Message Board -  Archive

[ Login ] [ Create Account ]
[ Board List ] [ View Board ] [ Post Reply ]
  Author  Subject: Re: system crash investigation

Posted on 08-20-2003 01:25 a.m. ET  reply

Original Poster: Mark Krentel

> Trying to avoid it happening again! It has the Red Hat 7.3 default
> scripts and cron jobs running. Can I get any info from those or the
> logs they produced at around the time of the crash?

System log messages usually go to /var/log/messages, they may or may
not say anything useful. (But to be honest, if you didn't know that
already, then you're not really qualified to be running Apache.)

Basically you need to get a feel for the normal activity of the server
so you can spot unusual activity. Try adding a cron job that takes a
single snapshot of "top" every 5-10 minutes. That will at least show
you the average load on the machine, which programs consume the most
resources and will spot a runaway process. Be sure to rotate this
file frequently (daily) since it will generate a fair amount of

And when the machine crashes, it helps to plug in a monitor.
That will tell you the difference between a panic and a freeze.

> primary tools called by scripts daily: tar (tape backup), rsync
> (backup other server on network), wget (sync with updates from remote
> FTP site)

Does this mean that you're updating parts of the system daily??
If so, that would be far and away my number one suspect.
Not the scripts, but the files they change.


< Previous 1 Next >

Site Contents