Ok so the qa environment we've been fooling around with rman for testing backup/recovery.
 
tonite they say " we want a backup restored from rman from 2 incarnations ago(for those of u who dont know what an incarnation is, its a new version of the database that gets created when you do open resetlogs <-- if thats wrong please RMAN gurus correct me.
 
So I get out the book, the docs say "in the event you have to do this, which should be rare", you must do the
"reset database to incarnation" command.
 
seems easy enough,   fire up rman, startup nomount, reset incarnation and it finds the old backups, and starts the restore.
 
so far so good,  but then the catch, being a rman newbie(8.1.7), i forgot to make a controlfile backup, now if i'd had that, this should have been cake,restore the control file in nomount mode, mount and restore the datafiles, recover the datafiles and open resetlogs, did i forget to say, this is a NOARCHIVELOGMODE database.
 
Well, for whatever reason,  the database wouldnt open, due to using a newer controlfile.
 
hmmm, i've done this before, i'll just dump the controlfile to trace(yes i did this BEFORE the restore, just in case of an actual emergency).
 
rebuild the control file, try the recover again(did i say we're going to a particular SCN based on the info from list backup/list incarnation commands), no deal, damn database will not open.
 
regroup,  wait a minute, rman aint anything special,  let's do this:
 
restore the files from rman backup again.
exit rman
sqlplus:  recover database until scn #######;
alter database open resetlogs;
 
rman: <check>  resync catalog;
 
CHOKE, but i expected that,
rman: reset database;
 
full catalog sync AND
 
we're good to go.
 
Where did go wrong in using RMAN to do the recover/restore ???
 
joe
 

Reply via email to