> Yes, I can; but it's somewhat annoying. I gather from your astonishment
> that nobody else is having this problem?

I haven't been paying a whole lot of attention, but I haven't noticed any.

> Um, it's a production server. I'm I safe? ;-)

If you know what's coming...  ;-)
I think you're pretty safe with the current state of things, though.

[ sn0: buffer problems ]
> It could be the traffic generated by Retrospect; on the other hand I
> don't see an unusually high packet count with netstat.

It's definitely traffic-related.  I have been seeing this with my A/UX
interrupt experiments, and it's about the only thing keeping me from
committing those.  I can reliably reproduce the error on my Q700 (with
the A/UX interrupts enabled) by using "ping -s 32760 <remote host>".

The errors are actually coming from status bits on the card.  We do need
to handle them better.  I'm not sure what the solution is or why we're
seeing them (or why I'm seeing them more when I should be handling
interrupts better).

Unfortunately, I'm probably not going to have time to look at this any
more for a while.  :-(  I hope someone else finds something before I can
look at it...

-allen