Mysterious 10.20 system halt

[Joseph Brazell]

I am experiencing a mysterious system halt on our HP 9000 D class server. The system runs fine while in active use, but if left idle for a long period of time (overnight) the system halts, and there is no activity on the console: no prompt (blank screen), no disk activity, nothing. The only way to recover is to cut the power and reboot.

This has happened two days in a row. Only the HP-UX 10.20 operating system is present on this server at this time, but we will be loading Oracle-HR and other applications in a few months, and I would like to get this problem solved ASAP. I am new to HP-UX system administration and have a lot to learn, but has anyone else experienced this event?

[Vincenzo Restuccia]

See /var/adm/syslog/syslog.log
run STM (diagnostic).

[Bill McNAMARA_1]

you want to look at your shutdown.log and cron tasks. The syslog may also be usefull.
Look for panics from BCH>SER pim hpmc
and /var/adm/crash/ts99 dates.
The /etc/rc.log may also help.

Later,
bill

[Steffi Jones_1]

Hi Joseph,

try to reset the console once before powering it off.

- turn console off
- push key 'd'
- turn console on
- release 'd'

See if that clears up the console.
When you have this "hang" situation, can you still telnet in from another box? Do you see any indication that the box is still working, like flashing lights, etc.
Check if any log files report problems. syslog is a good starting point, but check also the OLDsyslog.log in /var/adm/syslog. Also you can run dmesg and keep an eye on that if it actually fills up with messages.

Steffi Jones

[John Bolene]

If there was nothing in the log files, you can try to get a dump using the TOC button.

It sure sounds like hardware to me.

I have a D box running 10.20 that has been up 630 days now.

[MANOJ SRIVASTAVA]

Hi Joseph

Please call HP to replace the ON/OFF button , switch , it will solve your problem .

Manoj Srivastava

[Joseph Brazell]

The subject system halt occurred once again today - this time within an hour of rebooting. So, the halt is not being caused by the server just sitting idle for a long time. (The Common Desktop Interface screen saver was running at the time and the image was frozen on the screen.)

It has been impossible for me to use the system logs to research the problem because they are being restarted anew with each power off and reboot. Any event that points to the problem being written to the logs is being erased with each restart.

Anyway, I believe the problem is definitely hardware related and I'll be contacting HP service about it. Thanks to everyone for their advice.

[Mark Vollmers]

Joseph,

You should be able to grab the old log, unless if there are two reboots back to back before you can get in. Every reboot, it copies /var/adm/syslog/syslog.log to /var/adm/syslog/OLDsyslog.log. This is the log to look at. I would imagine that there is something in these that will point to the cause. Good luck.

Mark