[wplug] What to do with a flaky machine

[Tom Rhodes]

On Tue, 14 Aug 2007 11:00:36 -0400
"Michael Semcheski" <mhsemcheski at gmail.com> wrote:

> I thought I'd pose this question to the list...  I've got a Dell
> Precision workstation of an early 2000's vintage.  It was a pretty
> good machine back in the day.  Its got a 15,000 rpm SCSI, two Pentium
> 3's, and 512MB or 1GB of ECC memory (I think.)
> 
> About two years ago, it was made into a general purpose, nothing too
> critical server.  We had a CVS repository on it for a while, and a
> couple of mediawiki instances, nagios, and probably more stuff.  It
> was running OpenSuSE 9, which I'm not too fond of, but it did the job
> for a while.
> 
> Not too long ago, we got a brand new machine to do everything the old
> Precision did, and more.  The only thing the new machine couldn't do
> was nagios, because we had to put it on a restricted network, where it
> wouldn't have access to anything we'd want to check.
> 
> In between, some stuff happened.  Its a long story, involving YaST2, a
> major version upgrade, dependencies and the Knights Templar.  The
> short version is, I screwed up KDE to the point that it was completely
> broken.  But that's what ssh is for, anyway.
> 
> So this machine was running along, without having to do all that much,
> and everything was good.  The OS was somewhere in between OpenSuSE 9
> and 10.  It wasn't ideal, but it wasn't a problem that I planned on
> fixing.
> 
> Fast forward to last Friday, when it started acting really flaky.  By
> Sunday, it had to be rebooted.  It didn't seem to have booted after 3
> hours, but somehow, on Monday morning, it was running.  Slowly.
> 
> So, there's nothing too important on there.  We don't need the
> computer for anything, but if the hardware's OK, we'll find a use for
> it.  Does anyone have any suggestions on how to go about figuring out
> what's wrong?
> 
> I would like to proceed as if there were some critical service or data
> on there, and moving it off of that box was out of the question.  I am
> going to pretend that I have to figure out what is wrong with this
> machine without destroying its contents, just because it could be a
> useful exercise.  So, what would you do?

Check all logs.  Check RAM.  Check disks.  Make a temporary disk
and put it in while cleaning the others.

-- 
Tom Rhodes