Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Incomplete Recovery
Norman Dunbar <ndunbar_at_lynxfinancialsystems.co.uk> wrote in message
news:F43E6BAE5BB5D411A44C00805FBE740D46FEAF_at_apps.lynx-fsc.co.uk...
> >> Not true -not entirely, anyway. Backing up cold after a shutdown
abort
> >>simply means that the database, when restored in a noarchivelog
scenario,
> >> will immediately need Instance Recovery.
>
> I agree if running noarchivelog, but for incomplete recovery the
> scenario is totally different and requires archivelog and don't recover
> the control files :o)
>
> >>I don't disagree, because one can never be too careful. But it's not
> >> strictly necessary.
>
> We were shown how to break an instance on the B&R course. Shutdown abort
> and do a recovery, works until you come to open the database. Bang -
> file 'n' requires more recovery. Bye bye data - at least back to the
> previous full (cold) backup. Its fun :o)
>
Not true, Norman. If you have all redo from the point of the last checkpoint onwards, then shutdown abort will pose no greater difficulty than any other form of shutdown.
When do you recover binary backups of control files, Norman? Er, never... not unless you've lost a tablespace in the meantime, anyway. Or unless you have a deathwish.
I can show you a dozen ways of breaking a *database* (I know of NO ways of breaking an *Instance* beyond recovery, since an Instance is merely a structure in memory that can be reconstructed ab initio time and time again. Maybe if they taught you the difference between a database and an instance, I might trust your comments a good deal more).
I only mention this because I have spent the afternoon doing incomplete recoveries on a database that was repeatedly shutdown 'abort'. Nary a one went wrong (which I agree is some kind of record for me and my demos).
Regards
HJR
> Regards
> HJR
>
>
>
>
> > HTH.
> >
> > Regards, Norman.
> >
> >