[ https://issues.apache.org/activemq/browse/AMQ-1235?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hiram Chirino resolved AMQ-1235.
--------------------------------
Resolution: Fixed
Fix Version/s: 4.2.0
4.1.2
Good patch! Thanks for figuring this out. Applied.
> Scheduler.cancel uses incorrect argument to shutdown threads
> ------------------------------------------------------------
>
> Key: AMQ-1235
> URL: https://issues.apache.org/activemq/browse/AMQ-1235
> Project: ActiveMQ
> Issue Type: Bug
> Affects Versions: 4.1.1
> Reporter: Alex Burgel
> Assigned To: Hiram Chirino
> Priority: Critical
> Fix For: 4.1.2, 4.2.0
>
> Attachments: scheduler.patch
>
>
> looking at the code from 4.1.1 in org.apache.activemq.thread.Scheduler, in the cancel
method:
> the Runnable task argument is passed to clockDaemon.remove(). i think this is incorrect.
ScheduledFuture ticket should be passed to clockDaemon.remove().
> the javadocs of ScheduledThreadPoolExecutor.remove discuss the possibility that Runnables
might be stored in some other form internally, so calling remove with a plain Runnable might
not do anything. I think the solution is to call remove with a ScheduledFuture, which is how
they are stored internally in ScheduledThreadPoolExecutor.
> i came across this bug after upgrading to the java 5 version of backport-util-concurrent
3.0. that version makes more assumptions about the types that are passed into ScheduledThreadPoolExecutor.remove,
so when you pass in a regular Runnable you'll get a ClassCastException.
> this is trivial to fix, so i don't think a patch is necessary. also i think this might
address the memory leak mentioned in AMQ-1205
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
|