Restoring entire system with shared vgs

[hydrocct]

I will be restoring entirely my 2 node Service Guard cluster from backups. The vg00s of each node will be restored from ignite tapes. As for the shared vgs, they will be restored from Netbackup tapes (similar to omniback). The plan is to bring up the cluster in maintenance mode and mount all shared vgs on one node. Before launching the restore of the shared vgs, I must erase the old contents to ensure no unwanted files remain (even if the restore is done with overwrite switch on). My question is how? By doing rm -r * from each shared logical volume mount point?

[Patrick Wallek]

The easier way would be to umount each LV, do a newfs on it, and then remount.

That would be MUCH MUCH faster than an 'rm -rf'.

[hydrocct]

Will newfs have any impact on mirroring?

[Prashant Zanwar_4]

Yes newfs will be a good thing in order to erase the entire contents of LV.

you can do it like below: Make sure you want to go ahead:

then:

while read FS | `mount -v | awk '{print $1}'|egrep -v "vg00|floppy|cdrom"`
do
echo $FS
fuser -ku $FS
if [$(fuser -u $fs) -eq ""]; then
umount $FS
newfs -F vxfs -o options $FS
done

You are going to plan outage surely..This is something I thought..hope this helps
Prashant

[A. Clay Stephenson]

Running newfs will have no impact on mirroring. Mirroring is done at the logical volume level and is thus completely invisible to the overlying filesystem. In fact, mirroring of raw volumes works as well -- with no filesystem at all.

[Prashant Zanwar_4]

As I believe it should be taking care of mirrored LV.

Do the newfs $FS && lvsync $FS

as in this case I am referring to lvol for the file system.

Hope it helps

Thanks
Prashant