[Techtalk] machine hanging
B Seeger
seeger at prosensing.com
Thu Dec 13 17:14:06 UTC 2007
Hi,
After spending most of yesterday on this, I think I've figured out what
it is. And I sincerely hope we don't have another setup like this
anytime soon. :)
The machine had a SCSI hard drive operating through a PCI card. I
suspected that this might be part of the problem, so after testing the
memory, I pulled out the drive and used an ATA hardrive, which hooked
directly into the motherboard. Except for the ethernet, things were
much more stable (as the harddrive had a few versions of 2.6 linux on
it, I figured the problem with the ethernet was a driver issue.). Today
I swapped in another ATA drive that had a 2.4.25 kernel and the machine
acts like it never had a problem and the ethernet is fine.
Unfortunately I haven't gotten the cdrom drive to work, so I could never
boot in to knoppix to diagnose this more. Another thing to fix, but the
main problem is gone and I have a few cdrom drives I can swap in, if
necessary.
So, somewhere with the SCSI there's a hardware issue or maybe an IRQ
problem/misconfiguration. Unfortunately I'm out of time to play with
it.
One thing solved... :) Thanks again for all the advice received.
Thankfully I don't have to repartition anything and now have a fairly
stable system.
Thanks,
Bethany
> This even sounds to me like an old 'CPU fan error' : the processor's
> fan doesn't cool the heatsink enough and the CPU 'freezes' when the
> temperature high watermark is reached.
>
> [...]
>
> 'top' output seems quite correct for such a machine.
>
> > Thanks for your help so far - I feel like I've been set down on a
> > strange planet with this machine. (I'm used to much newer servers, so
> > this is probably a good learning lesson for me - okay, so this isn't
> > *that* old, but still...).
>
> No problem, we're here to help ;)
>
> Now, my diagnostic would be : hardware failure.
> I'd say this machine never was stable, be it because of a motherboard
> flaw or some on-board chip not being recognized and crashing
> unexpectedly.
> Usually, the linux kernel writes some output on the console on a
> software crash or on a hardware triggered kernel crash. That may lead
> you in some direction.
More information about the Techtalk
mailing list