On Fri, Jan 07, 2000 at 12:33:14AM +1100, Keith Owens wrote:
> On 06 Jan 2000 14:23:59 +0100,
> Arnaud Gomes-do-Vale <[EMAIL PROTECTED]> wrote:
> >The box was completely
> >frozen, no disk activity, no num lock, no alt+sysrq
> >combo. Unfortunately, I have no trace of oops or anything like that ;
> >however, as I was looking at the console, I guess I would have seen
> >one if it had occured :-)
>
> Not necessarily, if syslogd was frozen then you would not see any
> output, even if an oops occurred. A serial console will often capture
> the oops, even if syslog is dead.
>
> >Hardware is dual celeron 433 on an abit bp6 mb
>
> A lot of people are seeing hangs on bp6. So much so that there is a
> separate mailing list, email [EMAIL PROTECTED], "subscribe
> linux-abit". Unfortunately we have no handle on this problem yet, only
> guesses.
I'll take a stab in the dark here and place a $20 bet (just so you can
pay me) that the problem's related to the BP6 in some way. I've got a BP6
board here, also with dual Celerons (OT: is there anyone who has a BP6 who
doesn't run in dual?) and it locks up extremely occasionally. (Three times
in the past two months?). No panic message, no oops; just a complete
freeze.
Unfortunately, I can't trace the locks at all - they seem to occur in
almost random conditions - one time I was in X, running about 10 copies of
lame (mp3 encoder) in the background, another time I had no interactive
programs running and it froze up in zsh while I was typing, another time
when I was running only mutt and playing a .wav. I *think* that the problem
*may* be related to the HPT366 driver, but that's only a very, very wild
guess.
I'm currently running 2.2.13 with SMP, reiserfs, Andre Hedrick's
2.2.13-IDE patch and raid 0.90. The system shows no signs of abnormal
behaviour apart from this seemingly random lockup. If I get another one of
these (I haven't had one in over 2 weeks), I might just consider figuring
out how to use ksymoops :).
One thing which you might like to try is lucifer - search for 'lucifer'
on freshmeat. It's a stress-tester designed to test CPUs which are
overclocked.. I ran a little script which ran 10 copies of lucifer (load avg
of 16! boing) and got an oops a few minutes later. I put my bus rate down
from 78 to 75 and it was fine running after that 8) (Ran the whole night).
It might be worthwhile to try it and see if your system dies under that
test.
--
: Andre Pang <[EMAIL PROTECTED]> - Purruna Pty Ltd - ph# 0411.882299 :
: #ozone - http://www.vjolnir.org/ozone/ :
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
