[Techtalk] mysterious system halts - how to prevent/fix/detect?

[Julie Meloni]

JD> julie at i2ii.com said:
>>  have a SuSE-based web/db/mail server that has been alive and running
>> brilliantly for 10 months.  In that time, I rebooted it once, for no
>> reason that I recollect now.

>> However, in the last 3 days, it has halted itself for no particular
>> reason:

JD> What have you changed recently?

nothing.  swear to god!  haven't cracked the box since I bolted it in
the rack.  About 2 or 3 months ago I upgraded to the latest (at the
time) versions of sendmail and bind, but that's it.  It's probably
time to check versions again on everything, but I have another machine
right above it running the same versions of everything.

JD> Different workload?

Not really - access stats for the services, and bandwidth usage has
been the pretty darn similar the last three months.

JD> BTW, what do you mean by halting?  Hung, as in windows and everything are
JD> there, but the mouse doesn't move, or halted, like you ran 'halt'.

Halting meaning when I ping it remotely, see it isn't up, drive really
fast to the colo and plug in a monitor and keyboard, the screen is
blank.  Normally, the login prompt is sitting there, but in this case,
just a blank screen.

>> I'm pretty much terrified by kernels, so I haven't touched it.

JD> 2.2.16 is a bit old.  It would probably be a good idea to upgrade it,
JD> if for no other reason than to get over your fear of kernels...

Yes, a good point.  I will study hard and do so.

thanks,
julie