Superdome sx2000 hardware degraded

[Peter Marais]

When I reboot a particular npar on my superdome I receive a warning that my harware is degraded:

Keyword = CELL_HW_DEGRADED

Description:

A dimm or CPU has failed and is not operational for the system. This event is emitted prior to determining if the cell should be integrated into the Partition.

Cause/Action:

A deconfigured dimm or cpu has been detected. Examine earlier events to isolate the problem.


Recommendation:

None

Yet when I look at the system via parstatus or stm I can not find any degraded memory or cpu...
Also my SEL does not give me any more detail of what failed

any other sagestions of where I can look? Running HP-UX 11.23

[Torsten.]

Use "info all" from EFI or BCH, depending on the CPUs installed (RISC vs. Integrity server).

[Peter Marais]

Thanks Torsten is there any unix commands I can try as I do not have the luxery of bringinig the system down

[Torsten.]

You can check the memory with stm.

[Matti_Kurkela]

If you have currently unused iCOD/iCAP CPUs in your system, the system is smart enough to "swap" a failed CPU with an unused one.

The rationale is apparently something like: you've paid for X CPUs, so the system will provide you with X CPUs as long as there are enough usable CPUs to do that.

MK

[Andrew Rutter]

hi,

Did you run the info tool in stm? and look at the log for the memory or cpu's. Also try running the info tool on the system, usually the first item in the map in CSTM.

this should show any degraded/deconfigured items

Andy

[Peter Marais]

Hi Guys

Thanks for all the responces.

As I stated in my opening mail stm reveals no deconfigured memory or cpu's. The only message I get is from SEL on the mp stating the qouted worning (from opening statement)above and marks the cell board with a w in VFP. No ICOD running on this system
Machinfo also gives the correct amount of CPU's and memory.

[Torsten.]

So I would investigate the cells from MP with

MP:CM> ps

[Peter Marais]

Thanks Torston, ps only gives a power status. No joy with that.

[Torsten.]

I would expect an item in "failed" status.

However, there must be a related message in the diags logs, maybe in root's mailbox and in the MP logs.