rkhachatryan commented on a change in pull request #11515: [FLINK-16744][task] implement channel state persistence for unaligned checkpoints URL: https://github.com/apache/flink/pull/11515#discussion_r405119890
########## File path: flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/channel/ChannelStateWriteRequestExecutorImpl.java ########## @@ -0,0 +1,161 @@ +/* + * 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.runtime.checkpoint.channel; + +import org.apache.flink.annotation.VisibleForTesting; +import org.apache.flink.util.ExceptionUtils; +import org.apache.flink.util.function.RunnableWithException; + +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import javax.annotation.concurrent.ThreadSafe; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.List; +import java.util.concurrent.BlockingDeque; +import java.util.concurrent.CancellationException; +import java.util.concurrent.LinkedBlockingDeque; + +/** + * Executes {@link ChannelStateWriteRequest}s in a separate thread. Any exception occurred during execution causes this + * thread to stop and the exception to be re-thrown on any subsequent call. + */ +@ThreadSafe +class ChannelStateWriteRequestExecutorImpl implements ChannelStateWriteRequestExecutor { + + private static final Logger LOG = LoggerFactory.getLogger(ChannelStateWriteRequestExecutorImpl.class); + private static final int DEFAULT_HANDOVER_CAPACITY = 10_000; + + private final ChannelStateWriteRequestDispatcher dispatcher; + private final BlockingDeque<ChannelStateWriteRequest> deque; + private final Thread thread; + private volatile Exception thrown = null; + private volatile boolean wasClosed = false; + + ChannelStateWriteRequestExecutorImpl(ChannelStateWriteRequestDispatcher dispatcher) { + this(dispatcher, new LinkedBlockingDeque<>(DEFAULT_HANDOVER_CAPACITY)); Review comment: If the same buffer can be enqueued multiple times, then this overhead can be significant. It can happen because of a bug or (probably) with multiple checkpoints in-flight but without incremental checkpointing. In such cases we can: - throw an exception (current behavior) - block - allow to proceed (possibly exhausting memory / gc) (and no limit means `Integer.MAX_VALUE`, so it's just a higher limit) ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
