Technical Question

[durham.med.unc.edu">blanchar]

Please refer this to another mail list if this is
not the proper venue for such questions.

I am running SunSoft's Solaris 2.4 on a SPARCserver 20.
We do cpio backups to WANGdat drives.  I have tape
backups of several thousand home directories in two
directories which I did just prior to a disk failure. 
As good as this sounds, I cannot restore the entire 
filesystems from the tape.  Each tape cpio archive
quits with and error "Bad Magic Number/Header."

I had though that perhaps the new disk did not have
enough inodes (a problem with our home partitions...),
but I have packed as many inodes as possible in the
new partitions, but the errors are the same.

I have tried to restore from a tape cut several days
earlier, but we come to the same end at the same place
each time.  I have copied everything to another
SPARCserver 20 with different disks - same problems.

All this work has taken many hours, since one tape pass
takes over three hours.  I'm running low on sleep and
patience.  The logs of the backups show normal and
successful completion of the backups each time.

WHAT I REALLY NEED TO KNOW:  what does the error from
cpio "Bad Magic Number/Header." mean, what triggers
it??  We are using cpio as a transition from afio to
something more - ummm - fashionable - Solaris Networker
is what we are going to set up when we get through this
patch of disk problems...

In case my return address gets nibbled, it is as follows:

                  blancharmed.unc.edu

                           Thanks in advance,

                           Anne C. Blanchard

                           UNIX SysProg/SysAdmin
                           UNC School of Medicine
                           Office of Information Systems