After a bit of nailbiting we realized our fault was backing up the
client box, and not the virtual address.  So, we made sure that all
three boxes in our cluster have the same Netbackup directory, and
.rhosts, and bp.conf, then we created new classes to backup the two NFS
shares based on the automounted virtual ip point, not the client
itself.  This worked like a charm.  Now it dosn't matter if the share
fails over because Netbackup sees the same filesystem that our other
clients see.

Thanks to all who responded.

Robert Westling