This isn't correct - it'll only cross NFS mountpoints if
"Follow NFS" is checked, regardless of "Cross Mountpoints".
We backup "/" and have "Cross Mountpoints" checked and
"Follow NFS" off. We have exclude lists for oracle database files,
typically just excluding any file name "*.dbf". Yeah, we get a couple of
database-related files this way but it's small.
Rather than run multiple backups per server, we run many
simultaneous backups across the servers (we have 266 unix boxes spread across
about 5 policies).
We're chosing to go wide this way. Otherwise, as
mentioned, you're thrashing your spindles and increasing, IMO, complexity since
a failed job has to be diagnosed to find out what filesystem
failed.
-M
I use all local drives for the unix
backups, and on small systems I limit how may jobs can run at a
time.
I have no issue with this at all and has
worked great for years.
But the Cross mount points – you should
think a little on that.
The way I understand that is that it will
backup NFS mounts. So if you backup the server the NFS comes from and the
server it is mounted on you are backing it up
twice.
If you use the all local drives and multi
streaming you will get 1 job per fs. Then just exclude the ones you
don’t/can’t backup like /tmp and /proc.
|