HPMC chasis codes.

[b65ran]

I have rp2470 A-class server. I am getting this errors in GSP.

GSP> sl


SL

Select Chassis Code Buffer to be displayed:
Incoming, Activity, Error, Current boot or Last boot? (I/A/E/C/L) e
e

Set up filter options on this buffer? (Y/[N])


The first entry is the most recent Chassis Code
Type + CR and CR to go up (back in time),
Type - CR and CR to go down (forward in time),
Type Q/q CR to quit.


Log Entry # 0 :
SYSTEM NAME: stcssh02-st1039
DATE: 10/29/2007 TIME: 13:48:32
ALERT LEVEL: 3 = System blocked waiting for operator input

SOURCE: 1 = processor
SOURCE DETAIL: 1 = processor general SOURCE ID: 0
PROBLEM DETAIL: 0 = no problem detail

CALLER ACTIVITY: E = system warning STATUS: 0
CALLER SUBACTIVITY: F0 = implementation dependent
REPORTING ENTITY TYPE: E = HP-UX REPORTING ENTITY ID: 00

0xF8E000301100EF00 00000000 0000EF00 type 31 = legacy PA HEX chassis-code
0x58E008301100EF00 00006B09 1D0D3020 type 11 = Timestamp 10/29/2007 13:48:32
Type CR for next entry, Q CR to quit.


I am attaching HPMC chassis codes logs. Please somebody take a look and help me out here. If anybody knows the complete list of HPMC chassis codes that would be great. Because I couldn't find any kind of list like that.

Thanks.

[Prashanth.D.S]

Hi,

I have analysed the ts99 posted found the issue to be with SYSTEM BOARD, involve the HP Onsite team and have it replaced...

CW : coherent write error found in ASTRO chip
Usually this indicates a bad SYSTEM BOARD. However if the SYSTEM BOARD has already been changed you will have to look into a possible SOFTWARE problem.

System Board Part Number..
A5570-69316

Best Regards,
Prashanth

[Stefan Stechemesser]

Hi,

I would strongly recomend that you open a case with HP support and let them analyze the server logs in detail.
The HPMC is interesting. CPU #1 wanted data from I/O address 0xc180_80ff_fb60_203c which belongs to PCI slots 3 and 4 (shared slots, HW pathes 0/6/2 & 0/6/0 ).
In addition to this, we have a PCI error from the same PCI bus telling us that a PCI timeout ("PCI no DEVSEL#") has happened:
"Timestamp = Mon Oct 29 13:39:09 GMT 2007 (20:07:10:29:13:39:09)

OV RQ RS ESTAT A C D corr unc fe cw pf
-- -- -- ----- - - - ---- --- -- -- --
X ERR_TIMEOUT X

IO Requestor Address = 0xfffffffffed3c000
IO Target Address = 0x00000000fb60203c
IO Responder Address = 0x0000000000000000
IO Physical Location = 0x000000ffff03ff83
IO Hardware Path = 0x00ffffffffff0006

Module Error Register = 0x000000000000001c"

Based on this information I would not recomend to replace the systemboard but the card in either slot 3 or 4 (maybe a crash dump analysis could tell you what card was accessed before the HPMC happened.).

The interesting thing on this HPMC is that CPU #0 did not simply got a broadcast HPMC signal, but it logged itself a timeout "purge TLB timeout" which means, it wanted that CPU #1 should clear its table with physical to virtual address translations accidentaly when the HPMC has happened, which timed out. I guess this is only a sideeffect of the PCI related HPMC.

best regards

Stefan

[b65ran]

Thanks Prshant and Stfan for replying. I am gonna replace pci on slot 3. But how how do you understand the codes. I am pretty much new at HP 9000. If you please point me out how do read the codes that will be great.

Thanks

[tkc]

hi,

if this problem is reproducible, you just remove the card in slot 3 & 4, one at a time and confirm the culprit. if system still hpmc after both cards are removed, do replace the pci i/o backplane or the system board.

[tkc]

hi,

if this problem is reproducible, you just remove the card in slot 3 & 4, one at a time and confirm the culprit. if system still hpmc after both cards are removed, do replace the system board.