The Recovery Process
- Find out what is damaged.
- Restore the damaged files from backup.
- Recover them by applying archived and online redo.
- Open the database.
Scenario 1: Lost a Non-System Datafile (Database Open)
A datafile of a user tablespace is deleted. Users get errors, but the database stays up.
Only the objects in that datafile were unavailable. The rest of the database kept running.
Scenario 2: Lost a SYSTEM Datafile or Many Files
The SYSTEM tablespace cannot be taken offline, so the database must be closed first.
Scenario 3: Lost All Control Files
SELECT dbid FROM v$database; and store it with your documentation. You need it when the control file is lost and RMAN has no records.
Scenario 4: Lost the SPFILE
Scenario 5: Lost the Whole Server
- Install the same Oracle version and patch level on the new server.
- Copy the backups and autobackup to it.
STARTUP NOMOUNTwith a minimal parameter file.RESTORE SPFILE, restart,RESTORE CONTROLFILE,RESTORE DATABASE,RECOVER DATABASE.ALTER DATABASE OPEN RESETLOGS.
What is RESETLOGS?
OPEN RESETLOGS creates a new redo history. It is required after recovery with a restored or recreated control file or after incomplete recovery. After it, take a new full backup immediately.
| Command | Meaning |
|---|---|
RESTORE | Bring files back from the backup |
RECOVER | Apply redo to catch up |
RMAN> LIST FAILURE; | Ask Oracle what it thinks is wrong |
Conclusion
Recovery is: find the damage, restore from backup, recover with redo, open. Keep your DBID, SPFILE and control file autobackups safe and rehearse often.

