You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "lindong28 (via GitHub)" <gi...@apache.org> on 2023/04/12 00:34:00 UTC
[GitHub] [flink-ml] lindong28 commented on a diff in pull request #230: [FLINK-31753] Support DataStream CoGroup in stream mode with similar performance as DataSet CoGroup
lindong28 commented on code in PR #230:
URL: https://github.com/apache/flink-ml/pull/230#discussion_r1163449957
##########
flink-ml-core/src/main/java/org/apache/flink/ml/common/datastream/sort/CoGroupOperator.java:
##########
@@ -0,0 +1,313 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.flink.ml.common.datastream.sort;
+
+import org.apache.flink.api.common.ExecutionConfig;
+import org.apache.flink.api.common.functions.CoGroupFunction;
+import org.apache.flink.api.common.typeutils.TypeComparator;
+import org.apache.flink.api.common.typeutils.TypePairComparator;
+import org.apache.flink.api.common.typeutils.TypeSerializer;
+import org.apache.flink.api.java.functions.KeySelector;
+import org.apache.flink.api.java.tuple.Tuple2;
+import org.apache.flink.api.java.typeutils.runtime.RuntimePairComparatorFactory;
+import org.apache.flink.configuration.AlgorithmOptions;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.core.memory.DataOutputSerializer;
+import org.apache.flink.core.memory.ManagedMemoryUseCase;
+import org.apache.flink.runtime.io.disk.iomanager.IOManager;
+import org.apache.flink.runtime.memory.MemoryAllocationException;
+import org.apache.flink.runtime.memory.MemoryManager;
+import org.apache.flink.runtime.operators.sort.ExternalSorter;
+import org.apache.flink.runtime.operators.sort.NonReusingSortMergeCoGroupIterator;
+import org.apache.flink.runtime.operators.sort.PushSorter;
+import org.apache.flink.runtime.operators.sort.ReusingSortMergeCoGroupIterator;
+import org.apache.flink.runtime.operators.util.CoGroupTaskIterator;
+import org.apache.flink.streaming.api.graph.StreamConfig;
+import org.apache.flink.streaming.api.operators.AbstractUdfStreamOperator;
+import org.apache.flink.streaming.api.operators.BoundedMultiInput;
+import org.apache.flink.streaming.api.operators.Output;
+import org.apache.flink.streaming.api.operators.TimestampedCollector;
+import org.apache.flink.streaming.api.operators.TwoInputStreamOperator;
+import org.apache.flink.streaming.api.watermark.Watermark;
+import org.apache.flink.streaming.runtime.streamrecord.StreamRecord;
+import org.apache.flink.streaming.runtime.tasks.StreamTask;
+import org.apache.flink.util.MutableObjectIterator;
+import org.apache.flink.util.TraversableOnceException;
+
+import java.io.Serializable;
+import java.util.Iterator;
+
+/**
+ * An operator that implements the co-group logic.
+ *
+ * @param <IN1> The class type of the first input.
+ * @param <IN2> The class type of the second input.
+ * @param <KEY> The class type of the key.
+ * @param <OUT> The class type of the output values.
+ */
+public class CoGroupOperator<IN1, IN2, KEY extends Serializable, OUT>
+ extends AbstractUdfStreamOperator<OUT, CoGroupFunction<IN1, IN2, OUT>>
+ implements TwoInputStreamOperator<IN1, IN2, OUT>, BoundedMultiInput {
+
+ private PushSorter<Tuple2<byte[], StreamRecord<IN1>>> sorterA;
+ private PushSorter<Tuple2<byte[], StreamRecord<IN2>>> sorterB;
+ private TypeComparator<Tuple2<byte[], StreamRecord<IN1>>> comparatorA;
+ private TypeComparator<Tuple2<byte[], StreamRecord<IN2>>> comparatorB;
+ private KeySelector<IN1, KEY> keySelectorA;
+ private KeySelector<IN2, KEY> keySelectorB;
+ private TypeSerializer<Tuple2<byte[], StreamRecord<IN1>>> keyAndValueSerializerA;
+ private TypeSerializer<Tuple2<byte[], StreamRecord<IN2>>> keyAndValueSerializerB;
+ private TypeSerializer<KEY> keySerializer;
+ private DataOutputSerializer dataOutputSerializer;
+ private long lastWatermarkTimestamp = Long.MIN_VALUE;
+ private int remainingInputNum = 2;
+
+ public CoGroupOperator(CoGroupFunction<IN1, IN2, OUT> function) {
+ super(function);
+ }
+
+ @Override
+ public void setup(
+ StreamTask<?, ?> containingTask,
+ StreamConfig config,
+ Output<StreamRecord<OUT>> output) {
+ super.setup(containingTask, config, output);
+ ClassLoader userCodeClassLoader = containingTask.getUserCodeClassLoader();
+ MemoryManager memoryManager = containingTask.getEnvironment().getMemoryManager();
+ IOManager ioManager = containingTask.getEnvironment().getIOManager();
+
+ keySelectorA = config.getStatePartitioner(0, userCodeClassLoader);
+ keySelectorB = config.getStatePartitioner(1, userCodeClassLoader);
+ keySerializer = config.getStateKeySerializer(userCodeClassLoader);
+ int keyLength = keySerializer.getLength();
+
+ TypeSerializer<IN1> typeSerializerA = config.getTypeSerializerIn(0, userCodeClassLoader);
+ TypeSerializer<IN2> typeSerializerB = config.getTypeSerializerIn(1, userCodeClassLoader);
+ keyAndValueSerializerA = new KeyAndValueSerializer<>(typeSerializerA, keyLength);
+ keyAndValueSerializerB = new KeyAndValueSerializer<>(typeSerializerB, keyLength);
+
+ if (keyLength > 0) {
+ dataOutputSerializer = new DataOutputSerializer(keyLength);
+ comparatorA = new FixedLengthByteKeyComparator<>(keyLength);
+ comparatorB = new FixedLengthByteKeyComparator<>(keyLength);
+ } else {
+ dataOutputSerializer = new DataOutputSerializer(64);
+ comparatorA = new VariableLengthByteKeyComparator<>();
+ comparatorB = new VariableLengthByteKeyComparator<>();
+ }
+
+ ExecutionConfig executionConfig = containingTask.getEnvironment().getExecutionConfig();
+ double managedMemoryFraction =
+ config.getManagedMemoryFractionOperatorUseCaseOfSlot(
+ ManagedMemoryUseCase.OPERATOR,
+ containingTask.getEnvironment().getTaskConfiguration(),
+ userCodeClassLoader)
+ / 2;
+ Configuration jobConfiguration = containingTask.getEnvironment().getJobConfiguration();
+
+ try {
+ sorterA =
+ ExternalSorter.newBuilder(
+ memoryManager,
+ containingTask,
+ keyAndValueSerializerA,
+ comparatorA,
+ executionConfig)
+ .memoryFraction(managedMemoryFraction)
+ .enableSpilling(
+ ioManager,
+ jobConfiguration.get(AlgorithmOptions.SORT_SPILLING_THRESHOLD))
+ .maxNumFileHandles(
+ jobConfiguration.get(AlgorithmOptions.SPILLING_MAX_FAN))
+ .objectReuse(executionConfig.isObjectReuseEnabled())
+ .largeRecords(
+ jobConfiguration.get(
+ AlgorithmOptions.USE_LARGE_RECORDS_HANDLER))
+ .build();
+ sorterB =
+ ExternalSorter.newBuilder(
+ memoryManager,
+ containingTask,
+ keyAndValueSerializerB,
+ comparatorB,
+ executionConfig)
+ .memoryFraction(managedMemoryFraction)
+ .enableSpilling(
+ ioManager,
+ jobConfiguration.get(AlgorithmOptions.SORT_SPILLING_THRESHOLD))
+ .maxNumFileHandles(
+ jobConfiguration.get(AlgorithmOptions.SPILLING_MAX_FAN))
+ .objectReuse(executionConfig.isObjectReuseEnabled())
+ .largeRecords(
+ jobConfiguration.get(
+ AlgorithmOptions.USE_LARGE_RECORDS_HANDLER))
+ .build();
+ } catch (MemoryAllocationException e) {
+ throw new RuntimeException(e);
+ }
+ }
+
+ @Override
+ public void endInput(int inputId) throws Exception {
+ if (inputId == 1) {
+ sorterA.finishReading();
+ remainingInputNum--;
+ } else if (inputId == 2) {
+ sorterB.finishReading();
+ remainingInputNum--;
+ } else {
+ throw new RuntimeException("Unknown inputId " + inputId);
+ }
+
+ if (remainingInputNum > 0) {
+ return;
+ }
+
+ MutableObjectIterator<Tuple2<byte[], StreamRecord<IN1>>> iteratorA = sorterA.getIterator();
+ MutableObjectIterator<Tuple2<byte[], StreamRecord<IN2>>> iteratorB = sorterB.getIterator();
+ TypePairComparator<Tuple2<byte[], StreamRecord<IN1>>, Tuple2<byte[], StreamRecord<IN2>>>
+ pairComparator =
+ (new RuntimePairComparatorFactory<
+ Tuple2<byte[], StreamRecord<IN1>>,
+ Tuple2<byte[], StreamRecord<IN2>>>())
+ .createComparator12(comparatorA, comparatorB);
+
+ CoGroupTaskIterator<Tuple2<byte[], StreamRecord<IN1>>, Tuple2<byte[], StreamRecord<IN2>>>
+ coGroupIterator;
+ if (getExecutionConfig().isObjectReuseEnabled()) {
+ coGroupIterator =
+ new ReusingSortMergeCoGroupIterator<
+ Tuple2<byte[], StreamRecord<IN1>>, Tuple2<byte[], StreamRecord<IN2>>>(
+ iteratorA,
+ iteratorB,
+ keyAndValueSerializerA,
+ comparatorA,
+ keyAndValueSerializerB,
+ comparatorB,
+ pairComparator);
+ } else {
+ coGroupIterator =
+ new NonReusingSortMergeCoGroupIterator<
+ Tuple2<byte[], StreamRecord<IN1>>, Tuple2<byte[], StreamRecord<IN2>>>(
+ iteratorA,
+ iteratorB,
+ keyAndValueSerializerA,
+ comparatorA,
+ keyAndValueSerializerB,
+ comparatorB,
+ pairComparator);
+ }
+
+ coGroupIterator.open();
+ TupleUnwrappingIterator<IN1, byte[]> unWrappediteratorA = new TupleUnwrappingIterator<>();
+ TupleUnwrappingIterator<IN2, byte[]> unWrappediteratorB = new TupleUnwrappingIterator<>();
+ while (coGroupIterator.next()) {
+ unWrappediteratorA.set(coGroupIterator.getValues1().iterator());
+ unWrappediteratorB.set(coGroupIterator.getValues2().iterator());
+ userFunction.coGroup(
+ unWrappediteratorA, unWrappediteratorB, new TimestampedCollector<>(output));
+ }
+ coGroupIterator.close();
+
+ Watermark watermark = new Watermark(lastWatermarkTimestamp);
+ if (getTimeServiceManager().isPresent()) {
+ getTimeServiceManager().get().advanceWatermark(watermark);
+ }
+ output.emitWatermark(watermark);
+ }
+
+ @Override
+ public void processWatermark(Watermark watermark) throws Exception {
Review Comment:
According to the `BoundedMultiInput#endInput` API, it is called when `no more data will arrive from the input identified`. It does not necessarily means the event time is MAX_LONG.
While the current Flink implementation typically use watermark=`MAX_LONG` to indicate end of stream, I am not sure it will always be the case in the future. So it seems more extensible not to have this operator assume the watermark == MAX_LONg when end-of-input is reached.
This operator extends `AbstractStreamOperator`. And `AbstractStreamOperator#processWatermark1` will invoke `processWatermark `. So it seem simpler to implement `processWatermark()` here?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org