How to uncatalog a failed RMAN FULL Backup

[volker_borowski]

Hello,

if a brbackup RMAN FULL Backup fails, somewhere is tracked that the

attempt was made to do a FULL backup.

If you try to do a brbackup with INCR afterwards it fails.

BR0516E Last full database backup (level 0) bebovmuj.fnr 2009-09-30 16.11.21 was not successful

BR0520I Please start a full database backup (level 0) first

Instead of doing a full backup I'd like to tell brbackup to do an

INCR backup while taking the last successfull FULL backup as a base

and not the failed one.

I already did an uncatalog of the failed backup pieces using native rman

and erased the entry from backSID.log, but brbackup still keeps track

somewhere else and seems to be able to query other tables (SDBA* ?)

to know that the attempt "(level 0) bebovmuj.fnr" has been made.

Any Idea how to get rid of the information concerning the unsuccessfull attempt?

Thanks a lot

Volker

[Former Member · Accepted solution]

Dear Volker,

unfortunately, this is not so easy. Just allowing brbackup to run would involve quite some hacking effort in table SDBAD (and it's not done with just deleting a single entry). Apart from this, you would also have to ensure consistency with the other SDB* tables in order to convince brrestore from using the successful backup in case of problems...

To sum it up: I am afraid you would have to do another full backup.

Best Regards,

Michael

[volker_borowski]

Hello Michael,

this are bad news. My DB is some 30+ TB big and a FULL backup is running roughly 48- hours.

An INCR is usually ready in 6-8 hours.

For month closing we usually schedule a backup a week before months end.

This time we have been rather unlucky (defective tape on first and defective tape-drive on second)

for two attempts to do a full backup.

So this time, we ended up doing the closing without a backup, because we

would not have been able to finish it in time (allthough an INCR would have been possible).

So my desaster procdure would include 2 weeks of redologs

It is not, that I'd like to use this all the time, but if it is crucial to have a backup for a certain time,

this would be at least a second best option.

Since BRCONNECT takes care of consistent maintanance of the SDBA* tables in it's cleanup function,

it would be just a nice option to do a cleanup with the backup-id as a parameter to get rid of it.

best regards

Volker

[Former Member]

Dear Volker,

I agree that the current approach could use some flexibility...

If you open a customer message on the component BC-DB-ORA-DBA and reference this thread, we could check with development whether improvments are possible here. Your arguments are totally valid here esspecially since other tools (for example the Enterprise Manager) do not have a problem with such a situation.

Best Regards,

Michael

[LieveRavyts]

Dear Volker, Michael,

We are now 5 years later, is there now a possibility to uncatalog a failed RMAN FULL backup ?

Kind regards,

Lieve