You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by Albert Strasheim <13...@sun.ac.za> on 2007/02/28 11:02:51 UTC
Kaha persistence/UsageManager.waitForSpace and fast producer/slow consumer again
Hello all/Rob
We're still having problems with Kaha persistence when running a fast
producer/slow consumer.
The last time I ran our test, we saw that messages would arrive out of
order, presumably when they were being retrieved from the Kaha store
and sent out (since they were definately being sent in order). Rob,
looks like you checked in a test for this kind of thing in revision
512643? Were you seeing problems of this nature?
Having updated to revision 512643, when I run our test, it runs for
about two seconds before the only remaining ActiveMQ Session Task
thread hangs in UsageManager.waitForSpace.
Rob, could you give a summary of the broker/VM/application parameters
that we should try to tweak to avoid this problem?
By the way, we're still seeing the exact same issues with the test case
attached to the AMQ-1148 issue. Maybe the same parameters can be
tweaked to make that test pass? In that case, we're trying to
VMPendingSubscriberMessageStoragePolicy instead of having messages
persisted to the Kaha store.
Thanks for any feedback.
Cheers,
Albert
Re: Kaha persistence/UsageManager.waitForSpace and fast producer/slow consumer again
Posted by Rob Davies <ra...@gmail.com>.
btw Albert - the test case attached to AMQ-1148 runs better if there
are separate connections for the publishers and consumers.
cheers,
Rob
On 28 Feb 2007, at 10:02, Albert Strasheim wrote:
> Hello all/Rob
>
> We're still having problems with Kaha persistence when running a fast
> producer/slow consumer.
>
> The last time I ran our test, we saw that messages would arrive out of
> order, presumably when they were being retrieved from the Kaha store
> and sent out (since they were definately being sent in order). Rob,
> looks like you checked in a test for this kind of thing in revision
> 512643? Were you seeing problems of this nature?
>
> Having updated to revision 512643, when I run our test, it runs for
> about two seconds before the only remaining ActiveMQ Session Task
> thread hangs in UsageManager.waitForSpace.
>
> Rob, could you give a summary of the broker/VM/application parameters
> that we should try to tweak to avoid this problem?
>
> By the way, we're still seeing the exact same issues with the test
> case
> attached to the AMQ-1148 issue. Maybe the same parameters can be
> tweaked to make that test pass? In that case, we're trying to
> VMPendingSubscriberMessageStoragePolicy instead of having messages
> persisted to the Kaha store.
>
> Thanks for any feedback.
>
> Cheers,
>
> Albert
Re: Kaha persistence/UsageManager.waitForSpace and fast producer/slow consumer again
Posted by Rob Davies <ra...@gmail.com>.
Hi albert,
I'm just going to start using your test case and see if I can figure
out whats going on
cheers,
Rob
On 28 Feb 2007, at 10:02, Albert Strasheim wrote:
> Hello all/Rob
>
> We're still having problems with Kaha persistence when running a fast
> producer/slow consumer.
>
> The last time I ran our test, we saw that messages would arrive out of
> order, presumably when they were being retrieved from the Kaha store
> and sent out (since they were definately being sent in order). Rob,
> looks like you checked in a test for this kind of thing in revision
> 512643? Were you seeing problems of this nature?
>
> Having updated to revision 512643, when I run our test, it runs for
> about two seconds before the only remaining ActiveMQ Session Task
> thread hangs in UsageManager.waitForSpace.
>
> Rob, could you give a summary of the broker/VM/application parameters
> that we should try to tweak to avoid this problem?
>
> By the way, we're still seeing the exact same issues with the test
> case
> attached to the AMQ-1148 issue. Maybe the same parameters can be
> tweaked to make that test pass? In that case, we're trying to
> VMPendingSubscriberMessageStoragePolicy instead of having messages
> persisted to the Kaha store.
>
> Thanks for any feedback.
>
> Cheers,
>
> Albert