HP 9000 L-Class rebooting at will

[Robert_Jewell]

I would agree you have a flaky power monitor board and should have it replaced. The system is simply resetting, which is why you have limited logs at the OS level. Likely the PM board is not holding a stable voltage at some level and at times this just triggers a log, while others it forces a reset of the server.

Each time this happens you stand chance of having your root filesystem corrupted so I would not delay the replacement if you have a choice.

Good luck!

-Bob

[Gary_O]

Dumb question, but is the platform monitor card and the power monitor card the same thing.

I found these 2 parts for the platform monitor card, but wondering if there really is a power monitor card.

New Part # Exch. Part #
A5191-60010 A5191-69010 Platform Monitor Board

[Andrew Rutter]

hi,

yes thats it

it could be A5191-60110 or 60210 as well, depending upon the revision

-60210 is the latest

Andy

[cnb]

Hi Gary,

I came across another L-class server doing the exact same thing as yours, but without the power monitor failure messages.

Did you replace the Platform Monitor and did it resolve the intermittent reboot problem?

Best Regards,


-cnb

[Gary_O]

Hi cnb,
I did 2 things.
I pulled the server apart and reseated
the power monitor card a couple times.
I looked at the contacts on the card, they
seemed to be good / clean.

When I booted up the system, I rebuilt my vmunix. I wasnt sure with all the reboots if something wasnt getting corrupted.

In any case, I have NOT seen the power monitor
error since.

I think I got lucky. Good luck to you!!

[Gary_O]

I acquired a power monitor card, and replaced the old one. It has rebooted twice since all on its own.
I am not getting anything in the /var/adm/crash log, nor
a message in the /etc/shutdownlog. Nothing new
in the GSP. By the way, i reset the GSP, but still see info from Dec 2010? Is there a way to clear this stuff

Cant figure it out.
I've included some cstm stuff...

Please Help. THanks!

[cnb]

Hi Gary,


CSTM did not capture all of the status INFORMATION from each component. You may want to go back into cstm and issue:

cstm> sel all
cstm> info all
cstm> map (and then map again to wait until all components show INFORMATION XXXX) and then;
cstm> il

just to completely check all of the components. This may or may not buy you anything.

To clear the PIM data from cstm, you can issue CLEARPIM in the BCH SERVICE menu. This will clear old PIM data but won't clear the GSP chassis log files.

FWIW -
The system has really old PDC (system firmware) and although it may not resolve the issue, it might pull more accurate data from the system when experiencing the intermittent shutdowns. Later versions addressed some shutdown issues:

http://bizsupport1.austin.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=15351&prodSeriesId=2512017&swItem=PF-CRHW4411&prodNameId=322770&swEnvOID=7&swLang=13&taskId=135&mode=4&idx=0

HTH


Rgds,

[Gary_O]

Updated firmware via patch PHSS_32705
new PDC Firmware Revision: 44.28

We'll see if she behaves any better.
Hopefully I'll get a crash dump if it happens again.

Thanks all

[cnb]

Hi Gary,

I can't recall if you ever upgraded the GSP to A.01.12?


There was a fix in that version for the an interconnect medium not responding issue. Like we said before it may or may not be the resolution but it should be considered as another option.

Best Rgds,

-cnb

[Gary_O]

Rebooted 5 more times since I upgraded the firmware.
Apr 15 12:11:50 acp /sbin/init.d/sendmail[1197]: #### rebooted ####
Apr 15 12:31:10 acp /sbin/init.d/sendmail[1197]: #### rebooted ####
Apr 15 12:47:59 acp /sbin/init.d/sendmail[1197]: #### rebooted ####
Apr 15 13:46:27 acp /sbin/init.d/sendmail[1197]: #### rebooted ####
Apr 15 15:33:39 acp /sbin/init.d/sendmail[1198]: #### rebooted ####

I still have no idea why its doing this.
Nothing in /etc/shutdownlog and nothing in /var/adm/crash

Attached is everything from cstm

Thank you very much!