I've got a server which I recently dist-upgraded from Woody to Sarge. It's crashing about once per week. I haven't been able to find the cause of it yet. Nothing unusual in /var/syslog or /var/messages that I can tell.

Any advice on finding out what's causing a machine to go down? My guess is that logs may not actually contain anything useful at the time of the crash, since the machine may not have time to write to them when things go wrong.

My plan is to set up a script on root's crontab to periodically record some system vitals in log file, and then sit tight, wait for the next crash, then pore over said log file looking for a pattern.

Any tips greatly appreciated.

A big question is: what shows up on the main console when this happens? Does it lock, spontaneously reboot, or give a kernel panic?

I'd expect it to be a hardware fault, but with so little info it's only possible to guess.

Matir, it's a remote server, and it runs for a week or so before seemingly randomly going down. So I'm not logged in when it happens.

yowi, I'm not so much looking for someone to offer a reason as to why it's having an issue, but rather looking for suggestions on how I might ferret out the cause.

So far, I'm trying my special log file idea and seeing what it tells me.

Thanks.