.expect(taskStorage.getTask(successfulTask0.getId())).andReturn(Optional.of(successfulTask0)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).anyTimes();
    EasyMock.expect(taskQueue.add(EasyMock.capture(postSplitCaptured))).andReturn(true).times(1);
    replayAll();
    supervisor.runInternal();
    verifyAll();
    EasyMock.reset(taskStorage);
    EasyMock.reset(taskClient);
    EasyMock.expect(taskClient.getStatusAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(SeekableStreamIndexTaskRunner.Status.NOT_STARTED)).anyTimes();
    EasyMock.expect(taskClient.getStartTimeAsync(EasyMock.anyString())).andReturn(Futures.immediateFuture(DateTimes.nowUtc())).anyTimes();
    TreeMap<Integer, Map<String, String>> checkpointsGroup0 = new TreeMap<>();
    checkpointsGroup0.put(0, ImmutableMap.of(SHARD_ID2, "100"));
    // there would be 1 task, only task group 0 has a shard
    EasyMock.expect(taskClient.getCheckpointsAsync(EasyMock.contains("sequenceName-0"), EasyMock.anyBoolean())).andReturn(Futures.immediateFuture(checkpointsGroup0)).times(1);
    List<Task> postSplitTasks = postSplitCaptured.getValues();
    EasyMock.expect(taskStorage.getActiveTasksByDatasource(DATASOURCE)).andReturn(postSplitTasks).anyTimes();
    for (Task task : postSplitTasks) {
        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.replay(taskStorage);
    EasyMock.replay(taskClient);
    supervisor.runInternal();
    verifyAll();
    // Check that shardId-000000000000 which has hit EOS is not included in the sequences sent to the task for group 0
    SeekableStreamStartSequenceNumbers<String, String> group0ExpectedStartSequenceNumbers = new SeekableStreamStartSequenceNumbers<>(STREAM, ImmutableMap.of(SHARD_ID2, "100"), ImmutableSet.of(SHARD_ID2));
    SeekableStreamEndSequenceNumbers<String, String> group0ExpectedEndSequenceNumbers = new SeekableStreamEndSequenceNumbers<>(STREAM, ImmutableMap.of(SHARD_ID2, KinesisSequenceNumber.NO_END_SEQUENCE_NUMBER));
    Assert.assertEquals(1, postSplitTasks.size());
    KinesisIndexTaskIOConfig group0Config = ((KinesisIndexTask) postSplitTasks.get(0)).getIOConfig();
    Assert.assertEquals((Integer) 0, group0Config.getTaskGroupId());
    Assert.assertEquals(group0ExpectedStartSequenceNumbers, group0Config.getStartSequenceNumbers());
    Assert.assertEquals(group0ExpectedEndSequenceNumbers, group0Config.getEndSequenceNumbers());
    Map<Integer, Set<String>> expectedPartitionGroups = ImmutableMap.of(0, ImmutableSet.of(SHARD_ID2), 1, ImmutableSet.of());
    ConcurrentHashMap<String, String> expectedPartitionOffsets = new ConcurrentHashMap<>(ImmutableMap.of(SHARD_ID2, "-1", SHARD_ID1, "-1", SHARD_ID0, "-1"));
    Assert.assertEquals(expectedPartitionGroups, supervisor.getPartitionGroups());
    Assert.assertEquals(expectedPartitionOffsets, supervisor.getPartitionOffsets());
}