Search in sources :

Example 1 with TestFirehose

use of org.apache.druid.indexing.common.TestFirehose in project druid by druid-io.

the class RealtimeIndexTaskTest method testTransformSpec.

@Test(timeout = 60_000L)
public void testTransformSpec() throws Exception {
    final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
    final TransformSpec transformSpec = new TransformSpec(new SelectorDimFilter("dim1", "foo", null), ImmutableList.of(new ExpressionTransform("dim1t", "concat(dim1,dim1)", ExprMacroTable.nil())));
    final RealtimeIndexTask task = makeRealtimeTask(null, transformSpec, true, 0);
    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(1, task.getMetrics().processed());
    Assert.assertEquals(2, task.getMetrics().thrownAway());
    Assert.assertEquals(0, task.getMetrics().unparseable());
    // Do some queries.
    Assert.assertEquals(1, sumMetric(task, null, "rows").longValue());
    Assert.assertEquals(1, sumMetric(task, new SelectorDimFilter("dim1t", "foofoo", null), "rows").longValue());
    if (NullHandling.replaceWithDefault()) {
        Assert.assertEquals(0, sumMetric(task, new SelectorDimFilter("dim1t", "barbar", null), "rows").longValue());
    } else {
        Assert.assertNull(sumMetric(task, new SelectorDimFilter("dim1t", "barbar", null), "rows"));
    }
    Assert.assertEquals(1, 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());
}
Also used : Period(org.joda.time.Period) TaskStatus(org.apache.druid.indexer.TaskStatus) DataSegment(org.apache.druid.timeline.DataSegment) TransformSpec(org.apache.druid.segment.transform.TransformSpec) TestFirehose(org.apache.druid.indexing.common.TestFirehose) TaskToolbox(org.apache.druid.indexing.common.TaskToolbox) Executor(java.util.concurrent.Executor) TestIndexerMetadataStorageCoordinator(org.apache.druid.indexing.test.TestIndexerMetadataStorageCoordinator) SelectorDimFilter(org.apache.druid.query.filter.SelectorDimFilter) SegmentDescriptor(org.apache.druid.query.SegmentDescriptor) ExpressionTransform(org.apache.druid.segment.transform.ExpressionTransform) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) Pair(org.apache.druid.java.util.common.Pair) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Example 2 with TestFirehose

use of org.apache.druid.indexing.common.TestFirehose 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.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());
    }
    // Corrupt the data:
    final File smooshFile = new File(StringUtils.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(), StringUtils.toUtf8("oops!"));
    // 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(org.apache.druid.indexing.common.TaskToolbox) TestIndexerMetadataStorageCoordinator(org.apache.druid.indexing.test.TestIndexerMetadataStorageCoordinator) ListenableFuture(com.google.common.util.concurrent.ListenableFuture) TaskStatus(org.apache.druid.indexer.TaskStatus) File(java.io.File) ParseException(org.apache.druid.java.util.common.parsers.ParseException) EntryExistsException(org.apache.druid.metadata.EntryExistsException) ExecutionException(java.util.concurrent.ExecutionException) ExpectedException(org.junit.rules.ExpectedException) TestFirehose(org.apache.druid.indexing.common.TestFirehose) HashSet(java.util.HashSet) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Example 3 with TestFirehose

use of org.apache.druid.indexing.common.TestFirehose in project druid by druid-io.

the class RealtimeIndexTaskTest method testRestoreAfterHandoffAttemptDuringShutdown.

@Test(timeout = 60_000L)
public void testRestoreAfterHandoffAttemptDuringShutdown() throws Exception {
    final TaskStorage taskStorage = new HeapMemoryTaskStorage(new TaskStorageConfig(null));
    final TestIndexerMetadataStorageCoordinator mdc = new TestIndexerMetadataStorageCoordinator();
    final File directory = tempFolder.newFolder();
    final RealtimeIndexTask task1 = makeRealtimeTask(null);
    final DataSegment publishedSegment;
    // First run:
    {
        final TaskToolbox taskToolbox = makeToolbox(task1, taskStorage, 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")));
        // Stop the firehose, this will trigger a finishJob.
        firehose.close();
        // Wait for publish.
        while (mdc.getPublished().isEmpty()) {
            Thread.sleep(50);
        }
        publishedSegment = Iterables.getOnlyElement(mdc.getPublished());
        // Do a query.
        Assert.assertEquals(1, sumMetric(task1, null, "rows").longValue());
        // Trigger graceful shutdown.
        task1.stopGracefully(taskToolbox.getConfig());
        // Wait for the task to finish. The status doesn't really matter.
        while (!statusFuture.isDone()) {
            Thread.sleep(50);
        }
    }
    // Second run:
    {
        final RealtimeIndexTask task2 = makeRealtimeTask(task1.getId());
        final TaskToolbox taskToolbox = makeToolbox(task2, taskStorage, 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);
        }
        // Stop the firehose again, this will start another handoff.
        final TestFirehose firehose = (TestFirehose) task2.getFirehose();
        // Stop the firehose, this will trigger a finishJob.
        firehose.close();
        // publishedSegment is still published. No reason it shouldn't be.
        Assert.assertEquals(ImmutableSet.of(publishedSegment), mdc.getPublished());
        // Wait for a handoffCallback to show up.
        while (handOffCallbacks.isEmpty()) {
            Thread.sleep(50);
        }
        // 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());
    }
}
Also used : TaskStorageConfig(org.apache.druid.indexing.common.config.TaskStorageConfig) HeapMemoryTaskStorage(org.apache.druid.indexing.overlord.HeapMemoryTaskStorage) TaskStatus(org.apache.druid.indexer.TaskStatus) DataSegment(org.apache.druid.timeline.DataSegment) TestFirehose(org.apache.druid.indexing.common.TestFirehose) TaskToolbox(org.apache.druid.indexing.common.TaskToolbox) Executor(java.util.concurrent.Executor) TaskStorage(org.apache.druid.indexing.overlord.TaskStorage) HeapMemoryTaskStorage(org.apache.druid.indexing.overlord.HeapMemoryTaskStorage) TestIndexerMetadataStorageCoordinator(org.apache.druid.indexing.test.TestIndexerMetadataStorageCoordinator) SegmentDescriptor(org.apache.druid.query.SegmentDescriptor) ListenableFuture(com.google.common.util.concurrent.ListenableFuture) File(java.io.File) Pair(org.apache.druid.java.util.common.Pair) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Example 4 with TestFirehose

use of org.apache.druid.indexing.common.TestFirehose 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());
}
Also used : Period(org.joda.time.Period) TaskStatus(org.apache.druid.indexer.TaskStatus) DataSegment(org.apache.druid.timeline.DataSegment) TestFirehose(org.apache.druid.indexing.common.TestFirehose) TaskToolbox(org.apache.druid.indexing.common.TaskToolbox) Executor(java.util.concurrent.Executor) TestIndexerMetadataStorageCoordinator(org.apache.druid.indexing.test.TestIndexerMetadataStorageCoordinator) SegmentDescriptor(org.apache.druid.query.SegmentDescriptor) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) Pair(org.apache.druid.java.util.common.Pair) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Example 5 with TestFirehose

use of org.apache.druid.indexing.common.TestFirehose 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();
}
Also used : TaskToolbox(org.apache.druid.indexing.common.TaskToolbox) TestIndexerMetadataStorageCoordinator(org.apache.druid.indexing.test.TestIndexerMetadataStorageCoordinator) TaskStatus(org.apache.druid.indexer.TaskStatus) TestFirehose(org.apache.druid.indexing.common.TestFirehose) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Aggregations

TaskStatus (org.apache.druid.indexer.TaskStatus)8 TaskToolbox (org.apache.druid.indexing.common.TaskToolbox)8 TestFirehose (org.apache.druid.indexing.common.TestFirehose)8 TestIndexerMetadataStorageCoordinator (org.apache.druid.indexing.test.TestIndexerMetadataStorageCoordinator)8 InitializedNullHandlingTest (org.apache.druid.testing.InitializedNullHandlingTest)8 Test (org.junit.Test)8 Executor (java.util.concurrent.Executor)5 Pair (org.apache.druid.java.util.common.Pair)5 SegmentDescriptor (org.apache.druid.query.SegmentDescriptor)5 DataSegment (org.apache.druid.timeline.DataSegment)5 ImmutableMap (com.google.common.collect.ImmutableMap)3 ListenableFuture (com.google.common.util.concurrent.ListenableFuture)3 File (java.io.File)3 Map (java.util.Map)3 ConcurrentHashMap (java.util.concurrent.ConcurrentHashMap)3 Period (org.joda.time.Period)3 HashSet (java.util.HashSet)2 ParseException (org.apache.druid.java.util.common.parsers.ParseException)2 ExecutionException (java.util.concurrent.ExecutionException)1 TaskStorageConfig (org.apache.druid.indexing.common.config.TaskStorageConfig)1