topic Re: tombstone file in Operating System - HP-UX

tombstone file

AUJ — Tue, 16 Jul 2002 08:41:07 GMT

Hi everybody,

I just want to check something about the output of the "pdcinfo" and the STM utility. I have a system, when I checked tombstone file, the timestamp stated before "the chassis code information" appear is MAY 2002 and then I try to move all the tombstone files to /tmp directory and then reboot the server, then I checked the ts99 file unfortunately when I read the said file, I got the same timestamp "May 2002". And when I run STM utility, I got the same result (timestamp is May 2002).
Now, my question:
1) Is these utilities reliable?
2) Where can I find the correct information?
3) Can you recommend other tools, I can use to check if there is hardware related problem on my server?

Thank you very much in Advance.

AUJ

Re: tombstone file

BFA6 — Tue, 16 Jul 2002 08:55:43 GMT

Hi,

We have a box that when rebooted has tombstone info time stamped Dec 2000

We are running an old version of the diagnostics (version 15 I think) and HP have recommended upgrading to at least version 24.

Don't know if this helps you.

Regards,

Hilary

Re: tombstone file

AUJ — Tue, 16 Jul 2002 09:02:51 GMT

Hi,

The STM version installed in my server is A.29.00
and pdcinfo is 2.35.

Thanks,
AUJ

Re: tombstone file

melvyn burnard — Tue, 16 Jul 2002 09:51:02 GMT

What exactly are you saying?
Is this the timestamp of the file as shown by ll?
or are you reading hte file?
It is a simple text file so you can vi or more ts99.
If needed you can even attach it to this thread

Re: tombstone file

Patrick Wessel — Tue, 16 Jul 2002 11:14:20 GMT

The HPMC handler has written the timestamp in the ts99 file as your server crashed the last time. You will see no change in the timestamp if your server runs stable. It takes another unpleasant HPMC to overwrite it.
To answer your questions:
1) Yes, it is reliable
2) These information are correct
3) STM (cstm, mstm, xstm)

Re: tombstone file

AUJ — Wed, 17 Jul 2002 00:05:06 GMT

Hi Patrick,

Does it mean that even the server has been rebooted and the system produce new ts99, it's still displaying the old crash file, with old timestamp?

What about STM? When I run cstm, the get the info of CPU, I got the same information. Is STM reading the same file?

SAMPLE:

Timestamp = Thu May 23 05:12:03 GMT 2002 (20:02:05:23:05:12:03)

HPMC Chassis Codes

Chassis Code Extension
------------ ---------
0x0000082000ff6242 0x0000000000000000
0x1800082011006312 0xcb81000000000000
0x0000082000ff6462 0x0000000000000000

Thanks,
AUJ

Re: tombstone file

Patrick Wessel — Wed, 17 Jul 2002 11:06:50 GMT

AUJ,
The PIM dump is stored in the NVRAM on the system card of your server. When ever you use the pdcinfo program or you run the info tool of STM on the CPU, pulls the system the information out of the NVRAM and does not delete the NVRAM. The only time the context of this section of stable storage changes is when a new HPMC occurs (or you replace the systemcard).

STM and pdcinfo use the same resources (PDC calls) to get these information and pdcinfo is used to write the ts99 file at reboot. At reboot reads the system the pim data and compares it to the ts99 file. If the ts99 is not present or the timestamps differ to the current pim, it creates a new ts99

Re: tombstone file

Christopher Meenan — Wed, 17 Jul 2002 12:23:52 GMT

Hello Everyone,

It is true that this information is stored on the system board itself and is a record of the last failure.

If you wish to clear this information, stop at the boot console menu. Go to the service menu, and select the CLEAR PIM option. This will clear the contents of the onboard storage. You will still get a new ts99 each time you boot ( you may have noticed the "copying processor logs to /var/tombstones" during startup. From now on, your ts99 files should reflect "No relevant timestamps" for each processor.

Best regards,

Chris Meenan

Re: tombstone file

AUJ — Thu, 18 Jul 2002 02:27:29 GMT

Guys,

So, how can I determined let say I have a hardware related problem? As you have said pdcinfo and STM is reading the same resources to get the information and the information reflected here is the record of the last failure.

Thanks a lot.

AUJ