Examples with CompactorRequest org.apache.flink.connector.file.sink.compactor.operator.CompactorRequest used on opensource projects

Example 1 with CompactorRequest

use of org.apache.flink.connector.file.sink.compactor.operator.CompactorRequest in project flink by apache.

the class FileSink method addPreCommitTopology.

@Override
public DataStream<CommittableMessage<FileSinkCommittable>> addPreCommitTopology(DataStream<CommittableMessage<FileSinkCommittable>> committableStream) {
    FileCompactStrategy strategy = bucketsBuilder.getCompactStrategy();
    if (strategy == null) {
        // not enabled, handlers will be added to process the remaining states of the compact
        // coordinator and the compactor operators.
        SingleOutputStreamOperator<Either<CommittableMessage<FileSinkCommittable>, CompactorRequest>> coordinatorOp = committableStream.forward().transform("CompactorCoordinator", new EitherTypeInfo<>(committableStream.getType(), new CompactorRequestTypeInfo(bucketsBuilder::getCommittableSerializer)), new CompactCoordinatorStateHandlerFactory(bucketsBuilder::getCommittableSerializer)).setParallelism(committableStream.getParallelism()).uid("FileSinkCompactorCoordinator");
        return coordinatorOp.forward().transform("CompactorOperator", committableStream.getType(), new CompactorOperatorStateHandlerFactory(bucketsBuilder::getCommittableSerializer, bucketsBuilder::createBucketWriter)).setParallelism(committableStream.getParallelism()).uid("FileSinkCompactorOperator");
    }
    // explicitly rebalance here is required, or the partitioner will be forward, which is in
    // fact the partitioner from the writers to the committers
    SingleOutputStreamOperator<CompactorRequest> coordinatorOp = committableStream.rebalance().transform("CompactorCoordinator", new CompactorRequestTypeInfo(bucketsBuilder::getCommittableSerializer), new CompactCoordinatorFactory(strategy, bucketsBuilder::getCommittableSerializer)).setParallelism(1).uid("FileSinkCompactorCoordinator");
    // parallelism of the compactors is not configurable at present, since it must be identical
    // to that of the committers, or the committable summary and the committables may be
    // distributed to different committers, which will cause a failure
    TypeInformation<CommittableMessage<FileSinkCommittable>> committableType = committableStream.getType();
    return coordinatorOp.transform("CompactorOperator", committableType, new CompactorOperatorFactory(strategy, bucketsBuilder.getFileCompactor(), bucketsBuilder::getCommittableSerializer, bucketsBuilder::createBucketWriter)).setParallelism(committableStream.getParallelism()).uid("FileSinkCompactorOperator");
}

Example 2 with CompactorRequest

use of org.apache.flink.connector.file.sink.compactor.operator.CompactorRequest in project flink by apache.

the class CompactCoordinatorTest method testSizeThreshold.

@Test
public void testSizeThreshold() throws Exception {
    FileCompactStrategy strategy = Builder.newBuilder().setSizeThreshold(10).build();
    CompactCoordinator coordinator = new CompactCoordinator(strategy, getTestCommittableSerializer());
    try (OneInputStreamOperatorTestHarness<CommittableMessage<FileSinkCommittable>, CompactorRequest> harness = new OneInputStreamOperatorTestHarness<>(coordinator)) {
        harness.setup();
        harness.open();
        FileSinkCommittable committable0 = committable("0", ".0", 5);
        FileSinkCommittable committable1 = committable("0", ".1", 6);
        harness.processElement(message(committable0));
        Assert.assertEquals(0, harness.extractOutputValues().size());
        harness.processElement(message(committable1));
        List<CompactorRequest> results = harness.extractOutputValues();
        Assert.assertEquals(1, results.size());
        assertToCompact(results.get(0), committable0, committable1);
        harness.processElement(message(committable("0", ".2", 5)));
        harness.processElement(message(committable("1", ".0", 5)));
        Assert.assertEquals(1, harness.extractOutputValues().size());
    }
}

Example 3 with CompactorRequest

use of org.apache.flink.connector.file.sink.compactor.operator.CompactorRequest in project flink by apache.

the class CompactCoordinatorTest method testCompactOverMultipleCheckpoints.

@Test
public void testCompactOverMultipleCheckpoints() throws Exception {
    FileCompactStrategy strategy = Builder.newBuilder().enableCompactionOnCheckpoint(3).build();
    CompactCoordinator coordinator = new CompactCoordinator(strategy, getTestCommittableSerializer());
    try (OneInputStreamOperatorTestHarness<CommittableMessage<FileSinkCommittable>, CompactorRequest> harness = new OneInputStreamOperatorTestHarness<>(coordinator)) {
        harness.setup();
        harness.open();
        FileSinkCommittable committable0 = committable("0", ".0", 5);
        FileSinkCommittable committable1 = committable("0", ".1", 6);
        harness.processElement(message(committable0));
        harness.processElement(message(committable1));
        Assert.assertEquals(0, harness.extractOutputValues().size());
        harness.prepareSnapshotPreBarrier(1);
        harness.snapshot(1, 1);
        harness.prepareSnapshotPreBarrier(2);
        harness.snapshot(2, 2);
        Assert.assertEquals(0, harness.extractOutputValues().size());
        harness.prepareSnapshotPreBarrier(3);
        harness.snapshot(3, 3);
        List<CompactorRequest> results = harness.extractOutputValues();
        Assert.assertEquals(1, results.size());
        assertToCompact(results.get(0), committable0, committable1);
    }
}

Example 4 with CompactorRequest

use of org.apache.flink.connector.file.sink.compactor.operator.CompactorRequest in project flink by apache.

the class CompactCoordinatorTest method testCompactOnEndOfInput.

@Test
public void testCompactOnEndOfInput() throws Exception {
    FileCompactStrategy strategy = Builder.newBuilder().setSizeThreshold(10).build();
    CompactCoordinator coordinator = new CompactCoordinator(strategy, getTestCommittableSerializer());
    try (OneInputStreamOperatorTestHarness<CommittableMessage<FileSinkCommittable>, CompactorRequest> harness = new OneInputStreamOperatorTestHarness<>(coordinator)) {
        harness.setup();
        harness.open();
        FileSinkCommittable committable0 = committable("0", ".0", 5);
        harness.processElement(message(committable0));
        Assert.assertEquals(0, harness.extractOutputValues().size());
        harness.prepareSnapshotPreBarrier(1);
        harness.snapshot(1, 1);
        Assert.assertEquals(0, harness.extractOutputValues().size());
        harness.endInput();
        List<CompactorRequest> results = harness.extractOutputValues();
        Assert.assertEquals(1, results.size());
        assertToCompact(results.get(0), committable0);
    }
}

Example 5 with CompactorRequest

use of org.apache.flink.connector.file.sink.compactor.operator.CompactorRequest in project flink by apache.

the class CompactCoordinatorTest method testStateHandler.

@Test
public void testStateHandler() throws Exception {
    FileCompactStrategy strategy = Builder.newBuilder().setSizeThreshold(10).build();
    CompactCoordinator coordinator = new CompactCoordinator(strategy, getTestCommittableSerializer());
    // with . prefix
    FileSinkCommittable committable0 = committable("0", ".0", 5);
    FileSinkCommittable committable1 = committable("0", ".1", 6);
    // without . prefix
    FileSinkCommittable committable2 = committable("0", "2", 6);
    OperatorSubtaskState state;
    try (OneInputStreamOperatorTestHarness<CommittableMessage<FileSinkCommittable>, CompactorRequest> harness = new OneInputStreamOperatorTestHarness<>(coordinator)) {
        harness.setup();
        harness.open();
        harness.processElement(message(committable0));
        Assert.assertEquals(0, harness.extractOutputValues().size());
        harness.prepareSnapshotPreBarrier(1);
        state = harness.snapshot(1, 1);
    }
    CompactCoordinatorStateHandler handler = new CompactCoordinatorStateHandler(getTestCommittableSerializer());
    try (OneInputStreamOperatorTestHarness<CommittableMessage<FileSinkCommittable>, Either<CommittableMessage<FileSinkCommittable>, CompactorRequest>> harness = new OneInputStreamOperatorTestHarness<>(handler)) {
        harness.setup(new EitherSerializer<>(new SimpleVersionedSerializerTypeSerializerProxy<>(() -> new CommittableMessageSerializer<>(getTestCommittableSerializer())), new SimpleVersionedSerializerTypeSerializerProxy<>(() -> new CompactorRequestSerializer(getTestCommittableSerializer()))));
        harness.initializeState(state);
        harness.open();
        Assert.assertEquals(1, harness.extractOutputValues().size());
        harness.processElement(message(committable1));
        harness.processElement(message(committable2));
        List<Either<CommittableMessage<FileSinkCommittable>, CompactorRequest>> results = harness.extractOutputValues();
        Assert.assertEquals(3, results.size());
        // restored request
        Assert.assertTrue(results.get(0).isRight());
        assertToCompact(results.get(0).right(), committable0);
        // committable with . prefix should also be passed through
        Assert.assertTrue(results.get(1).isLeft() && results.get(1).left() instanceof CommittableWithLineage);
        Assert.assertEquals(((CommittableWithLineage<FileSinkCommittable>) results.get(1).left()).getCommittable(), committable1);
        // committable without . prefix should be passed through normally
        Assert.assertTrue(results.get(2).isLeft() && results.get(2).left() instanceof CommittableWithLineage);
        Assert.assertEquals(((CommittableWithLineage<FileSinkCommittable>) results.get(2).left()).getCommittable(), committable2);
    }
}