Re: RMAN frustration

"Rogbaker" <Rogbaker_at_gdi.net> schrieb im Newsbeitrag news:e40a12ea.0301281207.6ffa2973_at_posting.google.com...
> I have been playing with RMAN for 3 days trying to learn how it works
> and I am having a real difficult time with it. I am trying to do a
> point-in-time recovery of an Oracle 9i database. I can recover, but
> not a point-in-time.
>
> I basically have a table scott.junk that I load with a bunch of data.
> I commit, then note the time, say 2:38pm.
>
> Then I do a RMAN backup with the following:
> connect catalog rmanuser/mypassword_at_rmnr
> connect target sys/systest_at_test
>
> run {
> allocate channel disk1 type disk format
> 'c:\rmanbakup\backup_012803';
> backup database;
> sql 'ALTER SYSTEM ARCHIVE LOG CURRENT':
> }
>
> this is giving me a successful backup.
> Then, I truncate my table and get rid of all the data.
> Now, my thinking is that if I restore to before the truncate time, I
> should get the data back.
>
> I shut down the database, and start it up in mount mode.
> to restore, I am using the following:
>
> run {
> allocate channel ch1 type disk;
> set until time '28-JAN-2003 14:38:00';
> restore database;
> recover database;
> }
>
> Then I get an error,
>
> datafile 1 must be restored from backup created before 28-Jan-2003
> 14:38:00
>
> I experiment with the times and I get other messages such as:
> ...no backup or copy of datafile XX found to restore...
>
> Can I only restore to times listed when I use the 'LIST BACKUP'
> command?
> I can only seem to restore to the latest state the database was in.
>
> Thanks for any help,
> Roger

The error message is absolutely correct, the earliest time-point for point-in-time
recovery is the completion time of the level 0 (full) backup. You can get the
completion time from the "LIST BACKUP" output and then You can use this time as the point-in-time. The time must be before Your truncation-time.
Look at the format attribute of Your example, if You write more than one backup with this format, You will destroy Your old backup.

--
Eberhard