For the past several months, we have been plagued by the subject failure,
especially in backing up large UNIX boxes, but even on the Macs and PCs.
Our ADSM server is running under VM.  I've tried both levels 8 and 14
of the server, with no apparent difference in the failure rate.  The
problem has become bad enough that users are talking of finding another
backup system.  NOTHING points to the network as the culprit.  I also
can locate nothing that has changed that might have caused us to start
seeing this problem.  I suspect that it may be sensitive to the load on
the server.  We have 1500 clients.  (I have also been trying for
several months to split the server, but the server crashes so often
that I have been unable to EXPORT the nodes that I want to move.)

Advice on how to attack this problem?

Melinda Varian
Princeton University