current + mpt = panic: Bad link elm 0xffffff80002d6480 next->prev != elm

[Ståle Kristoffersen]

On 2010-07-20 at 14:16, Svein Skogen (Listmail account) wrote:
> Sorry for the late response here, but what you're describing matches
> fairly well what I saw with RELENG_8 (just after 8.0 was released), but
> luckily I didn't have any disks on my MPT, just my tape autoloader.
> 
> Random timeouts, and then bus resets (that made tape IO unreliable).
> 
> The bad news, is that I had the exact same trouble with OpenSolaris
> (134), and something-similar with Linux (can't remember versions), at
> the time.
> 
> I never did find a solution, and ended up throwing windows on the box,
> just to get reliable backups.
> 
> My MPT is a 3801 LSI1068e based card running the latest bios.

Hmm, that does not sound good. Did windows work on the same hardware
without problems?

I -might- have solved my problem. It has now ran for 24h without timeouts,
and with a bit of load on it. I think I might have ran into the seagate +
NCQ-problem, even tho seagate's webpage told me my drives was not affected
(according to the serial numbers). I did however update the following
num drives       firmware 
6x  ST31000340AS SD15
4x  ST31500341AS SD17

to firmware SD1B (old SD17) and SD1A (old SD15), and that looks like it has
done the trick. I'll report back in a week or so if the problem has not
reappeared.

-- 
Ståle Kristoffersen