HP-UX11.11 JDK131_06: not able to get the thread dump when server freezes

[Manmohan]

We are running a weblogic application on HP-UX11.11 and JDK 131_06. We are also using a third party content management system which is deployed in the same weblogic.

The problem is that sometimes the weblogic server freezes i.e. simply stops responding and we are not even able to get the thread dumps during this situation. Without thread dumps its hard to find the last activity when the weblogic server freezes.

we also upgraded to JDK131_08 and then to 131_13 but still the same issues.

Please advise why this ? Its critical and in production.

[Manmohan]

To add more information on this:
Thw weblogic server used in 6.1 SP4. we also tried upgrading it to wls 6.1 sp7 but the issue still remains.

The latest configuration of the sytem are:
wls6.1sp7
jdk131_13
HP11.11

Also this issue is very random with no pattern which can be drawn out of this.

[Steven E. Protter]

Does the third party product have a log? It might be somewhat useful in this situation.

Questions:

1) What does the /var/adm/syslog/syslog.log say?
2) What is the curent state of java patching? Are all required patches in. http://www.hp.com/go/java
3) Is the system responsive in this situation? ie do other applications like ssh or telent still work? Can the machine be pinged?
4) What is the kmtune output for the kernel? Just wondering if the box has been tuned in compliance with the new java release/application combination. I no of no kernel requirements for java, but this other app, may have issues.

Sounds to me a lot like an application issue.

Now if by thread dumps you mean core dumps.

Try this:

If you have this line in /etc/profile or .profile for ANY user:

ulimit -Sc 0

Comment it out. It prevents core dumps.

SEP

[Manmohan]

thanks for the interest.

the third party product as well as the application logs shows all fine and no pointer to any error.

your answers:
1)feel all is fine here.
2)all pathces are up to date.
3)the machine is responive in this case.the machine can be pinged.
4)not sure about this issue.

actually there is no logs pointing to any specific reason for the issue. The bad scene is that we are not even able to take thread dumps (core dumps) by signalling the JVM using various options of KILL.

At this point we are interested in some way to take the thread dumps of latest activity when the server goes into this "freeze" condition so as to nail down the reason causing this. otherwise we have no clue why this is happening.

Thanks.