System rebooting

[Paul Gorniak]

I have a D220 and out of no where the system just
reboots in the middle of the day. It is running HP-UX 11.00. Anyone know what could be the problem.

[Mark Grant]

Every time I've seen an HP box do this it's been currupted RAM.

However, I imagine it could be lots of things. Have a look in the old syslog. Also check cron jobs. Check "at" too (I once saw an at job that installed another copy of itself and then re-booted the machine).

[MarkSyder]

Have you checked the last entries in /var/adm/syslog/OLDsyslog.log?

Alternatively, is there anything in /var/adm/crash?

Mark Syder (like the drink but spelt different)

[Shaikh Imran]

Hi,
Apart from syslog & dmesg.
Please run STM and take the info of all the devices including memory.
Also see if you get some mail to root.

Regards,

[Victor BERRIDGE · Accepted solution]

Hi,
What is your configuration?
Do you have an external disk subsystem attached ? Do you have mirrored internal system disks?
is there anything special in /etc/shutdownlog?


All the best
Victor

[Bill Hassell]

The most common problems that cause HP-UX to reboot are:

1) HPMC (High Priority Machine Check) which is a hardware failure (CPU, RAM, backplane, power supply)

2) Power interruption (no UPS or UPS did not work or someone pulled out the power cable)

3) System crash (a software failure) which is almost always fixed with a patch (or patches)

Nothing will be logged under any of the 3 above conditions because nothing is running. A system crash will jump to a routine (if it hasn't been corrupted) that tries to dump all of RAM into the dump area (seen with: lvlnboot -v). Then the processor starts to reboot. During the reboot, a program called savecrash will be run but only if the directory /var/adm/crash exists and SAVECRASH=1 is enabled in /etc/rc.config.d/savecrash. savecrash will also add an entry in /etc/shutdownlog (not syslog) concerning the crash event. shutdownlog will contain an entry for any of the above 3 events. If the problem is a software crash, /var/adm/crash will have very large files in it that need analyzing to determine the reason. Or you can bring the system up to date on patches and it will likely not crash again.

[Paul Gorniak]

This is the error in the shutdown log that I got.
Reboot after panic: TOC, pcsq.pcoq = 0.6a744, isr.ior = 0.6a740
08:26 Tue Jun 15 2004. Reboot after panic: TOC, pcsq.pcoq = 0.6a748, isr.ior = 0.6f04c8
09:00 Tue Jun 15 2004. Reboot after panic: TOC, pcsq.pcoq = 0.6a748, isr.ior = 0.6f04c8

[Bill Hassell]

This is number 3, a kernel panic or crash. You can send the crash dump to HP for analysis or simply bring all the patches up to date.