HPUX 11.23 (RX7620server) Panics

[Natesan_1]

Hi All,

RX7620 system installed with Oracle db 9i installed on HPUX 11.23 , system Panics with following strings:

panic Spinlock held too long!

I have checked following,
1. No Hardware error
a.checked MPlogs
b,Logtool output
c,no Err dump
d,no error mesg on syslog about hardware

2. installed following patches which was mentioned in the ITRC for above problem,

PHKL_30146 s700_800 11.23 fcntl patch

PHKL_30245 s700_800 11.23 fixes for TLB fault and spinlock panics

Now still the problem persists with same panic string.

Did any one come across similar issues?.
I am trying to analyse the Dump with dump tool for 11.11 I am not getting most of the information from the dump,

Any Gentle folks has input can wtite to me.

Thanks in advance to all of them who would reply.

[Jeff Schussele]

Hi,

You need to engage the HP performance team IF this is a critical app AND you have the appropriate support contract.
They have kernel tracing tools that will drill down to the source of the problem.
Else your best other option is to wrap tusc around the binaary.

Rgds,
Jeff

[Brian Hackley]

Hi,

Contact your local HP Support Center and have them get you the "crashinfo" package to run against the dump. This is the replacement for the "old" q4 package. That will give them all the info that is needed to analyze the spinlock dump and determine if it is a known bug or a newly found bug.

You can TRY to use q4:
. /usr/contrib/Q4/bin/set_env
cd /var/adm/crash
ls (decide which crashdump to look at e.g. crash.0)
cd crash.0
q4 .
At q4 prompt:
run WhatHappened > what.out
run SpinlockDepth > spin.out
Possibly:
run Analyze > au.out

The outputs may help your HP Support folks if you can't get a copy of crashinfo.

Hope that helps,
-> Brian Hackley