On 6/3/05, Andy Ruhl <acruhl@gmail.com> wrote:
> On 6/3/05, Robert Zagarello <bzag0@yahoo.com> wrote:
> > Andy,
> >
> > Sounds like a page fault trap.  UVM is the BSD virtual
> > memory system, meaning it manages the swap file.
> > Among the things that can cause a page fault is a bad
> > sector on your swap partition, but a driver bug can
> > cause it as well.  At this point I would be talking
> > out of my hat if I go any further.  Perhaps others
> > have ideas about eliminating the possible error
> > sources.
>=20
> It happened again just when I was trying to read your message... And
> again. Same problem. One time it was saying postgres caused it, the
> other perl.
>=20
> And I lied about the kernel. I'll have to see what I did. I'm back to
> GENERIC now and I haven't had a problem (yet). But I know the main
> reason I recompiled was to set it up as postgresql recommends
> (semaphore stuff).
>=20
> Another thing, I was able to get sync to work and finish successfully
> as it said, then after reboot when the system was coming up, savecore
> said it didn't have anything. Maybe I'm not set up right...

Ok, I think I have ghosts in the machine, I doubt it's NetBSDs fault.
I got hit pretty hard by lightning a week ago. Nothing broke just
then, but the next day the motherboard in my amd64 machine broke. I
bought a new one and it's back in action.

The other machine that was panicing (my server, the subject of this
thread), I had never actually powered it down since it was panicing.
WHen I did, the root disk didn't show up, and I've had various
problems making it work on a totally separate motherboard now (this is
when using a mostly GENERIC kernel really pays off).

Oh well, likely not NetBSD's fault. It's my hardware. And it's ugly.

Andy