You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@qpid.apache.org by "ASF subversion and git services (JIRA)" <ji...@apache.org> on 2016/12/02 14:17:59 UTC

[jira] [Commented] (DISPATCH-582) Memory Leak with reactor client

    [ https://issues.apache.org/jira/browse/DISPATCH-582?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15715241#comment-15715241 ] 

ASF subversion and git services commented on DISPATCH-582:
----------------------------------------------------------

Commit 0e3d1431fcdb5608428ec87e7c7b7687c6ec09ca in qpid-dispatch's branch refs/heads/master from [~tross@redhat.com]
[ https://git-wip-us.apache.org/repos/asf?p=qpid-dispatch.git;h=0e3d143 ]

DISPATCH-582 - Fixed leak of qdr_delivery_t objects when connections are lost.
There were three problems that needed to be solved:
  - In qdr_link_cleanup, cross-linkage of unsettled and undelivered deliveries were not
    accounted for.
  - With connection loss, linkage from abandoned pn_delivery_t to qdr_delivery_t was not
    accounted for.
  - The delivery refcount on forwarded deliveries was not properly initialized (this was only
    a problem if atomics fell back to the default lock-protected-integers).


> Memory Leak with reactor client
> -------------------------------
>
>                 Key: DISPATCH-582
>                 URL: https://issues.apache.org/jira/browse/DISPATCH-582
>             Project: Qpid Dispatch
>          Issue Type: Bug
>          Components: Routing Engine
>    Affects Versions: 0.7.0
>         Environment: centos 7.0, java 8, proton-c 0.16.0-snapshot
>            Reporter: Rohan Mars
>            Assignee: Ted Ross
>             Fix For: 0.8.0
>
>         Attachments: test-app-reactor-leak.tar.gz, valgrind.log.gz
>
>
> Running the attached client results in memory leaks in the dispatch router process.
> Differentiating factors include:
> * sender does not send settled
> * receiver, sets ack disposition then settles
> * sender receives a callback
> * test harness will run sender/receiver in loop, running for 10 seconds, killing the process after 10 seconds then restarting them.
> I'll attach a java test case and valgrind output.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@qpid.apache.org
For additional commands, e-mail: dev-help@qpid.apache.org