At boot up several error messages (see attached dmesg) indicate "invalid
raid superblock magic on hdcx".  However, later is the process the
bind<hdcx,n> process works although it report "nonpersistent superblock"  I
am then able to mount /dev/md0 /backup.  While I suspect I have some sort
of configuration error, the fact that it works make me think that: 

1.  The autodetection of RAID  arrays does not expect to find nonpersistent
superblock, since no one in their right mind would be using a RAID0
configuration. <grin>
2.  The logic used in the "bind<>" process is more forgiving.

This causes another problem when trying to do an Upgrade install of RH
7.0.  Anaconda finds the bad superblock. (see attached anaconda dump).

[I'm new to using bugzilla and have not figured out how to do attachments
to this form, it seems to me that cut-n-paste is not the way.  I'll try
commit and see if it leads to an attachment selection option.]