[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: system instablity



George Bonser wrote:
> 
> On Sun, 13 Dec 1998, Ed Cogburn wrote:
> 
> >       I realize this is a rare problem, because few if anyone is seeing it,
> > but its clear to me that its not a hardware problem, even though I
> > understand that you don't believe me, and you are not the only one to
> > tell me its not Debian software thats causing the problem.
> >
> >       It may be related to filesystem problems I had earlier, yet fsck says
> > that now my filesystem is fine.  I even ran a grep on all of my
> > filesystem files (except /proc & /dev) hoping to trigger an invalid
> > and/or broken file with errors, but that didn't find anything wrong
> > either.
> >
> 
> Ed, check your CPU fan/heatsink. Make sure your fan is not slowing or
> heatsink fins clogged with dust. You say you get the fault when using
> dselect/apt ... does it also fault when compiling a kernel? Some things
> are more CPU intensive than other and can result in the CPU getting hot.
> This can result is really weird crashes. I usually start to notice it on
> kernel builds when the heatsink fins start getting really packed with
> dust. People that do not do a lot of kernel builds might notice the
> problem later on a different program.
> 
> Also, if you have a computer such as a Gateway or AST that has only a CPU
> heatsink and a fan located some distance away that blows air over it, make
> sure that all covers are in place and all unused slots are covered. If air
> can "leak" before getting to the heatsink, you can loose cooling
> efficiency.
> 
> George Bonser


	Thanks George.  Well, I checked the things you suggested.  The CPU fan
and the case fan were not blocked by dust and both appeared to be
running at full speed.  There was a little dust on the tips of the fins,
but no fin was completely blocked or covered by dust.

	As for a kernel build, I'm doing a second one right now as I write this
in Netscape on X11. ... OK the second compile of 2.0.35 went off without
a hitch.

	[sigh] I really wish these errors would be repeatable.


-- 
Ed C.


Reply to: