Peter Nieberg wrote:
>
> Hello,
>
> I have a guestion about recovering an ORACLE-instance on a backup-server
> in case of the usual server being "disappeared" (fire, total system
> failure or whatever).
> We have ORACLEv7.1.3 and UNIXv 5.43B on a RM600-server. Every weekend we
> make a full backup on tape of the instance.
> Every hour we perform a "switch logfile" after which the created
> REDO-logfile is being dumped on tape by NSRORA (ORACLE-NETWORKER).
>
> In all the manuals is a lot of information on how to recover after a
> disk-crash fo after a file being corrupted. What i can't find is what to
> do when the ENTIRE SYSTEM is gone and the instance has to be rebuild on
> the backup-server. The problem is that i can't fool the controlfile by
> saying : "the instance is consistent, but we still want to go forward in
> time".
>
> For instance : On tuesday 13.30h the system is "gone". The last archived
> REDO-logfile on tape is from tuesday 11.00h. So we take our tapes of the
> full weekend-backup and the concerning REDO-tapes, we move to the
> location where our backup-server is placed and we start recovering the
> entire database from the full-backup-tapes. After this we have a
> consistent database, but it's not totally recovered until timestamp
> "tuesday 11.00h".
> We tried all the possible "recover database..." - statements, but there
> seems no way to tell the controlfile to go "back to the future". Or is
> there .......?????
>
> Thanks,
>
> Peter Nieberg
This is solved by the command:
RECOVER DATABASE UNTIL CANCEL USING BACKUP CONTROLFILE;
This way Oracle assumes infinite logs are needed and you cancel when the
last has been applied. Open with resetlogs specified.
Received on Tue Feb 11 1997 - 00:00:00 CST
