System Crash

[PROSANJIT MALLICK]

Hi all,

I have a 817 Server with 10.01 hp unix.
Sometimes my server simply reboots without any error messages. I have checked syslog and shutdownlog. There are no coredumps. No logs in preserve or crash.
There is no clue as to what happened. One of my servers happens to be on UPS so power fluctuation is ruled out.
Please let me know where can I get a clue as to what had gone wrong so that I can take precautions.

Thanks
Prosanjit

[CHRIS_ANORUO]

Check the .sh_history files for any reboot command and also your crontab file and cron log.

[Bill Hassell]

Be sure you have created /etc/shutdownlog. Without this file, there will be no record of power fails or system crashes. syslog will not have anything about a crash (or a powerfailure) since there is no way to write to anything when the operating system crashes or the computer loses power.

Since you are running 10.01, you should upgrade as soon as possible as 10.00, 10.01 and 10.10 have not had any development work done on them for many years. 10.20 is the minimum you should be running. Be sure to add the Support Plus CDROM of patches once you install 10.20.

[Roulet]

Hi,

Might also be /var/adm/shutdownlog to create.
You can also make a script witch make a copy of the dmesg to a logfile, you will get at least a kind of syslog file

Hope this will help.

PR

[PROSANJIT MALLICK]

I do have a shutdownlog. There are no cron jobs
running. The same problem had once occurred with my K-Class server which has 11.00 64-bit loaded. There we had two servers one D-Class with 10.01 and K-Class with 11.00. The K-class rebooted on its own.
There was no traces whatsoever.

[Roulet]

You should try to run a script like
dmesg | tail -f >> /etc/shutdmesg.log

You will get the last kernel message before the reboot

[PROSANJIT MALLICK]

But this will imply that the script should run
continously in the background as one never knows when the server will go down.

[Dan Hetzel]

Hi,

Everything has been said on shutdownlog file.

One question anyway: when did you have your UPS serviced/checked for the last time?

This really looks like UPS went down. It could be a dead battery.

Best regards,

Dan

[bhoopathi_1]

Even we had a similar problem. When consulted the HP Engineers, they asked us to check for the coredumps. But the core dumps was not there because of insufficient diskspace. This may be case with you also. After creating sufficient disk space try running savecore command.

[bhoopathi_1]

The space has to be created in /var/adm/crash. The coredump alone can give some direction to this problem.