[
https://issues.apache.org/jira/browse/FLINK-10042?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16586302#comment-16586302
]
ASF GitHub Bot commented on FLINK-10042:
----------------------------------------
azagrebin commented on a change in pull request #6556: FLINK-10042][state]
Extract snapshot algorithms from inner classes of RocksDBKeyedStateBackend into
full classes
URL: https://github.com/apache/flink/pull/6556#discussion_r211193248
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/state/AsyncSnapshotCallable.java
##########
@@ -0,0 +1,190 @@
+/*
+ * 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.state;
+
+import org.apache.flink.annotation.VisibleForTesting;
+import org.apache.flink.core.fs.CloseableRegistry;
+
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import javax.annotation.Nonnull;
+import javax.annotation.Nullable;
+
+import java.io.Closeable;
+import java.io.IOException;
+import java.util.concurrent.Callable;
+import java.util.concurrent.CancellationException;
+import java.util.concurrent.FutureTask;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+/**
+ * Base class that outlines the strategy for asynchronous snapshots.
Implementations of this class are typically
+ * instantiated with resources that have been created in the synchronous part
of a snapshot. Then, the implementation
+ * of {@link #callInternal()} is invoked in the asynchronous part. All
resources created by this methods should
+ * be released by the end of the method. If the created resources are {@link
Closeable} objects and can block in calls
+ * (e.g. in/output streams), they should be registered with the snapshot's
{@link CloseableRegistry} so that the can
+ * be closed and unblocked on cancellation. After {@link #callInternal()}
ended, {@link #logAsyncSnapshotComplete(long)}
+ * is called. In that method, implementations can emit log statements about
the duration. At the very end, this class
+ * calls {@link #cleanupProvidedResources()}. The implementation of this
method should release all provided resources
+ * that have been passed into the snapshot from the synchronous part of the
snapshot.
+ *
+ * @param <T> type of the result.
+ */
+public abstract class AsyncSnapshotCallable<T> implements Callable<T> {
+
+ /** Message for the {@link CancellationException}. */
+ private static final String CANCELLATION_EXCEPTION_MSG = "Async
snapshot was cancelled.";
+
+ private static final Logger LOG =
LoggerFactory.getLogger(AsyncSnapshotCallable.class);
+
+ /** This is used to atomically claim ownership for the resource
cleanup. */
+ @Nonnull
+ private final AtomicBoolean resourceCleanupOwnershipTaken;
+
+ /** Registers streams that can block in I/O during snapshot. Forwards
close from taskCancelCloseableRegistry. */
+ @Nonnull
+ private final CloseableRegistry snapshotCloseableRegistry;
+
+ protected AsyncSnapshotCallable() {
+ this.snapshotCloseableRegistry = new CloseableRegistry();
+ this.resourceCleanupOwnershipTaken = new AtomicBoolean(false);
+ }
+
+ @Override
+ public T call() throws Exception {
+ final long startTime = System.currentTimeMillis();
+
+ if (resourceCleanupOwnershipTaken.compareAndSet(false, true)) {
+ try {
+ T result = callInternal();
+ logAsyncSnapshotComplete(startTime);
+ return result;
+ } catch (Exception ex) {
+ if (!snapshotCloseableRegistry.isClosed()) {
+ throw ex;
+ }
+ } finally {
+ closeSnapshotIO();
+ cleanup();
+ }
+ }
+
+ throw new CancellationException(CANCELLATION_EXCEPTION_MSG);
+ }
+
+ @VisibleForTesting
+ protected void cancel() {
+ closeSnapshotIO();
+ if (resourceCleanupOwnershipTaken.compareAndSet(false, true)) {
+ cleanup();
+ }
+ }
+
+ /**
+ * Creates a future task from this and registers it with the given
{@link CloseableRegistry}. The task is
+ * unregistered again in {@link FutureTask#done()}.
+ */
+ public AsyncSnapshotTask toAsyncSnapshotFutureTask(@Nonnull
CloseableRegistry taskRegistry) throws IOException {
+ return new AsyncSnapshotTask(taskRegistry);
+ }
+
+ /**
+ * {@link FutureTask} that wraps a {@link AsyncSnapshotCallable} and
connects it with cancellation and closing.
+ */
+ public class AsyncSnapshotTask extends FutureTask<T> {
Review comment:
`AsyncSnapshotCallable` is always converted to `AsyncSnapshotTask` after its
creation. I wonder if they could be one task class in the first place, unless
there is a good reason and they have really different concerns.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> Extract snapshot algorithms from inner classes into full classes
> ----------------------------------------------------------------
>
> Key: FLINK-10042
> URL: https://issues.apache.org/jira/browse/FLINK-10042
> Project: Flink
> Issue Type: Sub-task
> Reporter: Stefan Richter
> Assignee: Stefan Richter
> Priority: Major
> Labels: pull-request-available
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
