unexpected HPMC (Non Production Box)

[Scott Frye_1]

HP-UX 11.i box has been rebooting since Saturday. I came in this morning to find our test box rebooting again and again. I noticed the Unexpected HPMC message. I searched the forums archives to find it is most likely a HW issue. How do I determine what I'm looking at. I'm assuming it may be a CPU but I was wondering if there is a way to tell for sure. **NOTE** this box never comes up completely to gain access to logs.

Thanks

Scott

[Robert-Jan Goossens_1]

Hi Scott,

Can you log into the gsp ?

Type CTRL b on the system console to enter GSP (Guardian Service
Processor) mode.

Access the GSP "error" log using the GSP "SL" command.

Regards,
Robert-Jan

[Steven E. Protter]

Yes Hardwre.

HPMC
High Priority Machine Check

System must boot immediately.

If it keeps booting it may be a CPU in a single CPU otherwise the box would come up with 1 CPU

To find out.

At the console interupt at the 10 second prompt

There is an In command (help may get you what you need)

There there are some basic hardware checks. You can check the cpu, memory and networking.

Then you can call hardware support with confidence.

SEP

[Scott Frye_1]

Robert-Jan,
I wish we had a GSP, this is an old K class and we don't have a GSP.

SEP,
I will interept the boot and report back.

The exact message is
Unexpected HPMC. Processor HPA FFFFFFFF'FFFA0000

Oh, and I also forgot to mention we are not paying for HW support so I guess we are on our own.

Thanks

Scott

[Bill Hassell]

No simple answer. The Kbox lacks the sophisticated processor fround in the GSP. You can try interrupting the boot process and see what diag commands may be present in the processor ROM code. There are offline diags available on the core/install CD. Since you'll need to spend some time learning about all this, you might start by booting into single user mode. If that works, you can list the ts99 log but you'll still need some help to interpret many of the registers. Another possibility is to use parts from a spare K-box. If the K-box has multiple processors, try pulling all but one. Of course, the bad one may be the one that's left so swap that. Also remove RAM down to just a couple of DIMMs.

However, if the problem is in the backplane or a power supply signal, you'll need to pay for a service call.

[Scott Frye_1]

Also, this is a 4 CPU box. with 4 gig of memory

[Scott Frye_1]

Shouldn't this be under sysadmin and not databases???

[Steven E. Protter · Accepted solution]

HP will fix this on a T&M basis, meaning time and materials.

It will may cost more than a hardware service contract, but you are not strictly speaking on your own.

Replacing a CPU is not the most difficult operation to attempt.

SEP

[Stefan Stechemesser]

Hi Scott,

mybe you can catch the chassis codes that appear during the processing of the HPMC. Sometimes it is possible to determine the problem cause without the full HPMC dump.
The best way is to connect to the console, set the key switch to service mode, press CTRL-B to get to the access port and then enter "sl" which will display the last 50 chassis codes. Post it here, and maybe someone can tell something about the error cause.

best regards

Stefan

[Scott Frye_1]

Thanks Stefan, trying it now.