🚀 go-pugleaf

Here is a brief description of our local system:
	4 3/280's acting as file servers
	60 3/50's distributed ~15 per server
	~2 MB user files distributed among the servers, and NFS mounted by
		*all* servers.

The problem:

If one of the servers goes down, it effectively stops all of the others
from working reasonably.  The problem seems to be that NFS requests queue
up for the dead disks, and are never served - and then legitimate requests
get stuck behind the "bad" requests.  The main cuplrit (we think) is the
call to quota made in the login sequence.  Thus we can't startup a new
shell on a 3/50 or anywhere else.

The questions:

Is anyone else running a system this large and interconnected?  If so, do
you have these problems?  Is there something we can do to make these
things reasonably independent if one goes down, but still maintain the
homogeneity (*all* usr partitions mounted by *all* servers, etc.) on the
rest of them?

Please e-mail and I will summarize if there is any response.

Thanks in advance...
-- 
David Bernholdt			bernhold@qtp.ufl.edu
Quantum Theory Project		bernhold@ufpine.bitnet
University of Florida
Gainesville, FL  32611		904/392 6365