Short version: is it possible in any way to change the VGID on a VG?
Long version: I am experimenting with using SAN snapshots to do backups of database LUNs.
Our nightly process requires 2 full cold backups of a given database, which resides on an EMC SAN.
The current process, for each cold, shuts down the database, does a full cold backup of it, then brings it back up.
The experimental process, for each cold, shuts down the database, performs a SAN-level snapshot, then brings the DB back up. The snapshot is then mounted to another machine to perform the actual disk-to-tape activity.
Both the DB server and my backup server are AIX.
The problem I am running into comes when I try to mount both snapshots to the backup server: since both are a snapshot of the original LUN, both have the same VGDA and therefore VGID. importvg works fine for the first, but when I add the second snapshot LUN, the PVs get added to the first imported VG.
So, something like a "generate new VGID" ability would be nice...
Yes I know the double-cold requirement is odd. Yes I know the trend is to move away from tape.
Long version: I am experimenting with using SAN snapshots to do backups of database LUNs.
Our nightly process requires 2 full cold backups of a given database, which resides on an EMC SAN.
The current process, for each cold, shuts down the database, does a full cold backup of it, then brings it back up.
The experimental process, for each cold, shuts down the database, performs a SAN-level snapshot, then brings the DB back up. The snapshot is then mounted to another machine to perform the actual disk-to-tape activity.
Both the DB server and my backup server are AIX.
The problem I am running into comes when I try to mount both snapshots to the backup server: since both are a snapshot of the original LUN, both have the same VGDA and therefore VGID. importvg works fine for the first, but when I add the second snapshot LUN, the PVs get added to the first imported VG.
So, something like a "generate new VGID" ability would be nice...
Yes I know the double-cold requirement is odd. Yes I know the trend is to move away from tape.
