System panic

[Unix Administrator_6]

I had a system panic:

K370 running 10.20

No errors, in syslog or tombstones

Shutdownlog had: "15:15 Tue Sep 05 2000. Reboot after panic: Data page fault"

Logged in mstm on console checked info and failure logs on all 4 CPU's and memory - nothing.

Trace log has this in it, but I don't know what it means:

stack trace for event 0
crash event was a panic
can't find unwind or stub descriptor for pc==0
panic+0x10
report_trap_or_int_and_panic+0xe8
trap+0xa48
$call_trap+0x20

I am ftp'ing core dump, kernel, trace, patch_list, ana.out (q4 thing) to HP to anylize, but I usually get more info from you guys then their reps.

Anything you guys see in this?

[Anthony deRito]

Seems like all hits are related to a kernel race condition and also prescribes patches to fix the problem. q4 output is key. Ask HP for required patches for this condition.

Tony

[Manju Kampli]

The way to look in to this crash is debug the core dump file. I did had similar page fault error and system crash. the result was with some bad patches on the system and HP suggested for installing superseeded patches which fixes the problem. Since you have already sent the core dump to HP, better you wait for their response and do accordinglty.

Thanks for your belief in FORUMS technical ability and we all are proud of it.

[Insu Kim]

It's HP responsibility to analyze core files but applying bunch of SCSI and LVM patches on the system would be helpful based on my experience.

PHCO_16591, PHCO_18563
PHKL_16751,16957,17858,20611,21595,21661,22224
PHNE_20834

Notice that some of them have dependencies.

[Vincente Fernandes]

You need to apply the latest patches on your system for which HP will suggest after doing analysis of the core dump. Sometimes it can happen on account of some process trying to accesses data in the memory. You need to wait till you get the feedback from HP about the exact cause/problem.

[Andy Monks]

Sherri,

There is not a lot of info from the stack trace. My guess is that your using version 1.52 of q4 (as shipped on 10.20). This version of q4 doesn't really know how to handle PA2.0 machines like the K370. This is why the stack trace is incomplete (well most likely).

A newer version of q4 is available via a patch ( PHCO_20261 ), or you can ask the RC to send you latest q4 package (even newer than the patch).