Re: Another twist on WRITE_DMA issues <- ProblemFound

At 11:32 AM +0100 12/7/04, Alexander Leidinger wrote:
>Zitat von Garance A Drosihn <drosih_at_rpi.edu>:
>
>  > Maybe it's an overheating issue, or maybe it's something else.
>  > But whatever it is, I am going to assume it's the fault of
>  > something in my PC.
>
>I had a similar problem... try to monitor the temperature of the
>disks with smartmontools. If they are below 50 degree celcius you
>may be on the safe side, but if the temperature grows over 55
>degrees I suggest to install some additional fans to cool the
>drives.

Well, I have a dual-boot system.  One system is a very recent
snapshot of -stable, and it is getting no errors.  The second
system was a copy of a snapshot I had built in late October.  It
was the one getting lots of WRITE_DMA errors.  After my previous
message in this thread, I copied the kernel from the first system
to the second, and was able to do an up-to-the-minute build of
6.x-current on it.  I installed that.  I then started up a sequence
of five consecutive buildworlds, closed up my office and went home
for the night.  My office always warms up a few degrees when the
door is closed for a few hours.  The PC finished all five of those
buildworlds without a single warning or error.  So, now I'm thinking
(hoping?) that at least some of my problems have been fixed by system
changes between late October and now.

Of course, now that I've said that, it will probably start giving
errors again...

One other long-shot guess on my problems.  After thinking about it
awhile, it seemed that I was much much more likely to have problems
when I was working on the serial console.  Also, this PC is the first
time that I have set up a serial console on FreeBSD.  So, the other
thing I did last night (before I did the first build of 6.x) was to
get rid of the serial-console setup.  This is one step short of using
tea-leaves to guess at my problem, but it seemed worth a try.  If the
stupid PC remains reliable under load for a few weeks, then I will
switch the serial console back on and see if my problems return.

I realize my reports so far have been pretty useless to Søren, but
they are all I have been able come up with...  Sorry about that.

-- 
Garance Alistair Drosehn            =   gad_at_gilead.netel.rpi.edu
Senior Systems Programmer           or  gad_at_freebsd.org
Rensselaer Polytechnic Institute    or  drosih_at_rpi.edu