topic Re: Crash Dump Analysis in Operating System - HP-UX

Crash Dump Analysis

RICK TAN JEE HOW_1 — Sun, 16 Jun 2002 02:02:31 GMT

Hi Anyone knows how to analyse crash dump using Q4 ? and how do we send it to HP ?

Re: Crash Dump Analysis

Amruth — Sun, 16 Jun 2002 02:27:52 GMT

Hi,

You have two options...

1. Tar it to a tape and send it to HP.
2. Using a Good ISP connection upload the dump to HP site.

For uploading consult the Hp-engineer...

Hope this helps..
AR

Re: Crash Dump Analysis

Bill Hassell — Sun, 16 Jun 2002 02:44:33 GMT

The HP Response Centers worldwide can analyze a crash dump for you. Contact your nearest HP Sales office or your HP Channel Partner for the location and contact information.

Re: Crash Dump Analysis

RAC_1 — Mon, 17 Jun 2002 04:58:40 GMT

At the outset I dont advise this is to use.

(Thoough I have used it. I will post detailed on later on)

Copy you crash dump files to another location.

run q4_prep
it creates the q4lib directory
go to that directory
say q4 now

it operns session in VI
find stack

Now it gives some messages saying stach contents, errors etc.

Please note that this might not be the exact one.

I will post detailed one afterwards.

Regards,

Re: Crash Dump Analysis

Trond Haugen — Mon, 17 Jun 2002 05:16:07 GMT

Follow Bill's advice.
If this is a ServiceGuard node first take a look at the end of OLDsyslog.log to see if cmcld performed a TOC.
Apart from the crash files some other files may be if interest:
/etc/shutdownlog - to see if this is the first crash.
Output of show_patches - to see what patches are installed.
rc.log - to see if there are any problems starting the system.
OLDsysylog.log - See above, or any other problems.
/var/tombstones/ts* - Check highest number if timestamp IN the file matches.

Regards,
Tron

Re: Crash Dump Analysis

Paula J Frazer-Campbell — Mon, 17 Jun 2002 06:10:37 GMT

Hi

crash commands
cd /var/adm/crash/core.*
q4 .
trace event 0

Stack trace from the first crash event:
stack trace for event 0

crash event was an HPMC <- typically hardware
or
crash event was a TOC <- hang or Serviceguard TOC
or

crash event was a Panic <- typically software

q4> examine panicstr using a
( fault here)

q4> load ???r crash_event_t from & crash_event_table until crash_event_ptr max 100
(loading 3 crash_event_t???s as an array (stopped bu ???until??? clause)

q4>runningprocs
(02 2 0x2)

q4> trace event 1

HTH

Paula