Search in sources :

Example 11 with TaskToolbox

use of io.druid.indexing.common.TaskToolbox 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.<InputRow>of(new MapBasedInputRow(now, ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("dim1", "foo", "met1", "1")), new MapBasedInputRow(now.minus(new Period("P1D")), ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("dim1", "foo", "met1", 2.0)), new MapBasedInputRow(now, ImmutableList.of("dim2"), ImmutableMap.<String, Object>of("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, "rows"));
    Assert.assertEquals(3, sumMetric(task, "met1"));
    // 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(TaskStatus.Status.SUCCESS, taskStatus.getStatusCode());
}
Also used : Period(org.joda.time.Period) TaskStatus(io.druid.indexing.common.TaskStatus) DataSegment(io.druid.timeline.DataSegment) TaskToolbox(io.druid.indexing.common.TaskToolbox) Executor(java.util.concurrent.Executor) TestIndexerMetadataStorageCoordinator(io.druid.indexing.test.TestIndexerMetadataStorageCoordinator) SegmentDescriptor(io.druid.query.SegmentDescriptor) MapBasedInputRow(io.druid.data.input.MapBasedInputRow) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) Pair(io.druid.java.util.common.Pair) Test(org.junit.Test)

Example 12 with TaskToolbox

use of io.druid.indexing.common.TaskToolbox 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.<InputRow>asList(// Good row- will be processed.
    new MapBasedInputRow(now, ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("dim1", "foo", "met1", "1")), // Null row- will be unparseable.
    null, // Bad metric- will count as processed, but that particular metric won't update.
    new MapBasedInputRow(now, ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("dim1", "foo", "met1", "foo")), // Bad row- will be unparseable.
    new MapBasedInputRow(now, ImmutableList.of("dim1", FAIL_DIM), ImmutableMap.<String, Object>of("dim1", "foo", "met1", 2.0)), // Old row- will be thrownAway.
    new MapBasedInputRow(now.minus(new Period("P1D")), ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("dim1", "foo", "met1", 2.0)), // Good row- will be processed.
    new MapBasedInputRow(now, ImmutableList.of("dim2"), ImmutableMap.<String, Object>of("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, "rows"));
    Assert.assertEquals(3, sumMetric(task, "met1"));
    // 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(TaskStatus.Status.SUCCESS, taskStatus.getStatusCode());
}
Also used : Period(org.joda.time.Period) TaskStatus(io.druid.indexing.common.TaskStatus) DataSegment(io.druid.timeline.DataSegment) TaskToolbox(io.druid.indexing.common.TaskToolbox) Executor(java.util.concurrent.Executor) TestIndexerMetadataStorageCoordinator(io.druid.indexing.test.TestIndexerMetadataStorageCoordinator) SegmentDescriptor(io.druid.query.SegmentDescriptor) MapBasedInputRow(io.druid.data.input.MapBasedInputRow) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) Pair(io.druid.java.util.common.Pair) Test(org.junit.Test)

Example 13 with TaskToolbox

use of io.druid.indexing.common.TaskToolbox in project druid by druid-io.

the class RealtimeIndexTaskTest method testRestoreCorruptData.

@Test(timeout = 60_000L)
public void testRestoreCorruptData() throws Exception {
    final File directory = tempFolder.newFolder();
    final RealtimeIndexTask task1 = makeRealtimeTask(null);
    // 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.<InputRow>of(new MapBasedInputRow(now, ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("dim1", "foo"))));
        // Trigger graceful shutdown.
        task1.stopGracefully();
        // 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(TaskStatus.Status.SUCCESS, taskStatus.getStatusCode());
        // Nothing should be published.
        Assert.assertEquals(Sets.newHashSet(), mdc.getPublished());
    }
    // Corrupt the data:
    final File smooshFile = new File(String.format("%s/persistent/task/%s/work/persist/%s/%s_%s/0/00000.smoosh", directory, task1.getId(), task1.getDataSource(), Granularities.DAY.bucketStart(now), Granularities.DAY.bucketEnd(now)));
    Files.write(smooshFile.toPath(), "oops!".getBytes(Charsets.UTF_8));
    // 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 the task to finish.
        boolean caught = false;
        try {
            statusFuture.get();
        } catch (Exception e) {
            caught = true;
        }
        Assert.assertTrue("expected exception", caught);
    }
}
Also used : TaskToolbox(io.druid.indexing.common.TaskToolbox) TestIndexerMetadataStorageCoordinator(io.druid.indexing.test.TestIndexerMetadataStorageCoordinator) MapBasedInputRow(io.druid.data.input.MapBasedInputRow) InputRow(io.druid.data.input.InputRow) ListenableFuture(com.google.common.util.concurrent.ListenableFuture) MapBasedInputRow(io.druid.data.input.MapBasedInputRow) TaskStatus(io.druid.indexing.common.TaskStatus) File(java.io.File) EntryExistsException(io.druid.metadata.EntryExistsException) ParseException(io.druid.java.util.common.parsers.ParseException) IOException(java.io.IOException) ExecutionException(java.util.concurrent.ExecutionException) ExpectedException(org.junit.rules.ExpectedException) Test(org.junit.Test)

Example 14 with TaskToolbox

use of io.druid.indexing.common.TaskToolbox 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, 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.<InputRow>of(new MapBasedInputRow(now, ImmutableList.of("dim1"), ImmutableMap.<String, Object>of("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();
}
Also used : TaskToolbox(io.druid.indexing.common.TaskToolbox) TestIndexerMetadataStorageCoordinator(io.druid.indexing.test.TestIndexerMetadataStorageCoordinator) MapBasedInputRow(io.druid.data.input.MapBasedInputRow) TaskStatus(io.druid.indexing.common.TaskStatus) Test(org.junit.Test)

Example 15 with TaskToolbox

use of io.druid.indexing.common.TaskToolbox in project druid by druid-io.

the class ThreadPoolTaskRunner method run.

@Override
public ListenableFuture<TaskStatus> run(final Task task) {
    final TaskToolbox toolbox = toolboxFactory.build(task);
    final Object taskPriorityObj = task.getContextValue(TaskThreadPriority.CONTEXT_KEY);
    int taskPriority = 0;
    if (taskPriorityObj != null) {
        if (taskPriorityObj instanceof Number) {
            taskPriority = ((Number) taskPriorityObj).intValue();
        } else if (taskPriorityObj instanceof String) {
            try {
                taskPriority = Integer.parseInt(taskPriorityObj.toString());
            } catch (NumberFormatException e) {
                log.error(e, "Error parsing task priority [%s] for task [%s]", taskPriorityObj, task.getId());
            }
        }
    }
    // Ensure an executor for that priority exists
    if (!exec.containsKey(taskPriority)) {
        final ListeningExecutorService executorService = buildExecutorService(taskPriority);
        if (exec.putIfAbsent(taskPriority, executorService) != null) {
            // favor prior service
            executorService.shutdownNow();
        }
    }
    final ListenableFuture<TaskStatus> statusFuture = exec.get(taskPriority).submit(new ThreadPoolTaskRunnerCallable(task, location, toolbox));
    final ThreadPoolTaskRunnerWorkItem taskRunnerWorkItem = new ThreadPoolTaskRunnerWorkItem(task, location, statusFuture);
    runningItems.add(taskRunnerWorkItem);
    Futures.addCallback(statusFuture, new FutureCallback<TaskStatus>() {

        @Override
        public void onSuccess(TaskStatus result) {
            runningItems.remove(taskRunnerWorkItem);
        }

        @Override
        public void onFailure(Throwable t) {
            runningItems.remove(taskRunnerWorkItem);
        }
    });
    return statusFuture;
}
Also used : TaskStatus(io.druid.indexing.common.TaskStatus) TaskToolbox(io.druid.indexing.common.TaskToolbox) ListeningExecutorService(com.google.common.util.concurrent.ListeningExecutorService)

Aggregations

TaskToolbox (io.druid.indexing.common.TaskToolbox)16 TaskStatus (io.druid.indexing.common.TaskStatus)13 Test (org.junit.Test)12 DataSegment (io.druid.timeline.DataSegment)9 TestIndexerMetadataStorageCoordinator (io.druid.indexing.test.TestIndexerMetadataStorageCoordinator)8 MapBasedInputRow (io.druid.data.input.MapBasedInputRow)7 File (java.io.File)6 TaskLock (io.druid.indexing.common.TaskLock)5 LockListAction (io.druid.indexing.common.actions.LockListAction)5 Interval (org.joda.time.Interval)5 SegmentInsertAction (io.druid.indexing.common.actions.SegmentInsertAction)4 Pair (io.druid.java.util.common.Pair)4 SegmentDescriptor (io.druid.query.SegmentDescriptor)4 ListenableFuture (com.google.common.util.concurrent.ListenableFuture)3 InputRow (io.druid.data.input.InputRow)3 TaskActionClient (io.druid.indexing.common.actions.TaskActionClient)3 AbstractFixedIntervalTask (io.druid.indexing.common.task.AbstractFixedIntervalTask)3 IndexTask (io.druid.indexing.common.task.IndexTask)3 KillTask (io.druid.indexing.common.task.KillTask)3 RealtimeIndexTask (io.druid.indexing.common.task.RealtimeIndexTask)3