I am running Weblogic 10.3.4 on Red Hat Linux 5.6. I have a JMS cluster of 2 nodes and I connect to the JMS server with something like "t3://ip1,ip2".
I'm running a long duration test over a few days and I notice that after a while, my JMS message insertion rate slows down and I see some messages like these in my log files:
####<Apr 5, 2012 2:42:52 PM EDT> <Info> <WorkManager> <linpe685001> <linpe685001-Server1> <weblogic.cluster.MessageReceiver> <<WLS Kernel>> <> <> <1333651372247> <BEA-002936> <maximum thread constraint ClusterMessaging is reached>
I've looked around for a way to increase this thread constraint but I have not been able to find anything. Can someone please tell me how to change this configuration in hopes to fix this message?
I took a quick look at the code and the thread pool in question appears to be related to WebLogic's multicast cluster implementation; furthermore, it looks like it can only safely have a single thread (so adding threads is not an option). There's step-by-step cluster troubleshooting guidance in the WebLogic documentation, or, alternatively, I'd bet that reconfiguring your cluster to use "unicast" instead of "multicast" mode would take care of the log message.
That said, I think it's hard to say whether the log message is related to the lowered performance. Thread dumps, JVM profiles, and wl statistic dumps could help with diagnosis.
Looks like problem with workmanager which you configured for application. The number of threads assigned to this work manager has reached the configured value in the maximum threads constraint. Please check your workmanager.
I recommend leaving this particular internal WM alone and avoiding any attempt to add threads - the code it invokes requires single threading. See my other post for more explanation, diagnosis advice, and a potential work around.
I agree with u but I would say, user has to chek MaxThread Constraint in server as well, Summary of Deployments > <application Ejb or wahtever>Monitoring>Workload to what size has defined
But of course bottom line need to collect Thread Dump, Logs etc to analyze completely to get the root cause.