[
https://issues.apache.org/activemq/browse/AMQ-1083?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_39326
]
James Strachan commented on AMQ-1083:
-------------------------------------
BTW which timestamp of 5.0.0 are you using? we've been doing a resonable amount
of development & fixing lately.
Incidentally are your producer/consumers in separate VMs? Just wanna know which
JVM is running out of RAM (the broker or consumer etc).
> Heap overflow when durable consumer closes
> ------------------------------------------
>
> Key: AMQ-1083
> URL: https://issues.apache.org/activemq/browse/AMQ-1083
> Project: ActiveMQ
> Issue Type: Bug
> Components: Message Store
> Affects Versions: 5.0.0
> Reporter: [EMAIL PROTECTED]
>
> (Not sure if bug is in broker or store, but my best guess is in the store, so
> that is how I have assigned the bug. I have attempted to track down in the
> code where the leak is occurring, but have not yet been successful.)
> This is with the Kaha store and the store durable pending cursor. Messages
> size is 10 kilobytes. Heap size set to 512MB. activemq.xml configured for
> 100 MB. Using example producer and consumer programs.
> 1. Start a consumer for a durable topic and stop it immediately.
> 2. Start the producer for the durable topic (leave it running for remainder
> of test).
> 3. Wait until 50000 messages have been enqueued.
> 4. Restart the consumer from Step #1, have it consume 8000 messages.
> 5. Once the consumer has ended (producer still running), heap usage will rise
> until out of heap (forcing GC does not affect).
> Heap usage was about 80 MB at the time the consumer closed, and rose to > 500
> MB within one minute of the consumer closing.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
