HPE 9000 and HPE e3000 Servers
cancel
Showing results for 
Search instead for 
Did you mean: 

help in analyzing HPMC of K380 server

 

help in analyzing HPMC of K380 server

Hi Folks

I have a problem with one of my K380 server
Server reboots with panic message. Unfortunately it is not in AMC. I have the following messages in /var/tombstones/ts99

CPU-ID( Model ) = 0xe

----------------- Processor 0 HPMC Information ------------------

Timestamp = Mon Nov 19 08:21:00 GMT 2001 (20:01:11:19:08:21:00)

HPMC Chassis Codes = 0xcbf0 0x5002 0x7d03 0x5408 0x5508 0xcbfb


-----------------
The ts99 file was created at the time of reboot
now i have following doubts/requirements

1. The timestamp shown is of 2001 ,is it mean the HPMC error is very old or it indicates something else
2. Can anybody give me the list of HPMC codes and it meanings

Best Regards,
Ruban
7 REPLIES 7
Stefan Stechemesser
Honored Contributor

Re: help in analyzing HPMC of K380 server

Hi Ruban,

1.) You are right, the timstamp is very old. You can ignore this tombstone because it has no relation to your actual system panic.

2.) You should contact the next HP support center for HPMC analysis if an HPMC would have an actual timestamp. HPMC analysis is much more than simply taking a look on chassis codes. A simple rule of what chassis codes on K-Class servers mean: 0xcbfb: HPMC happened, 0x5...: I/O or CPU, 0x7... Memory, 0x2... Cache, 0x5y08: Broadcast Error seen by Runway Bus Module with Master ID y, 0xcbfb: branching back to operating system (HPUX performs a system panic then).

The chassis codes are only a summary that the firmware generates during the high priority machine check procedure. In most cases it is nessecary to decode several status and diagnose registers from CPU, I/O Controllers and Memory Controller to determine the error cause. Unfortunately most of the needed manuals are not available for the public and I don't know any internet location with information about this. However in your case, I can tell you that 0x5002 is a "path error" (Data was requested but an error came back from I/O or memory) and 0x7d03 is an "MSI read timeout" which indicates a memory request to a non existing location during a DMA request and is most likely caused ba a bad kernel driver (SCSI and network card drivers are most common). To fix this old problem, install the newest Quality Pack patch bundle from the Support+ CD on your computer.

But do not forget that this error happened 2 years ago ! I would suggest to concentrate on the current problem and let someone from HP analyze the dump that should be created in /var/adm/crash.

Best regards

Stefan

Re: help in analyzing HPMC of K380 server

Hi Stefan
Thank you verymuch for the reply, i have been told by one of my colleque that we need to see when the ts99 file was created and whatever timestamp shows inside the file is nothing but some firmware date.
so now as per ur info, though the ts99 file was created during the time of reboot, since the timestamp inside is old we can ignore it.
Unfortunately this m/c is not in AMC, so no qtn of getting support from HP.
I have analysed the core also using q4. Below are the trace event 0 output .

stack trace for event 0
crash event was a panic
panic+0x14
report_trap_or_int_and_panic+0x84
trap+0xd9c
nokgdb+0x8
EIsKil+0x60
E2IDIs+0xc0
otsstop_blast_routing_database+0xf0
force_otsstop+0x90
OsiDmnIoctl+0x47c
Osxwuput+0xa4
putnext+0xcc
wait_iocack+0x68
str_istr_ioctl+0x72c
hpstreams_ioctl_int+0x370
hpstreams_ioctl+0x50
spec_ioctl+0xac
vno_ioctl+0x90
ioctl+0x78
syscall+0x28c
$syscallrtn+0x0

Please have a look and suggest if any,

Best Regards,
Ruban
Stefan Stechemesser
Honored Contributor

Re: help in analyzing HPMC of K380 server

Hi Ruban,

the startup script /sbin/init.d/pdcinfo creates a new ts99 file EVERY REBOOT . You have do check if it differs from the ts98 to determine if you have to take a look on it.
Unfortunately I cannot help you with the q4 output, because as an hardware engineer, I only have basic knowledge in dump analysis.

Best regards

Stefan
Robert_Jewell
Honored Contributor

Re: help in analyzing HPMC of K380 server

Could you also post your INDEX file from the corresponding crash directory. If not that, then at least the contents of /etc/shutdown.

Looking for the initial panic string which may tell something else.

-Bob
----------------
Was this helpful? Like this post by giving me a thumbs up below!

Re: help in analyzing HPMC of K380 server

Hi Bob

/etc/shutdownlog
12:33 Thu Sep 11 2003. Reboot after panic: Data page fault

INDEX
---------
comment savecrash crash dump INDEX file
version 2
hostname toyota
modelname 9000/800/K380
panic Data page fault
dumptime 1063262945 Thu Sep 11 12:19:05 IST 2003
savetime 1063263538 Thu Sep 11 12:28:58 IST 2003
release @(#)B2352B/9245XB HP-UX (B.11.00) #1: Wed Nov 5 22:38:19 PST 1997

memsize 2147475456
chunksize 67108864
module /stand/vmunix vmunix 23280280 1748517587
image image.1.1 0x0000000000000000 0x0000000003ffe000 0x0000000000000000 0x0000000000004977 3754277296
image image.1.2 0x0000000000000000 0x0000000003ff7000 0x0000000000004978 0x0000000000018a1f 103542320
image image.1.3 0x0000000000000000 0x0000000003ff6000 0x0000000000018a20 0x000000000003c387 1286219828
image image.1.4 0x0000000000000000 0x0000000003ff9000 0x000000000003c388 0x000000000004d437 3467225074
image image.1.5 0x0000000000000000 0x0000000003ff6000 0x000000000004d438 0x0000000000079927 1462605602
image image.1.6 0x0000000000000000 0x0000000002a7d000 0x0000000000079928 0x000000000007fffd 1518503178

Best Regards,
Ruban
Eugeny Brychkov
Honored Contributor

Re: help in analyzing HPMC of K380 server

Ruban,
too many words but information is gathered partially. Please generate the following outputs:
ll /var/tombstones > list.txt
cat /var/tombstones/* > ts_all.txt
zip them (zip, not gzip) and attach to your next reply
Eugeny

Re: help in analyzing HPMC of K380 server

Hi Eugency

Attached the list of ts files and their contect

Please check and reply if you have any clues

Best Regards
Ruban