Hi Alex
My preference would be to find out what the problem is and see what
needs to be done to rectify it. We certainly didn't see anything like
this in any of the performance tests we did run.
Let's discuss this at the Open MQ Users alias -- (We'll repost back
here, when we've got something definitive to report.)
Thanks,
-- Ed Bratt
Alex Sherwin said the following on 4/1/2009 11:56 AM:
> I've posted previously about bad performance of OpenMQ 4.3 since
> glassfish 2.1 came out. I've never had these issues with the previous
> version of OpenMQ bundled with glassfish v2ur2.
>
> Basically I see this most often when I re-deploy an EAR that contains
> an EJB-JAR with 3 MDBs (initial pool size of 25 each). I've tried
> this with larger and smaller pool sizes etc, and it always happens
> regardless.
>
> When it begins to occur, all MDB's seemingly just wait to process a
> message for 30 seconds or so (no java processes in app server or imq
> are doing anything at all.) Then it will begin processing the message.
> This slowdown is per message. I.E. If i throw in 10 messages, all 10
> will wait ~ 30 seconds, then all process. If I stagger them for ~ 5
> seconds inbetween, they are processed similarly 5 seconds apart.
>
> I can switch to glassfish v2ur2 with the exact same EAR and never see
> this behavior.
>
> Is is possible to downgrade the OpenMQ binaries in 2.1? There are
> other bug fixes in glassfish v2.1 that we are relying on.
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>