HTT/SMP Panic

[Michael Giegerich]

On Thu, Mar 03, 2005 at 08:15:10PM -0600, Stephen Montgomery-Smith wrote:
...
> As to the core dump - I am using a Stable 5.3 dated about Feb 19.  I did 
> a dumpon /dev/twed0s3b, which is 4G of swap space, since my system has 

Swap should be big enough to hold all of
your RAM plus the kernel; i.e. 4 GB plus
e.g. 8 MB for the kernel (should be
enough, but have a look at your kernel
size).

> 4G of memory (although it only sees 3G of it right now).  So perhaps 

Most 32 bit systems map drivers etc. within
the 2-4 GB RAM area. The system finding only
3 GB hints to that (or faulty RAM).
Do your XEONs have the 64EMT extensions?

Check your motherboard's bios for an option
to remap RAM above 4 GB. If not, you may be
out of luck.

You may also try to run the system with 2 GB
of RAM only (either to check above or find
faulty RAM).

> core dumps don't work with raid 0 systems?  I don't know.  savecore -f gives
> savecore: unable to force dump - bad magic
> savecore: no dumps found
> Maybe I am making some very obvious mistake, I don't know.

Have you enabled "dumpon" in /boot/loader.conf?
Needed for panics. Putting this only in rc.conf
isn't enough.
(Needs to be done w/ 4.x - have a look at man
loader.conf for 5.x).

One other thing (that drove me crazy at a time):
check your power supply (powerfull enough?) and
your wall plug (I had once one delivering unstable
power - took weeks to find out).

-- 
M. Giegerich, mail: migieger at web.de, phone: +33.(0)1.39692920