Chasis codes ( CPU )

[Satish Y]

Hi All,

Greeting for the day!

We have one HP-UX10.20, 9000/899 , K-class machine, which got reboted after panic.
I have found following chasis codes in stm logs.

HPMC Chassis Codes = 0xcbf0 0x5002 0x5408 0x5508 0x5608 0x5708
0xcbfb

Can any one tell me what these codes means, are they anything serious?.

I have already placed a software call with HP, it takes time to get response.

Regards,
satish.

[G.Kumar · Accepted solution]

hi satish;

It can be a hardware problem also because hpmc error 5xxx is related to I/O cards
.

[Satish Y]

Thanks for information.
I know it is a harware problem, but I placed software call to analyze core dumps generated by system, to traceout corect device and problem.

Regards,
Satish.

[Tom Wittenberg_1]

You have a runaway bus error and do not have enough information here to solve the problem!

[Rainer von Bongartz]

you don't need a software call , place a hardware call

HPMC is a High Priority Machine Check related to HW errors

look in /var/tomstones to find more info ln files ts

with date of your crash

[Ajitkumar Rane]

hi satish,

Its very difficult to guess what the problem would be as the single line o/p you have given.
Basically you need to have a HPMC decoder to analyse this. You should check your /var/tombstones/tsxx file to get full HPMC, the ts99 file is created everytime you reboot the server.so check for the date for the tsxx file when it happened.
If you do not have a HPMC decoder install on your server then only HP can help you to diagnose the error.

Good Luck.

[Satish Y]

Thanks to all for ur response.
There is no /var/tombstones directory in our machine....
how to enable to generate a file ts99 in /var/tombstones?.

Cheers...
Satish.

[James R. Ferguson]

Hi Satish:

In '/etc/rc.config.d/pdcinfo' change '0" to "1" for the 'PDCINFO' variable. This will toggle the generation of tombstone files.

...JRF...

[Johannes_K]

Hello Satish,

You may have a older diagnostic version, which does not generate the /var/tombstone/ts99 file at each reboot.You can run /usr/contrib/bin/pdcinfo instead, that will read all the required information. I suspect the most significant information can be found in the chapter "IO Module Error Log". If one of the lines in the columns "Sev/Estat" is different than "he 0x0d" it will indicate the problem. Most likely the "Responder" address is the is failed component then.

regards
Johannes