[
https://issues.apache.org/jira/browse/ACCUMULO-1858?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13824040#comment-13824040
]
Jared Winick commented on ACCUMULO-1858:
----------------------------------------
No, the NoClassDefFoundError doesn't seem to cause a problem besides the log
message. I definitely agree the sleep should be the responsibility of the user
and not the ZooKeeperInstance.
I think this would be a reasonable Zookeeper ticket which I am happy to write
up. Without intimate knowledge of their code, it seems like a sendThread.join()
and eventThread.join() after line 1304
http://svn.apache.org/viewvc/zookeeper/tags/release-3.3.6/src/java/main/org/apache/zookeeper/ClientCnxn.java?revision=1368082&view=markup
would do the trick, right? Either way, the fix I think we would want is to
make the ZooKeeper.close() (and internally the ClientCnxn.close()) block until
the threads have stopped.
BTW, someone brought this up on the ZK mailing list this summer but didn't get
a response
http://mail-archives.apache.org/mod_mbox/zookeeper-user/201306.mbox/%[email protected]%3E
If you guys think this sounds right, I can write up a ZooKeeper issue
referencing the testing done here. I will also verify the behavior on v3.4.5.
> Backport fix for Accumulo-1379 PermGen Leak to 1.4 and 1.5
> ----------------------------------------------------------
>
> Key: ACCUMULO-1858
> URL: https://issues.apache.org/jira/browse/ACCUMULO-1858
> Project: Accumulo
> Issue Type: Bug
> Affects Versions: 1.4.3, 1.5.0
> Reporter: Aaron Cordova
> Assignee: Sean Busbey
> Priority: Minor
> Fix For: 1.4.5, 1.5.1
>
> Attachments: ACCUMULO-1858-no-patch.png,
> ACCUMULO-1858-patch-close.png, ACCUMULO-1858.1.patch.txt
>
>
> Apply bug fix for the way zookeeper client is handled in Accumulo-1379 to
> Accumulo versions 1.4 and 1.5.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
