K460 server tombstone errors ts99

[Rui Medeiros_1]

Could someone tell me what this means.

[Ken Hubnik_2]

HPMC errors, basically the PIM. this file is created if your system rebooted or panic'd and has any HPMC errors. This file is usually what HP support uses to find out why the system panic.

[Stuart Abramson_2 · Accepted solution]

It means you have a problem, and if you don't fix it, it will happen again.

5. vi /etc/shutdownlog

15:37 Sun Jul 11, 1999. Reboot: (by gbo302-d!root)
11:26 Mon Jul 12, 1999. Reboot: (by gbo302-d!root)
13:19 Mon Jul 12, 1999. Reboot: (by gbo302-d!root)
14:34 Mon Jul 12, 1999. Reboot: (by gbo302-d!root)
12:10 Tue Jul 13 1999. Reboot after panic: Data memory protect fault
22:31 Wed Jul 14, 1999. Reboot: (by gbo302-d!root)
21:06 Sat Jul 17, 1999. Reboot:
20:19 Sun Jul 18 1999. Reboot after panic: Data memory protect fault
05:09 Mon Jul 19 1999. Reboot after panic: Data memory protect fault


6. cd /etc/tombstones
ll

-rw-r--r-- 1 root root 25615 Jul 11 15:49 ts91
-rw-r--r-- 1 root root 25615 Jul 12 11:39 ts92
-rw-r--r-- 1 root root 25615 Jul 12 13:31 ts93
-rw-r--r-- 1 root root 25615 Jul 12 14:47 ts94
-rw-r--r-- 1 root root 25615 Jul 13 12:02 ts95
-rw-r--r-- 1 root root 25615 Jul 14 22:44 ts96
-rw-r--r-- 1 root root 23603 Jul 17 21:19 ts97
-rw-r--r-- 1 root root 23603 Jul 18 20:19 ts98
-rw-r--r-- 1 root root 23603 Jul 19 05:09 ts99

a. "ts99" is always the last tombstone written.

b. After every reboot, HP-UX reads the "PIM' info and puts it in ts99.

c. Here is what it looks like:

HP-UX gbo318-d B.11.00 A 9000/800 464574381

CPU-ID( Model ) = 0xe

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

Timestamp = Mon Jul 19 07:40:22 GMT 1999 (19:99:07:19:07:40:22)

HPMC Chassis Codes = 0xcbf0 0x5008 0x7d05 0x5404 0x5504 0x5604
0x5704 0xcbfb

General Registers 0 - 31
00-03 0000000000000000 00000000000003f0 00000000f006ef54 000000000000017

04-07 0000000000000055 fafefffefc0755e0 00000000f017b87f 00000000000fc07

d. If the date in the first (most recent) HPMC, matches the date of the
last reboot, then you have a reboot caused by an HPMC: This implies
a hardware error. Call a CE.

e. If you have a panic with NO HPMC, then you have a software induced
panic, and you should make a software call and look at the DUMP in
/var/adm/crash.

7. cd /var/adm/crash
ll

root@gbo318-d:/var/adm/crash# ll
total 4
-rwxr-xr-x 1 root root 1 Jul 19 16:56 bounds
drwxr-xr-x 2 root root 1024 Jul 19 16:58 crash.6

8. Read syslogs:

cd /var/adm/syslog/
ll
vi OLDsyslog.log
vi syslog.log

[Michael Steele_2]

HPMC = High Priority Machine Check, and no one want's one of these. '/var/tombstones' collects HPMC's and for K class servers 'tombstones' are the sole source of all HPMC information. For later servers, like L and N class, the same information can be viewed in the GSP logs.

Refer to the PDC level. PDC firmware is important for correct tombstone logs. Here is the PDC firmware patch for a K460.

HP9000 Kx60,979KS/x00 Firmware Revision 41.33 PF_CMHK4133

http://www1.itrc.hp.com/service/patch/patchDetail.do?patchid=PF_CMHK4133&context=firmware:cpu

[Michael Steele_2]

Regarding "...sole source...", there are HP CE commands like 'pim info' that can be run from the service processor.