Again GSP + mirror disk

[Andrey Vorobiev]

Hi, everyone !
I'm new to this forum but nevertheless I have 2 questions :)

Under my supervision I have goat a 3 L-class servers (1 RP5400, and 2 L-2000).
After 2-3 days of operations I receieve daily a mail in /var/mail/root:

permrm5 sent Event Monitor notification information:

/system/events/core_hw/core_hw is >= 3.
Its current value is SERIOUS(4).

Event data from monitor:

Event Time : Thu Jul 18 11:18:21 2002
Hostname : permrm5 IP Address : 10.2.1.153
Event Id : 0x003d36a3fd00000000 Monitor : dm_core_hw
Event # : 32 Event Class : System
Severity : SERIOUS

Service processor not responding

Associated OS error log entry id(s):
0x3d36a3fd00000000

Latest information on this event:
http://docs.hp.com/hpux/content/hardware/ems/dm_core_hw.htm#32

Description of Error:

The service processor is not functioning.

Probable Cause / Recommended Action:

There could be a problem with the service processor. Verify support
processor operation. If the service processor has failed, repair or
replace the core I/O board.
There could be a problem with the system board. Repair or replace the
system board.

v-v-v-v-v-v-v-v-v-v-v-v-v D E T A I L S v-v-v-v-v-v-v-v-v-v-v-v-v

The PDC_PAT_EVENTS function reports that the service processor is not
functioning.

I ve checked the GSP it's ok. Moreover I managed to change the IP of lanconsole and resest the GSP. Messages are still coming. What should be done in this case?

The second question is about mirror disk. It seems that one primary boot and one alternative data disk have failed. WHich is the best action to restore the disks after I get a new ones.

Many thanks in advance,
Andrey

[Cheryl Griffin · Accepted solution]

Andrey,
A couple of things for EMS:
Since you have verified the GSP functioning, change the events to be informational only:
Temporarily disable montoring, then
# vi /var/stm/config/tools/monitor/default_dm_core_hw.clcfg

Searh for the line with:
/EQ:32

# EVENT_PE_SERVICE_PROC_DEAD:
EQ:32 :SERIOUS :TRUE :1440 :60 :4 :NONE :NO_OP:NO_OP:NONE

Change the line from ":SERIOUS" to ":INFORMATION". The line will now read:

# EVENT_PE_SERVICE_PROC_DEAD:
EQ:32 :INFORMATION :TRUE :1440 :60 :4 :NONE :NO_OP:NO_OP:NONE

Restart monitoring.

The real issue will be addressed in an upcoming EMS version.

Next for replacing the disks, look in the ITRC Knowledge Base. There are several documents that describe how to use vgcfgrestore to recover the LVM information for your replaced disks. Example:
"Replacing a mirrored root disk (no hot swap)" DocId: LVMKBRC00006116
"Replacing a bad mirrored disk by hot-swapping in a new disk" DocId: KBRC00009115


Best Wishes,
Cheryl

[eric watson]

Hi Andrey,

You might check to see which version of GSP firmware you are on. The revision A GSP should have firmware A.01.10 installed. The revision B GSP should have B.02.14 installed. Both versions of firmware have fixes for EMS event 32's.

Eric

[eric watson]

FYI:

Here is a short decoder ring for which GSP firmware goes with which system:

N and L-Class (rp54xx and rp7400) use the same GSP firmware but there are 2 revisions of GSP.

N-Class that are 440Mhz or slower always use revision A firmware( A.01.10 is Current)

N-Class manufactured prior to Oct 2001 that are 550Mhz will also use rev A firmware, they wil have a product number of A3639A or A3639B

N-Class Manufactured after Oct 2001 will a CPU speed of 550Mhz or greater use revision B firmware(B.02.07 current).

L-Class with a A6696A GSP needs revision A firmware.

L-Class with a A6696B GSP needs revision B firmware.

A-Class is similar to N-Class older ones will use the A.01.10 firmware.

Newer A-Class with a built in SWC use revision C firmware. ( C.02.03 current)

A quick way to check which you have would be look at your console hardware path, if the console path is 0/0/4/0 you need rev A firmware and if the console path is 0/0/4/1 you need revision B on L or N-Class and revision C on A-Class.

[Andrey Vorobiev]

Cheryl and Eric -
thanks for your clear and profound answers

Andrey