When taking backups of your rman repository, you can safeguard the backup metadata (located in the controlfiles) by using the controlfile autobackup feature.
When enabled (configure command in rman) it will make automatically create backup of the controlfile after backing up the database and use a "predictable" generated name for it.
When restoring, you only need the dbid of the rman catalog to allow rman to restore it from backup.
Once the controlfile is restored, you have access again to the metadata of the backups of your recovery catalog and you can restore the rest of the database as usual.
Oracle Database Administrator
tel +32(0)3 451 23 82http://www.uptime.be
From: email@example.com On Behalf Of Storey, Robert (DCSO)
Sent: donderdag 20 oktober 2011 19:25
Subject: Backup best practice
Okay, looking at a best practices kind of question.
I am setting up a machine to hold the catalog/repository for RMAN
backups of my dbs. And, I even will schedule the backups of the backup
machine to be done by RMAN.
However, what is the best practice for storing this machines backup
metadata? It would seem that you have to store it in the controlfile
and use other methodologies to safeguard the information. To me, you
can't use the catalog because if it dies, how do you do a restore of
your backup machine if your metadata and such about the backup....is in
the dead machine?
A suggestion in my class to day was that you could still use RMAN to
restore the backup machine using the DUPLICATE command, as long as you
had access to the the proper backup set pieces.