(unknown charset) Re: Processes blocked on ufs or getblk

On Wed, Jan 28, 2004 at 07:22:37PM +0100, Dag-Erling Smrgrav wrote:
> Ken Smith <kensmith_at_cse.Buffalo.EDU> writes:
> > Is anyone having this problem on a machine that does NOT have the
> > disk(s) attached to an aac controller?
> 
> *blink* *blink*
> 
> You're having trouble with an aac getting stuck?  See if you get
> timeout messages on the console (it might take a long time though).

If "long time" is less than two or three days I'm not getting any
timeout messages to the console, I have left the machine in a
wedged state for two or three days at times.  When I was trying
to back the machine up it never lasted that long before starting
to have maxpipekvma issues but now that I'm not bothering to back
it up it hasn't been quite as grumpy once things start to wedge.

> I had to replace the Adaptec 2410SA in my backup server with a Promise
> SATA150 TX4 because it kept getting wedged.  I had a long email
> discussion with scottl_at_ about this but we never managed to track down
> the problem.  There is no doubt however that either the adapter or the
> aac driver was at fault.

Ok, this is still a possibility despite lack of timeout messages.  Is
there some way I could confirm one way or another that this is the
problem other than swapping out the controller?  Should I be able to
find some process somewhere on the system that's wedged inside of
the aac's device driver?  If yes is there a simple way to locate
which one it is?  So far the processes I can find wedged in ufs or
getblk have had no signs of the aac device driver in their stack
traces but I'm not checking every process on the system (yet :-).

Thanks.

-- 
						Ken Smith
- From there to here, from here to      |       kensmith_at_cse.buffalo.edu
  there, funny things are everywhere.   |
                      - Theodore Geisel |