zookeeper-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From lvfangmin <...@git.apache.org>
Subject [GitHub] zookeeper pull request #590: [ZOOKEEPER-1177] Add the memory optimized watch...
Date Sun, 23 Sep 2018 04:57:52 GMT
Github user lvfangmin commented on a diff in the pull request:

    https://github.com/apache/zookeeper/pull/590#discussion_r219688430
  
    --- Diff: src/java/main/org/apache/zookeeper/server/watch/WatcherCleaner.java ---
    @@ -0,0 +1,176 @@
    +/**
    + * 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.zookeeper.server.watch;
    +
    +import java.util.Set;
    +import java.util.HashSet;
    +import java.util.Random;
    +import java.util.concurrent.atomic.AtomicInteger;
    +
    +import org.apache.zookeeper.common.Time;
    +import org.apache.zookeeper.server.RateLogger;
    +import org.apache.zookeeper.server.WorkerService;
    +import org.apache.zookeeper.server.WorkerService.WorkRequest;
    +
    +import org.slf4j.Logger;
    +import org.slf4j.LoggerFactory;
    +
    +/**
    + * Thread used to lazily clean up the closed watcher, it will trigger the
    + * clean up when the dead watchers get certain number or some number of
    + * seconds has elapsed since last clean up.
    + *
    + * Cost of running it:
    + *
    + * - need to go through all the paths even if the watcher may only
    + *   watching a single path
    + * - block in the path BitHashSet when we try to check the dead watcher
    + *   which won't block other stuff
    + */
    +public class WatcherCleaner extends Thread {
    +
    +    private static final Logger LOG = LoggerFactory.getLogger(WatcherCleaner.class);
    +    private final RateLogger RATE_LOGGER = new RateLogger(LOG);
    +
    +    private volatile boolean stopped = false;
    +    private final Object cleanEvent = new Object();
    +    private final Random r = new Random(System.nanoTime());
    +    private final WorkerService cleaners;
    +
    +    private final Set<Integer> deadWatchers;
    +    private final DeadWatcherListener listener;
    +    private final int watcherCleanThreshold;
    +    private final int watcherCleanIntervalInSeconds;
    +    private final int maxInProcessingDeadWatchers;
    +    private final AtomicInteger totalDeadWatchers = new AtomicInteger();
    +
    +    public WatcherCleaner(DeadWatcherListener listener) {
    +        this(listener,
    +            Integer.getInteger("zookeeper.watcherCleanThreshold", 1000),
    +            Integer.getInteger("zookeeper.watcherCleanIntervalInSeconds", 600),
    +            Integer.getInteger("zookeeper.watcherCleanThreadsNum", 2),
    +            Integer.getInteger("zookeeper.maxInProcessingDeadWatchers", -1));
    +    }
    +
    +    public WatcherCleaner(DeadWatcherListener listener,
    +            int watcherCleanThreshold, int watcherCleanIntervalInSeconds,
    +            int watcherCleanThreadsNum, int maxInProcessingDeadWatchers) {
    +        this.listener = listener;
    +        this.watcherCleanThreshold = watcherCleanThreshold;
    +        this.watcherCleanIntervalInSeconds = watcherCleanIntervalInSeconds;
    +        int suggestedMaxInProcessingThreshold =
    +                watcherCleanThreshold * watcherCleanThreadsNum;
    +        if (maxInProcessingDeadWatchers > 0 &&
    +                maxInProcessingDeadWatchers < suggestedMaxInProcessingThreshold) {
    +            maxInProcessingDeadWatchers = suggestedMaxInProcessingThreshold;
    +            LOG.info("The maxInProcessingDeadWatchers config is smaller " +
    +                    "than the suggested one, change it to use {}",
    +                    maxInProcessingDeadWatchers);
    +        }
    +        this.maxInProcessingDeadWatchers = maxInProcessingDeadWatchers;
    +        this.deadWatchers = new HashSet<Integer>();
    +        this.cleaners = new WorkerService("DeadWatcherCleanner",
    +                watcherCleanThreadsNum, false);
    +
    +        LOG.info("watcherCleanThreshold={}, watcherCleanIntervalInSeconds={}" +
    +                ", watcherCleanThreadsNum={}, maxInProcessingDeadWatchers={}",
    +                watcherCleanThreshold, watcherCleanIntervalInSeconds,
    +                watcherCleanThreadsNum, maxInProcessingDeadWatchers);
    +    }
    +
    +    public void addDeadWatcher(int watcherBit) {
    +        // Wait if there are too many watchers waiting to be closed,
    +        // this is will slow down the socket packet processing and
    +        // the adding watches in the ZK pipeline.
    +        while (maxInProcessingDeadWatchers > 0 && !stopped &&
    +                totalDeadWatchers.get() >= maxInProcessingDeadWatchers) {
    +            try {
    +                RATE_LOGGER.rateLimitLog("Waiting for dead watchers cleaning");
    +                synchronized(totalDeadWatchers) {
    +                    totalDeadWatchers.wait(100);
    +                }
    +            } catch (InterruptedException e) {
    +                LOG.info("Got interrupted while waiting for dead watches " +
    +                        "queue size");
    +            }
    +        }
    +        synchronized (this) {
    +            if (deadWatchers.add(watcherBit)) {
    +                totalDeadWatchers.incrementAndGet();
    +                if (deadWatchers.size() >= watcherCleanThreshold) {
    +                    synchronized (cleanEvent) {
    +                        cleanEvent.notifyAll();
    +                    }
    +                }
    +            }
    +        }
    +    }
    +
    +    @Override
    +    public void run() {
    +        while (!stopped) {
    +            synchronized (cleanEvent) {
    +                try {
    +                    if (deadWatchers.size() < watcherCleanThreshold) {
    +                        int maxWaitMs = (watcherCleanIntervalInSeconds +
    +                            r.nextInt(watcherCleanIntervalInSeconds / 2 + 1)) * 1000;
    +                        cleanEvent.wait(maxWaitMs);
    +                    }
    +                } catch (InterruptedException e) {
    +                    LOG.info("Received InterruptedException while " +
    +                            "waiting for cleanEvent");
    +                    break;
    +                }
    +            }
    +
    +            if (deadWatchers.isEmpty()) {
    +                continue;
    +            }
    +
    +            synchronized (this) {
    +                // Snapshot of the current dead watchers
    +                final Set<Integer> snapshot = new HashSet<Integer>(deadWatchers);
    +                deadWatchers.clear();
    --- End diff --
    
    Clean the dead watchers need to go through all the current watches, which is pretty heavy
and may take a second if there are millions of watches, that's why we're doing lazily batch
clean up here, we don't want to block addDeadWatcher, which is called from the Cnxn.close
while we're doing the clean work.
    
    The totalDeadWatchers is used to avoid OOM when the watcher cleaner cannot catch up (we
haven't seen this problem even with heavy reconnecting scenario), it is suggested to be set
to something like 1000 * watcherCleanThreshold. The watcherCleanThreshold is used to control
the batch size when doing the clean up, there is trade off between GC the dead watcher memory
and the time complexity of cleaning up, so we cannot set this too large. 


---

Mime
View raw message