You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by greenbean <Ke...@ngc.com> on 2006/08/22 01:26:40 UTC
Deadlock 4.0.1 Delivery
I have a mdp that attempts to connect to a soap server. The MDP does a
rollback and the message is redelivered on failure. The soap server was
down today by chance. I noticed that the rollback and redelivery stopped
after a period of time. I ran the Jboss through JProfiler and it reported a
deadlock situation with the ServerSessionPoolImp used by the ActiveMQ
Schedulers. Is this a known issue?
I can attempt to provide a test case if needed. However, I thought I would
provide some of the output from JProfiler first to see if that is enough to
explain it, or get it fixed.
We deploy activemq 4.0.1 into JBoss 4.0.3SP1 as a RAR. We use Jencks from
that I pulled from CVS a month or two ago so it would work correctly with
activemq 4.
http://www.nabble.com/user-files/235702/graph.png
http://www.nabble.com/user-files/235701/Current_monitor_usage.csv
Current_monitor_usage.csv
--
View this message in context: http://www.nabble.com/Deadlock-4.0.1-Delivery-tf2143448.html#a5916920
Sent from the ActiveMQ - User forum at Nabble.com.
Re: Deadlock 4.0.1 Delivery
Posted by greenbean <Ke...@ngc.com>.
A note: I was successful in reproducing this problem using activemq 4.0.2.
I did not explicitly mention that we set the maximumRedeliveries very high
since we want the message to be redelivered until the MDP is successul in
the request to a SOAP server. It usually takes serveral hundred rollback
and redeliver cycles for the deadlock to occur. So the chances of occurance
are pretty low.
greenbean wrote:
>
> I have a mdp that attempts to connect to a soap server. The MDP does a
> rollback and the message is redelivered on failure. The soap server was
> down today by chance. I noticed that the rollback and redelivery stopped
> after a period of time. I ran the Jboss through JProfiler and it reported
> a deadlock situation with the ServerSessionPoolImp used by the ActiveMQ
> Schedulers. Is this a known issue?
>
> I can attempt to provide a test case if needed. However, I thought I
> would provide some of the output from JProfiler first to see if that is
> enough to explain it, or get it fixed.
>
> We deploy activemq 4.0.1 into JBoss 4.0.3SP1 as a RAR. We use Jencks from
> that I pulled from CVS a month or two ago so it would work correctly with
> activemq 4.
>
> http://www.nabble.com/user-files/235702/graph.png
> http://www.nabble.com/user-files/235701/Current_monitor_usage.csv
> Current_monitor_usage.csv
>
--
View this message in context: http://www.nabble.com/Deadlock-4.0.1-Delivery-tf2143448.html#a5946077
Sent from the ActiveMQ - User forum at Nabble.com.