yunfengzhou-hub commented on a change in pull request #17: URL: https://github.com/apache/flink-ml/pull/17#discussion_r726816428
########## File path: flink-ml-iteration/src/main/java/org/apache/flink/ml/iteration/datacache/nonkeyed/DataCacheSnapshot.java ########## @@ -0,0 +1,201 @@ +/* + * 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.iteration.datacache.nonkeyed; + +import org.apache.flink.api.common.typeutils.TypeSerializer; +import org.apache.flink.api.java.tuple.Tuple2; +import org.apache.flink.core.fs.FSDataInputStream; +import org.apache.flink.core.fs.FSDataOutputStream; +import org.apache.flink.core.fs.FileSystem; +import org.apache.flink.core.fs.Path; +import org.apache.flink.core.memory.DataInputViewStreamWrapper; +import org.apache.flink.runtime.util.NonClosingInputStreamDecorator; +import org.apache.flink.runtime.util.NonClosingOutpusStreamDecorator; +import org.apache.flink.statefun.flink.core.feedback.FeedbackConsumer; +import org.apache.flink.util.IOUtils; +import org.apache.flink.util.function.SupplierWithException; + +import javax.annotation.Nullable; + +import java.io.DataInputStream; +import java.io.DataOutputStream; +import java.io.IOException; +import java.io.InputStream; +import java.io.OutputStream; +import java.util.ArrayList; +import java.util.Collections; +import java.util.List; + +import static org.apache.flink.util.Preconditions.checkState; + +/** The snapshot of a data cache. It could be written out or read from an external stream.O */ +public class DataCacheSnapshot { + + private static final int CURRENT_VERSION = 1; + + private final FileSystem fileSystem; + + @Nullable private final Tuple2<Integer, Integer> readerPosition; + + private final List<Segment> segments; + + public DataCacheSnapshot( + FileSystem fileSystem, + @Nullable Tuple2<Integer, Integer> readerPosition, + List<Segment> segments) { + this.fileSystem = fileSystem; + this.readerPosition = readerPosition; + this.segments = segments; + } + + public FileSystem getFileSystem() { + return fileSystem; + } + + @Nullable + public Tuple2<Integer, Integer> getReaderPosition() { + return readerPosition; + } + + public List<Segment> getSegments() { + return segments; + } + + public void writeTo(OutputStream checkpointOutputStream) throws IOException { + try (DataOutputStream dos = + new DataOutputStream(new NonClosingOutpusStreamDecorator(checkpointOutputStream))) { + dos.writeInt(CURRENT_VERSION); + dos.writeBoolean(readerPosition != null); + if (readerPosition != null) { + dos.writeInt(readerPosition.f0); + dos.writeInt(readerPosition.f1); + } + + dos.writeBoolean(fileSystem.isDistributedFS()); + if (fileSystem.isDistributedFS()) { + // We only need to record the segments itself + dos.writeInt(segments.size()); + for (Segment segment : segments) { + dos.writeUTF(segment.getPath().toString()); + dos.writeInt(segment.getCount()); + } + } else { + // We have to copy the whole streams. + int totalRecords = segments.stream().mapToInt(Segment::getCount).sum(); + checkState(totalRecords >= 0, "overflowed: " + totalRecords); + dos.writeInt(totalRecords); + + for (Segment segment : segments) { + try (FSDataInputStream inputStream = fileSystem.open(segment.getPath())) { + IOUtils.copyBytes(inputStream, checkpointOutputStream, false); + } + } + } + } + } + + public static <T> void replay( + InputStream checkpointInputStream, + TypeSerializer<T> serializer, + FileSystem fileSystem, + FeedbackConsumer<T> feedbackConsumer) + throws Exception { + try (DataInputStream dis = + new DataInputStream(new NonClosingInputStreamDecorator(checkpointInputStream))) { + int version = dis.readInt(); + checkState( + version == CURRENT_VERSION, + "Currently only support version " + CURRENT_VERSION); + parseReaderPosition(dis); + + boolean isDistributedFS = dis.readBoolean(); + if (isDistributedFS) { + List<Segment> segments = parseSegments(dis); + DataCacheReader<T> dataCacheReader = + new DataCacheReader<T>(serializer, fileSystem, segments); + while (dataCacheReader.hasNext()) { + feedbackConsumer.processFeedback(dataCacheReader.next()); + } + } else { + DataInputViewStreamWrapper dataInputView = new DataInputViewStreamWrapper(dis); + int totalRecords = dis.readInt(); + for (int i = 0; i < totalRecords; ++i) { + feedbackConsumer.processFeedback(serializer.deserialize(dataInputView)); + } + } + } + } + + public static DataCacheSnapshot recover( + InputStream checkpointInputStream, + FileSystem fileSystem, + SupplierWithException<Path, IOException> pathGenerator) + throws IOException { + try (DataInputStream dis = + new DataInputStream(new NonClosingInputStreamDecorator(checkpointInputStream))) { + int version = dis.readInt(); + checkState( + version == CURRENT_VERSION, + "Currently only support version " + CURRENT_VERSION); + Tuple2<Integer, Integer> readerPosition = parseReaderPosition(dis); + + boolean isDistributedFS = dis.readBoolean(); + checkState( + isDistributedFS == fileSystem.isDistributedFS(), + "Currently we do not support changing the cache file system. " + + "If required, please manually copy the directory from one filesystem to another."); + + List<Segment> segments; + if (isDistributedFS) { + segments = parseSegments(dis); + } else { + int totalRecords = dis.readInt(); + Path path = pathGenerator.get(); + try (FSDataOutputStream outputStream = + fileSystem.create(path, FileSystem.WriteMode.NO_OVERWRITE)) { + IOUtils.copyBytes(checkpointInputStream, outputStream, false); + } + segments = Collections.singletonList(new Segment(path, totalRecords)); + } + + return new DataCacheSnapshot(fileSystem, readerPosition, segments); + } + } Review comment: `replay()` is used only in `HeadOperator.initializeState()` and `recover` is not used except in tests. Can we just delete `recover()` and rename `replay()` to `recover()`? -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
