[
https://issues.apache.org/activemq/browse/AMQ-2483?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gary Tully resolved AMQ-2483.
-----------------------------
Resolution: Fixed
Fix Version/s: 5.3.1
resolved in r835874
revert queue back to pooled/dedicated task runner where multiple wake up calls
are consolidated. To ensure no call is ignored, the queue tracks pending
wakeupRequests and requests a repeat iteration till all requests are satisfied.
Run of unit tests with -DuseDedicatedTaskRunner=true threw up a few failures,
one of which was: https://issues.apache.org/activemq/browse/AMQ-2028
I exposed useDedicatedTaskRunner on connection factory to allow explicit unit
test verification and propagated broker.useDedicatedTaskRunner to make it more
easily configurable.
> OOM due to build of of iterate executions during bulk message consumption -
> asyncWakeup called excessivly
> ---------------------------------------------------------------------------------------------------------
>
> Key: AMQ-2483
> URL: https://issues.apache.org/activemq/browse/AMQ-2483
> Project: ActiveMQ
> Issue Type: Bug
> Components: Broker
> Affects Versions: 5.3.0
> Reporter: Gary Tully
> Assignee: Gary Tully
> Fix For: 5.3.1
>
>
> when consuming a large of messages,the method:asyncWakeup() is invoked
> crazily,so the executor has a great deal of runnable that callback
> Queue.iterate(), but
> Queue.iterate() is much slower than the increasing of runnable in the
> executor. this result in OOM.
> avoid to invoke the method:asyncWakeup() frequently???
> http://old.nabble.com/file/p26278228/oom.jpg
> Context: http://old.nabble.com/Out-of-Memory-on-5.3-tp26064098p26297020.html
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
