Examples with CompositeDataMovementEvent org.apache.tez.runtime.api.events.CompositeDataMovementEvent used on opensource projects

Example 11 with CompositeDataMovementEvent

use of org.apache.tez.runtime.api.events.CompositeDataMovementEvent in project tez by apache.

the class TestOrderedPartitionedKVOutput2 method testNonStartedOutput.

@Test(timeout = 5000)
public void testNonStartedOutput() throws IOException {
    OutputContext outputContext = OutputTestHelpers.createOutputContext(conf, workingDir);
    int numPartitions = 10;
    OrderedPartitionedKVOutput output = new OrderedPartitionedKVOutput(outputContext, numPartitions);
    output.initialize();
    List<Event> events = output.close();
    assertEquals(2, events.size());
    Event event1 = events.get(0);
    assertTrue(event1 instanceof VertexManagerEvent);
    Event event2 = events.get(1);
    assertTrue(event2 instanceof CompositeDataMovementEvent);
    CompositeDataMovementEvent cdme = (CompositeDataMovementEvent) event2;
    ByteBuffer bb = cdme.getUserPayload();
    ShuffleUserPayloads.DataMovementEventPayloadProto shufflePayload = ShuffleUserPayloads.DataMovementEventPayloadProto.parseFrom(ByteString.copyFrom(bb));
    assertTrue(shufflePayload.hasEmptyPartitions());
    byte[] emptyPartitions = TezCommonUtils.decompressByteStringToByteArray(shufflePayload.getEmptyPartitions());
    BitSet emptyPartionsBitSet = TezUtilsInternal.fromByteArray(emptyPartitions);
    assertEquals(numPartitions, emptyPartionsBitSet.cardinality());
    for (int i = 0; i < numPartitions; i++) {
        assertTrue(emptyPartionsBitSet.get(i));
    }
}

Example 12 with CompositeDataMovementEvent

use of org.apache.tez.runtime.api.events.CompositeDataMovementEvent in project tez by apache.

the class TestUnorderedPartitionedKVOutput2 method testNonStartedOutput.

@Test(timeout = 5000)
public void testNonStartedOutput() throws Exception {
    OutputContext outputContext = OutputTestHelpers.createOutputContext();
    int numPartitions = 1;
    UnorderedPartitionedKVOutput output = new UnorderedPartitionedKVOutput(outputContext, numPartitions);
    output.initialize();
    List<Event> events = output.close();
    assertEquals(1, events.size());
    Event event1 = events.get(0);
    assertTrue(event1 instanceof CompositeDataMovementEvent);
    CompositeDataMovementEvent dme = (CompositeDataMovementEvent) event1;
    ByteBuffer bb = dme.getUserPayload();
    ShuffleUserPayloads.DataMovementEventPayloadProto shufflePayload = ShuffleUserPayloads.DataMovementEventPayloadProto.parseFrom(ByteString.copyFrom(bb));
    assertTrue(shufflePayload.hasEmptyPartitions());
    byte[] emptyPartitions = TezCommonUtils.decompressByteStringToByteArray(shufflePayload.getEmptyPartitions());
    BitSet emptyPartionsBitSet = TezUtilsInternal.fromByteArray(emptyPartitions);
    assertEquals(numPartitions, emptyPartionsBitSet.cardinality());
    for (int i = 0; i < numPartitions; i++) {
        assertTrue(emptyPartionsBitSet.get(i));
    }
}

Example 13 with CompositeDataMovementEvent

use of org.apache.tez.runtime.api.events.CompositeDataMovementEvent in project tez by apache.

the class TestCompositeDataMovementEvent method testGetEvents.

@Test(timeout = 5000)
public void testGetEvents() {
    int numOutputs = 0;
    int startIndex = 1;
    CompositeDataMovementEvent cdme2 = CompositeDataMovementEvent.create(startIndex, numOutputs, userPayload);
    for (DataMovementEvent dme : cdme2.getEvents()) {
        numOutputs++;
    }
    Assert.assertEquals(numOutputs, cdme2.getCount());
}

Example 14 with CompositeDataMovementEvent

use of org.apache.tez.runtime.api.events.CompositeDataMovementEvent in project tez by apache.

the class TestCompositeDataMovementEvent method testGetCount.

@Test(timeout = 5000)
public void testGetCount() {
    int numPartitions = 2;
    int startIndex = 2;
    CompositeDataMovementEvent cdme1 = CompositeDataMovementEvent.create(startIndex, numPartitions, userPayload);
    Assert.assertEquals(numPartitions, cdme1.getCount());
    Assert.assertEquals(startIndex, cdme1.getSourceIndexStart());
}

Example 15 with CompositeDataMovementEvent

use of org.apache.tez.runtime.api.events.CompositeDataMovementEvent in project tez by apache.

the class TestReduceProcessor method testReduceProcessor.

@Test(timeout = 5000)
public void testReduceProcessor() throws Exception {
    final String dagName = "mrdag0";
    String mapVertexName = MultiStageMRConfigUtil.getInitialMapVertexName();
    String reduceVertexName = MultiStageMRConfigUtil.getFinalReduceVertexName();
    JobConf jobConf = new JobConf(defaultConf);
    setUpJobConf(jobConf);
    MRHelpers.translateMRConfToTez(jobConf);
    jobConf.setInt(MRJobConfig.APPLICATION_ATTEMPT_ID, 0);
    jobConf.set(MRFrameworkConfigs.TASK_LOCAL_RESOURCE_DIR, new Path(workDir, "localized-resources").toUri().toString());
    jobConf.setBoolean(MRJobConfig.MR_TEZ_SPLITS_VIA_EVENTS, false);
    Path mapInput = new Path(workDir, "map0");
    MapUtils.generateInputSplit(localFs, workDir, jobConf, mapInput, 10);
    InputSpec mapInputSpec = new InputSpec("NullSrcVertex", InputDescriptor.create(MRInputLegacy.class.getName()).setUserPayload(UserPayload.create(ByteBuffer.wrap(MRRuntimeProtos.MRInputUserPayloadProto.newBuilder().setConfigurationBytes(TezUtils.createByteStringFromConf(jobConf)).build().toByteArray()))), 1);
    OutputSpec mapOutputSpec = new OutputSpec("NullDestVertex", OutputDescriptor.create(OrderedPartitionedKVOutput.class.getName()).setUserPayload(TezUtils.createUserPayloadFromConf(jobConf)), 1);
    // Run a map
    TestUmbilical testUmbilical = new TestUmbilical();
    TezSharedExecutor sharedExecutor = new TezSharedExecutor(jobConf);
    LogicalIOProcessorRuntimeTask mapTask = MapUtils.createLogicalTask(localFs, workDir, jobConf, 0, mapInput, testUmbilical, dagName, mapVertexName, Collections.singletonList(mapInputSpec), Collections.singletonList(mapOutputSpec), sharedExecutor);
    mapTask.initialize();
    mapTask.run();
    mapTask.close();
    // One VME, One DME
    Assert.assertEquals(2, testUmbilical.getEvents().size());
    Assert.assertEquals(EventType.VERTEX_MANAGER_EVENT, testUmbilical.getEvents().get(0).getEventType());
    Assert.assertEquals(EventType.COMPOSITE_DATA_MOVEMENT_EVENT, testUmbilical.getEvents().get(1).getEventType());
    CompositeDataMovementEvent cdmEvent = (CompositeDataMovementEvent) testUmbilical.getEvents().get(1).getEvent();
    Assert.assertEquals(1, cdmEvent.getCount());
    DataMovementEvent dme = cdmEvent.getEvents().iterator().next();
    dme.setTargetIndex(0);
    LOG.info("Starting reduce...");
    JobTokenIdentifier identifier = new JobTokenIdentifier(new Text(dagName));
    JobTokenSecretManager jobTokenSecretManager = new JobTokenSecretManager();
    Token<JobTokenIdentifier> shuffleToken = new Token<JobTokenIdentifier>(identifier, jobTokenSecretManager);
    shuffleToken.setService(identifier.getJobId());
    jobConf.setOutputFormat(SequenceFileOutputFormat.class);
    jobConf.set(MRFrameworkConfigs.TASK_LOCAL_RESOURCE_DIR, new Path(workDir, "localized-resources").toUri().toString());
    jobConf.setBoolean(TezRuntimeConfiguration.TEZ_RUNTIME_OPTIMIZE_LOCAL_FETCH, true);
    FileOutputFormat.setOutputPath(jobConf, new Path(workDir, "output"));
    ProcessorDescriptor reduceProcessorDesc = ProcessorDescriptor.create(ReduceProcessor.class.getName()).setUserPayload(TezUtils.createUserPayloadFromConf(jobConf));
    InputSpec reduceInputSpec = new InputSpec(mapVertexName, InputDescriptor.create(OrderedGroupedInputLegacy.class.getName()).setUserPayload(TezUtils.createUserPayloadFromConf(jobConf)), 1);
    OutputSpec reduceOutputSpec = new OutputSpec("NullDestinationVertex", OutputDescriptor.create(MROutputLegacy.class.getName()).setUserPayload(TezUtils.createUserPayloadFromConf(jobConf)), 1);
    // Now run a reduce
    TaskSpec taskSpec = new TaskSpec(TezTestUtils.getMockTaskAttemptId(0, 1, 0, 0), dagName, reduceVertexName, -1, reduceProcessorDesc, Collections.singletonList(reduceInputSpec), Collections.singletonList(reduceOutputSpec), null, null);
    Map<String, ByteBuffer> serviceConsumerMetadata = new HashMap<String, ByteBuffer>();
    String auxiliaryService = jobConf.get(TezConfiguration.TEZ_AM_SHUFFLE_AUXILIARY_SERVICE_ID, TezConfiguration.TEZ_AM_SHUFFLE_AUXILIARY_SERVICE_ID_DEFAULT);
    serviceConsumerMetadata.put(auxiliaryService, ShuffleUtils.convertJobTokenToBytes(shuffleToken));
    Map<String, String> serviceProviderEnvMap = new HashMap<String, String>();
    ByteBuffer shufflePortBb = ByteBuffer.allocate(4).putInt(0, 8000);
    AuxiliaryServiceHelper.setServiceDataIntoEnv(auxiliaryService, shufflePortBb, serviceProviderEnvMap);
    LogicalIOProcessorRuntimeTask task = new LogicalIOProcessorRuntimeTask(taskSpec, 0, jobConf, new String[] { workDir.toString() }, new TestUmbilical(), serviceConsumerMetadata, serviceProviderEnvMap, HashMultimap.<String, String>create(), null, "", new ExecutionContextImpl("localhost"), Runtime.getRuntime().maxMemory(), true, new DefaultHadoopShim(), sharedExecutor);
    List<Event> destEvents = new LinkedList<Event>();
    destEvents.add(dme);
    task.initialize();
    OrderedGroupedInputLegacy sortedOut = (OrderedGroupedInputLegacy) task.getInputs().values().iterator().next();
    sortedOut.handleEvents(destEvents);
    task.run();
    task.close();
    sharedExecutor.shutdownNow();
    // MRTask mrTask = (MRTask)t.getProcessor();
    // TODO NEWTEZ Verify the partitioner has not been created
    // Likely not applicable anymore.
    // Assert.assertNull(mrTask.getPartitioner());
    // Only a task commit happens, hence the data is still in the temporary directory.
    Path reduceOutputDir = new Path(new Path(workDir, "output"), "_temporary/0/" + IDConverter.toMRTaskIdForOutput(TezTestUtils.getMockTaskId(0, 1, 0)));
    Path reduceOutputFile = new Path(reduceOutputDir, "part-v001-o000-00000");
    SequenceFile.Reader reader = new SequenceFile.Reader(localFs, reduceOutputFile, jobConf);
    LongWritable key = new LongWritable();
    Text value = new Text();
    long prev = Long.MIN_VALUE;
    while (reader.next(key, value)) {
        if (prev != Long.MIN_VALUE) {
            Assert.assertTrue(prev < key.get());
            prev = key.get();
        }
    }
    reader.close();
}