Examples with TaskStatus org.apache.druid.indexer.TaskStatus used on opensource projects

Example 81 with TaskStatus

use of org.apache.druid.indexer.TaskStatus in project druid by druid-io.

the class IndexTaskTest method testMultipleParseExceptionsFailureAtDeterminePartitions.

@Test
public void testMultipleParseExceptionsFailureAtDeterminePartitions() throws Exception {
    final File tmpDir = temporaryFolder.newFolder();
    final File tmpFile = File.createTempFile("druid", "index", tmpDir);
    try (BufferedWriter writer = Files.newWriter(tmpFile, StandardCharsets.UTF_8)) {
        writer.write("time,dim,dimLong,dimFloat,val\n");
        // unparseable
        writer.write("unparseable,a,2,3.0,1\n");
        // valid row
        writer.write("2014-01-01T00:00:10Z,a,2,3.0,1\n");
        // unparseable
        writer.write("9.0,a,2,3.0,1\n");
        // thrown away
        writer.write("3014-03-01T00:00:10Z,outsideofinterval,2,3.0,1\n");
        // unparseable
        writer.write("99999999999-01-01T00:00:10Z,b,2,3.0,1\n");
    }
    // Allow up to 3 parse exceptions, and save up to 2 parse exceptions
    final IndexTuningConfig tuningConfig = new IndexTuningConfig(null, null, null, null, null, null, null, null, null, null, new HashedPartitionsSpec(2, null, null), INDEX_SPEC, null, null, true, false, null, null, null, true, 2, 5, null, null);
    final TimestampSpec timestampSpec = new TimestampSpec("time", "auto", null);
    final DimensionsSpec dimensionsSpec = new DimensionsSpec(Arrays.asList(new StringDimensionSchema("dim"), new LongDimensionSchema("dimLong"), new FloatDimensionSchema("dimFloat")));
    final List<String> columns = Arrays.asList("time", "dim", "dimLong", "dimFloat", "val");
    final IndexIngestionSpec ingestionSpec;
    List<String> expectedMessages;
    if (useInputFormatApi) {
        ingestionSpec = createIngestionSpec(jsonMapper, tmpDir, timestampSpec, dimensionsSpec, new CsvInputFormat(columns, null, null, true, 0), null, null, tuningConfig, false, false);
        expectedMessages = Arrays.asList(StringUtils.format("Timestamp[99999999999-01-01T00:00:10Z] is unparseable! Event: {time=99999999999-01-01T00:00:10Z, dim=b, dimLong=2, dimFloat=3.0, val=1} (Path: %s, Record: 3, Line: 6)", tmpFile.toURI()), StringUtils.format("Timestamp[9.0] is unparseable! Event: {time=9.0, dim=a, dimLong=2, dimFloat=3.0, val=1} (Path: %s, Record: 2, Line: 4)", tmpFile.toURI()), StringUtils.format("Timestamp[unparseable] is unparseable! Event: {time=unparseable, dim=a, dimLong=2, dimFloat=3.0, val=1} (Path: %s, Record: 1, Line: 2)", tmpFile.toURI()));
    } else {
        ingestionSpec = createIngestionSpec(jsonMapper, tmpDir, new CSVParseSpec(timestampSpec, dimensionsSpec, null, columns, true, 0), null, null, tuningConfig, false, false);
        expectedMessages = Arrays.asList("Timestamp[99999999999-01-01T00:00:10Z] is unparseable! Event: {time=99999999999-01-01T00:00:10Z, dim=b, dimLong=2, dimFloat=3.0, val=1}", "Timestamp[9.0] is unparseable! Event: {time=9.0, dim=a, dimLong=2, dimFloat=3.0, val=1}", "Timestamp[unparseable] is unparseable! Event: {time=unparseable, dim=a, dimLong=2, dimFloat=3.0, val=1}");
    }
    IndexTask indexTask = new IndexTask(null, null, ingestionSpec, null);
    TaskStatus status = runTask(indexTask).lhs;
    Assert.assertEquals(TaskState.FAILED, status.getStatusCode());
    checkTaskStatusErrorMsgForParseExceptionsExceeded(status);
    IngestionStatsAndErrorsTaskReportData reportData = getTaskReportData();
    Map<String, Object> expectedMetrics = ImmutableMap.of(RowIngestionMeters.DETERMINE_PARTITIONS, ImmutableMap.of(RowIngestionMeters.PROCESSED_WITH_ERROR, 0, RowIngestionMeters.PROCESSED, 1, RowIngestionMeters.UNPARSEABLE, 3, RowIngestionMeters.THROWN_AWAY, useInputFormatApi ? 1 : 2), RowIngestionMeters.BUILD_SEGMENTS, ImmutableMap.of(RowIngestionMeters.PROCESSED_WITH_ERROR, 0, RowIngestionMeters.PROCESSED, 0, RowIngestionMeters.UNPARSEABLE, 0, RowIngestionMeters.THROWN_AWAY, 0));
    Assert.assertEquals(expectedMetrics, reportData.getRowStats());
    List<LinkedHashMap> parseExceptionReports = (List<LinkedHashMap>) reportData.getUnparseableEvents().get(RowIngestionMeters.DETERMINE_PARTITIONS);
    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("{time=99999999999-01-01T00:00:10Z, dim=b, dimLong=2, dimFloat=3.0, val=1}", "{time=9.0, dim=a, dimLong=2, dimFloat=3.0, val=1}", "{time=unparseable, dim=a, dimLong=2, dimFloat=3.0, val=1}");
    List<String> actualInputs = parseExceptionReports.stream().map((r) -> {
        return (String) r.get("input");
    }).collect(Collectors.toList());
    Assert.assertEquals(expectedInputs, actualInputs);
}

Example 82 with TaskStatus

use of org.apache.druid.indexer.TaskStatus in project druid by druid-io.

the class RealtimeIndexTaskTest method testBasics.

@Test(timeout = 60_000L)
public void testBasics() throws Exception {
    final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
    final RealtimeIndexTask task = makeRealtimeTask(null);
    final TaskToolbox taskToolbox = makeToolbox(task, mdc, tempFolder.newFolder());
    final ListenableFuture<TaskStatus> statusFuture = runTask(task, taskToolbox);
    final DataSegment publishedSegment;
    // Wait for firehose to show up, it starts off null.
    while (task.getFirehose() == null) {
        Thread.sleep(50);
    }
    final TestFirehose firehose = (TestFirehose) task.getFirehose();
    firehose.addRows(ImmutableList.of(ImmutableMap.of("t", now.getMillis(), "dim1", "foo", "met1", "1"), ImmutableMap.of("t", now.minus(new Period("P1D")).getMillis(), "dim1", "foo", "met1", 2.0), ImmutableMap.of("t", now.getMillis(), "dim2", "bar", "met1", 2.0)));
    // Stop the firehose, this will drain out existing events.
    firehose.close();
    // Wait for publish.
    while (mdc.getPublished().isEmpty()) {
        Thread.sleep(50);
    }
    publishedSegment = Iterables.getOnlyElement(mdc.getPublished());
    // Check metrics.
    Assert.assertEquals(2, task.getMetrics().processed());
    Assert.assertEquals(1, task.getMetrics().thrownAway());
    Assert.assertEquals(0, task.getMetrics().unparseable());
    // Do some queries.
    Assert.assertEquals(2, sumMetric(task, null, "rows").longValue());
    Assert.assertEquals(3, sumMetric(task, null, "met1").longValue());
    // Simulate handoff.
    for (Map.Entry<SegmentDescriptor, Pair<Executor, Runnable>> entry : handOffCallbacks.entrySet()) {
        final Pair<Executor, Runnable> executorRunnablePair = entry.getValue();
        Assert.assertEquals(new SegmentDescriptor(publishedSegment.getInterval(), publishedSegment.getVersion(), publishedSegment.getShardSpec().getPartitionNum()), entry.getKey());
        executorRunnablePair.lhs.execute(executorRunnablePair.rhs);
    }
    handOffCallbacks.clear();
    // Wait for the task to finish.
    final TaskStatus taskStatus = statusFuture.get();
    Assert.assertEquals(TaskState.SUCCESS, taskStatus.getStatusCode());
}

Example 83 with TaskStatus

use of org.apache.druid.indexer.TaskStatus in project druid by druid-io.

the class RealtimeIndexTaskTest method testHandoffTimeout.

@Test(timeout = 60_000L, expected = ExecutionException.class)
public void testHandoffTimeout() throws Exception {
    final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
    final RealtimeIndexTask task = makeRealtimeTask(null, TransformSpec.NONE, true, 100L);
    final TaskToolbox taskToolbox = makeToolbox(task, mdc, tempFolder.newFolder());
    final ListenableFuture<TaskStatus> statusFuture = runTask(task, taskToolbox);
    // Wait for firehose to show up, it starts off null.
    while (task.getFirehose() == null) {
        Thread.sleep(50);
    }
    final TestFirehose firehose = (TestFirehose) task.getFirehose();
    firehose.addRows(ImmutableList.of(ImmutableMap.of("t", now.getMillis(), "dim1", "foo", "met1", "1")));
    // Stop the firehose, this will drain out existing events.
    firehose.close();
    // Wait for publish.
    while (mdc.getPublished().isEmpty()) {
        Thread.sleep(50);
    }
    Assert.assertEquals(1, task.getMetrics().processed());
    Assert.assertNotNull(Iterables.getOnlyElement(mdc.getPublished()));
    // handoff would timeout, resulting in exception
    statusFuture.get();
}

Example 84 with TaskStatus

use of org.apache.druid.indexer.TaskStatus in project druid by druid-io.

the class RealtimeIndexTaskTest method testNoReportParseExceptions.

@Test(timeout = 60_000L)
public void testNoReportParseExceptions() throws Exception {
    final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
    final RealtimeIndexTask task = makeRealtimeTask(null, false);
    final TaskToolbox taskToolbox = makeToolbox(task, mdc, tempFolder.newFolder());
    final ListenableFuture<TaskStatus> statusFuture = runTask(task, taskToolbox);
    final DataSegment publishedSegment;
    // Wait for firehose to show up, it starts off null.
    while (task.getFirehose() == null) {
        Thread.sleep(50);
    }
    final TestFirehose firehose = (TestFirehose) task.getFirehose();
    firehose.addRows(Arrays.asList(// Good row- will be processed.
    ImmutableMap.of("t", now.getMillis(), "dim1", "foo", "met1", "1"), // Null row- will be thrown away.
    null, // Bad metric- will count as processed, but that particular metric won't update.
    ImmutableMap.of("t", now.getMillis(), "dim1", "foo", "met1", "foo"), // Bad row- will be unparseable.
    ImmutableMap.of("dim1", "foo", "met1", 2.0, TestFirehose.FAIL_DIM, "x"), // Old row- will be thrownAway.
    ImmutableMap.of("t", now.minus(Period.days(1)).getMillis(), "dim1", "foo", "met1", 2.0), // Good row- will be processed.
    ImmutableMap.of("t", now.getMillis(), "dim2", "bar", "met1", 2.0)));
    // Stop the firehose, this will drain out existing events.
    firehose.close();
    // Wait for publish.
    while (mdc.getPublished().isEmpty()) {
        Thread.sleep(50);
    }
    publishedSegment = Iterables.getOnlyElement(mdc.getPublished());
    // Check metrics.
    Assert.assertEquals(3, task.getMetrics().processed());
    Assert.assertEquals(1, task.getMetrics().thrownAway());
    Assert.assertEquals(2, task.getMetrics().unparseable());
    // Do some queries.
    Assert.assertEquals(3, sumMetric(task, null, "rows").longValue());
    Assert.assertEquals(3, sumMetric(task, null, "met1").longValue());
    // Simulate handoff.
    for (Map.Entry<SegmentDescriptor, Pair<Executor, Runnable>> entry : handOffCallbacks.entrySet()) {
        final Pair<Executor, Runnable> executorRunnablePair = entry.getValue();
        Assert.assertEquals(new SegmentDescriptor(publishedSegment.getInterval(), publishedSegment.getVersion(), publishedSegment.getShardSpec().getPartitionNum()), entry.getKey());
        executorRunnablePair.lhs.execute(executorRunnablePair.rhs);
    }
    handOffCallbacks.clear();
    // Wait for the task to finish.
    final TaskStatus taskStatus = statusFuture.get();
    Assert.assertEquals(TaskState.SUCCESS, taskStatus.getStatusCode());
}

Example 85 with TaskStatus

use of org.apache.druid.indexer.TaskStatus in project druid by druid-io.

the class RealtimeIndexTaskTest method testRestore.

@Test(timeout = 60_000L)
public void testRestore() throws Exception {
    final File directory = tempFolder.newFolder();
    final RealtimeIndexTask task1 = makeRealtimeTask(null);
    final DataSegment publishedSegment;
    // First run:
    {
        final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
        final TaskToolbox taskToolbox = makeToolbox(task1, mdc, directory);
        final ListenableFuture<TaskStatus> statusFuture = runTask(task1, taskToolbox);
        // Wait for firehose to show up, it starts off null.
        while (task1.getFirehose() == null) {
            Thread.sleep(50);
        }
        final TestFirehose firehose = (TestFirehose) task1.getFirehose();
        firehose.addRows(ImmutableList.of(ImmutableMap.of("t", now.getMillis(), "dim1", "foo")));
        // Trigger graceful shutdown.
        task1.stopGracefully(taskToolbox.getConfig());
        // Wait for the task to finish. The status doesn't really matter, but we'll check it anyway.
        final TaskStatus taskStatus = statusFuture.get();
        Assert.assertEquals(TaskState.SUCCESS, taskStatus.getStatusCode());
        // Nothing should be published.
        Assert.assertEquals(new HashSet<>(), mdc.getPublished());
    }
    // Second run:
    {
        final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
        final RealtimeIndexTask task2 = makeRealtimeTask(task1.getId());
        final TaskToolbox taskToolbox = makeToolbox(task2, mdc, directory);
        final ListenableFuture<TaskStatus> statusFuture = runTask(task2, taskToolbox);
        // Wait for firehose to show up, it starts off null.
        while (task2.getFirehose() == null) {
            Thread.sleep(50);
        }
        // Do a query, at this point the previous data should be loaded.
        Assert.assertEquals(1, sumMetric(task2, null, "rows").longValue());
        final TestFirehose firehose = (TestFirehose) task2.getFirehose();
        firehose.addRows(ImmutableList.of(ImmutableMap.of("t", now.getMillis(), "dim2", "bar")));
        // Stop the firehose, this will drain out existing events.
        firehose.close();
        // Wait for publish.
        while (mdc.getPublished().isEmpty()) {
            Thread.sleep(50);
        }
        publishedSegment = Iterables.getOnlyElement(mdc.getPublished());
        // Do a query.
        Assert.assertEquals(2, sumMetric(task2, null, "rows").longValue());
        // Simulate handoff.
        for (Map.Entry<SegmentDescriptor, Pair<Executor, Runnable>> entry : handOffCallbacks.entrySet()) {
            final Pair<Executor, Runnable> executorRunnablePair = entry.getValue();
            Assert.assertEquals(new SegmentDescriptor(publishedSegment.getInterval(), publishedSegment.getVersion(), publishedSegment.getShardSpec().getPartitionNum()), entry.getKey());
            executorRunnablePair.lhs.execute(executorRunnablePair.rhs);
        }
        handOffCallbacks.clear();
        // Wait for the task to finish.
        final TaskStatus taskStatus = statusFuture.get();
        Assert.assertEquals(TaskState.SUCCESS, taskStatus.getStatusCode());
    }
}