Diagnosing a Linux Crash

[walla]

I've been using Linux for about a month. (SUSE 9.1, x86_64, KDE). I've had more than a fair number of crashes, which I'm willing to blame on configuration cluelessness at this point.

What is the best way to diagnose a crash? Usually when my Linux crashes, it hard locks and I have to resort to flipping the power supply off (which I hate doing but see no other way...).

I dig around in the "/var/log/messages" usually. I then look for where the "restarts" occur and look at the preceeding few messages. Often, it contains nothing useful or legible as far as I can tell.

Is there a better way to diagnose a (random) crash in Linux.

And please say there is a way I can make my power button shut the computer off, and not the power supply switch. 🙂

 

[Emulex]

1. run setiathome.
2. Run memtest and suck up most of your memory.
3. get the kernel sources and run a continous make all on the kernel, if you see any segmentation faults
your box is unstable.

This is about how i burn in most all of my pc's these days. Cept i use freebsd and make buildworld over and over. Usually about 1 days worth without error and the above apps tearing up the memory/fpu along with a script to constantly tar up the /usr partition (4 at once tar cpf).

this gives me:
ram test sorta, heats up the ram
seti ensures 100% cpu on all cpu's (HT too)
kernel is a good way to eat up cpu and disk io to the max and any fails are a sign of a weak component.
tar is a good way to task the disk subsystem (raid/pata/scsi/etc)

Works great.

now i got a 478 p4-2.0ghz that wont run stock in the above anymore just flakes out, any ideas what i can do? Cant tell if its mobo or cpu, everything else swapped and it still does it.

🙁