Re: Linux 2.6.8 SegFaulting...
From: Frederik Dannemare
Date: Sat Sep 04 2004 - 08:49:38 EST
-----BEGIN PGP SIGNED MESSAGE-----
On Friday 03 September 2004 07:08, Richard Whittaker wrote:
> We have a Proliant DL360 G3 running Linux 2.6.8. This machine has two
> CPUs, 3GB of memory, a pair of 146GB disks on the SmartArray 5i
> contoller, a QL2314 Fibrechannel card, and is using abou 60GB on an
> HP VA7400 via the Fibrechannel card. Under 2.6.6 and 1GB of memory
> this machine had run for 4 months without even a hiccup... As of
> yesterday, I added the space on the VA, and moved our MRTG monitoring
> system to the machine. Our MRTG currently monitors about 300
> different elements, and is pretty I/O intensive. The website that the
> MRTG graphs are being written to is on the VA.
> This machine crashed and burned in spectacular fashion last night for
> the first time in 4 months, and I was really suprised. The machine
> SegFaulted, but I couldn't get a capture of the stack spew... I had
> to power the machine off, and restart it this morning. The machine
> ran for about 4 hours, then SegFaulted again... I was able to at
> least see the Segfault info, and it was saying something about
> swapper. Power cycled, and about 5 hours later, Segfaulted again, but
> this time I was able to capture the DMESG output... I thought it
> might have been the software RAID device I'd made out of the two LUNS
> on the VA, but after wiping the LUNS out, and rebuilding them, and
> moving to LVM with striping the problem still persists.
> Everything in this Kernel is statically linked, and everything is
> stock, right from ftp.kernel.org...
> Now, I have a feeling that this has something to do with the I/O that
> I'm loading on the QLogic controller, but can't be absolutely
> certain... I pose the question here for perusal by the experts, and
> hopefully some suggestions about what I should look at, or do next...
Can you rule out bad memory (try memtest86+ for 10-20 hours)? Or could
it be heat? Bad mem and heat problems have been the source of almost
all crashes I've witnessed.
Frederik Dannemare | mailto:frederik@xxxxxxxxxxxxx
http://frederik.dannemare.net | http://www.linuxworlddomination.dk
Key fingerprint: BB7B 078A 0DBF 7663 180A F84A 2D25 FAD5 9C4E B5A8
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
-----END PGP SIGNATURE-----
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/