Sudden thought - is your solaris server multihomed?

		David/absolute

	"Its just another ultimate battle of good against evil..."

On Wed, 10 Mar 1999, Urban Boquist wrote:

> >>>>> Frank van der Linden writes:
> 
> Frank> There is a problem with amd, because it uses TCP for mounts,
> 
> Maybe, but I think there must also be some other problem with amd. At
> our site we have severe amd problems and we are not using TCP
> mounts.
> 
> We have a -current i386 machine that automounts home directories from
> a few Solaris servers (except for two home directories that it mounts
> from itself). This machine will reproducibly hang within a day or so
> if we try to use the amd in -current. Downgrading to amd-6.0a13 (from
> early August 1998) fixes the problem (it has worked for two weeks
> now).
> 
> Strangely enough we had exactly the same problem in August, after
> Christos had upgraded amd. We then found that if we forced the new amd
> to use UDP with "proto=udp" in the amd mount options, then it would
> work just fine.
> 
> Everything was ok until a while ago when we took this machine to
> -current again (it does not track -current all the time). The udp
> workaround above did no longer help, and the machine hung every day
> with "nfs server dogbert not responding" (which incidentally is the
> name of the machine itself).
> 
> Currently we have to revert back to the August amd to make the machine
> usable at all.
> 
> See also the old PR bin/6037.
> 
>         -- Urban
>