system crash dump

[stephan berger]

Since about 6 weeks, I have got some trouble with my 9000 server (about 6 years old, HP-UX 9.07). Sometimes - appr. once a week - and without any visible reason, the systems crashes and reboots.
I've made some logfiles but I don't see any problems, and I don't find any core files.
Now I want to make and analyse a crash dumpfile, but i don't know how to do :-(
I tried "crash" but it doesn't work.
has anyone got an idea for me?
thanks
stephan berger

[Stefan Farrelly]

How to setup crashdumps and analysis on 9.07 ?? wow, not sure most of us can remember back that far.

Anwyay, the stats you attached are rather interesting. Looks like a lot of paging going on at times. You must have memory pressure. My guess is you slowly run out of memory during a week and this causes the OS to panic and crash/reboot. Can you find the source of the memory leak ? probably an application which you could periodically shutdown to fix the problem, or try to find an OS patch which would prevent the crash (but finding old 9.07 patches ! -tough one).

[Curt Thompson]

Hello Stephan,

Sure, a panic (crash) is one thing. How about
an HPMC crash? Do you have any valid tombstones in /var/tombstones? Maybe the system
is rebooting due to a hardware problem and that data will be in the 'ts99' file.
Just checking...

Good Luck,
Curt

[TrustNo1]

Does it realy crash or just reboot?
For what it's worth, I had a 9.07 (or 9.05) server that liked to reboot for no apparent reason.
It had nothing to do with the OS, rather it was a disk going bad.
Once the RPM's dropped below a certain point, the dang thing would reboot. Occurences of this became more frequent as time passed, finally allowing me to be listening one day when it happend.
~jdk