I do clusters a little differently.

I set up one class per physical server and set the filelist to contain only
the physical
drives that server owns.

I set the cluster resource to allow failover one hour prior to the start of
the backup window.

I am multistreaming to DLT/SDLT and I do not like the way the
all_local_drives multistreams.

Since I am already manually editing the filelist, it was just easier to
backup via the physical node

name.

Greg Guenther