Patrick Hunt commented on ZOOKEEPER-124:

This is failing on my machine as well. 

1) I think this is related to ZOOKEEPER-137 - there is a flaw in the watcher 
handling if the same watcher object is registered on the same node for data 
watches while async operations are performed on the node

2) the test as modified by the current attached patch is flawed. when bang is 
set to false all of the hammer threads will immediately close their zk clients 
and exit the run method. pending callbacks are dropped and the counters will 
not be updated properly. to make the test valid we'd have to have the hammer 
threads wait (for some bounded amt of time) while they still had outstanding 
requests. The testHammer() driver method would also have to remove the added 
sleep(5000) because I don't think this is enough time for all of the hammers to 
exit cleanly (before interrupting them).

> StatCallback is broken in trunk
> -------------------------------
>                 Key: ZOOKEEPER-124
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-124
>             Project: Zookeeper
>          Issue Type: Bug
>          Components: java client, server, tests
>         Environment: Linux corner-cube 2.6.24-19-generic #1 SMP Fri Jul 11 
> 23:41:49 UTC 2008 i686 GNU/Linux
> java version "1.6.0_06"
> Java(TM) SE Runtime Environment (build 1.6.0_06-b02)
> Java HotSpot(TM) Client VM (build 10.0-b22, mixed mode, sharing)
>            Reporter: Stu Hood
>            Assignee: Patrick Hunt
>             Fix For: 3.0.0
>         Attachments: stat-callback-fail-test.diff, stat-callback-test.diff
> StatCallback appears to be broken in trunk. I'll attach a patch for AsyncTest 
> that triggers the behaviour.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

Reply via email to