[TriLUG] diagnostic/stress test software

[Steve Litt]

On Thursday 07 June 2007 13:02, Jason Watts wrote:
> Gents,
>
> we have a linux server that keeps has hung twice on boot this month (its
> restarted 3times a week)
>
> we are looking for a self booting diagnostic tool that will do both
> diagnostic and stress testing.
>
> we have found one "
> http://www.smithmicro.com/default.tpl?group=product_full&sku=CKDWINEE&prodv
>iew=intro" , but i wanted to see if you guys had any other recomendations?
>
> jsn




If I had to fix this box, I'd try booting til it hangs, and see whether it 
completes memory counting or not. If it doesn't complete counting memory, 
it's not the OS.

I'd wait until I could reproduce it within 10 reboots. I would see whether it 
counts memory on boot, when it hangs. If it doesn't complete counting of 
memory, I'd do this:

I'd disconnect every peripheral from the motherboard, with only video card, 
RAM and video card still connected, and boot it about 50 times to see if it 
counts memory (obviously with the hard disks disconnected no OS will run). If 
it does not hang I'd start putting back peripherals, booting many times for 
each peripheral, until it hangs again. I'd keep a written log of the 
reconnections and boot attempts. Obviously, when it hangs again, suspect the 
last reconnected periperhal. If, after all periperhals were reconnected, it 
still doesn't hang even with 50 boots, then assume it was a bad connection, 
use electronic lubricant:

http://www.troubleshooters.com/tpromag/200310/200310.htm

If, with all peripherals disconnected, it still hangs, then it's either 
memory, video card, processor or mobo. Swap the video card. Try booting with 
one stick at a time and vary the stick. If it's the mobo or CPU, replace em.

If when hanging it always completes memory counting, throw in a different disk 
with a tiny OS, and ascertain that it will boot 50 times out of 50 times. If 
so, you've isolated it to your hard disk or your OS.

HTH

SteveT

Steve Litt
Author: Universal Troubleshooting Process books and courseware
http://www.troubleshooters.com/