[
https://issues.apache.org/jira/browse/ZOOKEEPER-3564?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16946578#comment-16946578
]
maoling commented on ZOOKEEPER-3564:
------------------------------------
[~denghongcai] Thanks for reporting for this
* What heap-dump tool were you using? how many size and memory occupation of
*waitingEvents?*
* You were creating too many watch events which had hit the bottle neck of zk
server? you can improve you client/application logic to avoid this?
* Do you have any good suggestions? A rate limiter/throttle for the watcher?
> org.apache.zookeeper.ClientCnxn EventThread Memory Problem
> ----------------------------------------------------------
>
> Key: ZOOKEEPER-3564
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3564
> Project: ZooKeeper
> Issue Type: Bug
> Components: java client
> Reporter: Hongcai Deng
> Priority: Critical
> Attachments: image-2019-10-01-10-02-28-228.png,
> image-2019-10-01-10-18-07-156.png
>
>
> recently i found some fullgc occur on my java app. i did heapdump and found
> that
> !image-2019-10-01-10-02-28-228.png!
> EventThread ate too much memory. I dig into zk code, found that
>
> {code:java}
> class EventThread extends ZooKeeperThread {
> private final LinkedBlockingQueue<Object> waitingEvents =
> new LinkedBlockingQueue<Object>();
> // code lines
> }{code}
> waitingEvents not set a boundary. is this for some reason?
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
