[jboss-jira] [JBoss JIRA] (JGRP-1888) FD_SOCK race condition during stop

[Dennis Reed (JIRA)]

    [ https://issues.jboss.org/browse/JGRP-1888?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011387#comment-13011387 ] 

Dennis Reed commented on JGRP-1888:
-----------------------------------

Does not appear to affect master, as the order was switched (pinger_thread stopped first), and it waits for the pinger_thread to stop before stopping the BroadcastTask.


> FD_SOCK race condition during stop
> ----------------------------------
>
>                 Key: JGRP-1888
>                 URL: https://issues.jboss.org/browse/JGRP-1888
>             Project: JGroups
>          Issue Type: Bug
>    Affects Versions: 2.6.16
>            Reporter: Dennis Reed
>            Assignee: Dennis Reed
>
> There is a race condition in FD_SOCK between stop() shutting down the pinger thread and the BroadcastTask.
> It first shuts down the broadcast task, and then the pinger thread.
> If the pinger thread generates a new suspected event in this interval, it re-starts the broadcast task, which creates a new thread suspecting that node repeatedly until the JVM exits.



--
This message was sent by Atlassian JIRA
(v6.3.1#6329)