Re: Backup Architecture question

Jeremy Billones wrote:
> OK, this one's pretty ugly. Let me start with some background.
>
> Oracle 9.2.0.5 server. One data instance, one repository instance.
> Database is up 24/7, with hot backups nightly run from OMS. Nightly,
> the OMS/agent/repository is shut down automatically, and a tape backup
> of the /opt partition is made. If the disk fries, we can restore the
> partition from the backups, start the repository instance (& agent & OMS),
> then restore the main database from the hot backups.
>
> That all works, we think. (System is still in development, but passed
> the surface testing I did.)
>
> Ah, but they have a new, cunning plan. In the event of an Oracle patch,
> they want to ship out a new hard drive with the patched $ORACLE_HOME,
> then restore the database from the backups. Of course, all they
> want restored from the backup is the actual database, not anything
> that might have been superceded by the patch.
>
> And we now hit the wall of my Oracle knowledge. The reason for the
> cold backup of the repository is that I couldn't get it to run after
> a restore taken while it was running, and I didn't know how to boot-
> strap it.
>
> What is the set of files that have to be restored from tape in order
> to permit the reconstruction of the repository database w/o impacting
> any files that may have been patched? (If I can get the repository
> up & running, I believe all I need is the most recent backup set made
> by the OMS to reconstitute the data instance and I'm back in business,
> right?)
>
> Or is that simply such a Wrong Way To Do It that I have to present them
> with a Plan B?

And why was using DataGuard and/or RMAN rejected?

-- 
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)