| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.misc -> Re: shutdown immediate or abort for cold-backup?
"Charles J. Fisher" <cfisher_at_rhadmin.org> wrote in message
news:Pine.LNX.4.33.0108211507480.29526-100000_at_galt.rhadmin.org...
> On Tue, 21 Aug 2001, Randy wrote:
>
> > I guess I would have less concern about shutdown abort if it forced the
> > checkpoint immediately before aborting the instance.
>
> Jason Couchman's 8i OCP text says the following on page 669:
>
> "When the database starts, all checkpoint sequence numbers in all
> datafiles, redo log files, and control files must match. If they do not,
> Oracle will not start, and you must perform media recovery on your
> database to get the files synchronized and in a consistent state."
>
> HJR says this is total horse-hockey and I should ignore it, but it
> implies that an abort during a checkpoint trashes the database
I wouldn't have said that Couchman's statement, as posted here, is horse-hockey (whatever that is), because it is totally correct. If the sequence numbers don't match, that will be because some form of file restore has happened, and that will require media recovery (I'm ignoring the permissible mismatches for read-only tablespaces).
But a shutdown abort does NOT result in inconsistent sequence numbers. An aborted database has beautifully consistent sequence numbers -because they were made so the last time a checkpoint was issued. The trouble with a shutdown abort is that there is no FINAL checkpoint issued, and hence there are transactions in the redo logs from a time AFTER the last checkpoint. They clearly need replaying -and that is *INSTANCE* recovery, not media recovery.
Shutdown aborts will always require instance recovery. They will not require media recovery.
Conceivably, it would be possible to abort whilst a checkpoint is actually being issued, and CKPT has only got round to timestamping half the data files... that would indeed require media recovery, but once again, since all requisite redo is available in the current redo log, it wouldn't be a big deal.
Regards
HJR
.
>
> Bearing this in mind, HAS ANYBODY EVER BEEN FORCED INTO MEDIA RECOVERY
> BECAUSE OF A SHUTDOWN ABORT?
>
> And, more importantly, DID THIS RECOVERY REQUIRE ANYTHING MORE THAN
> "RECOVER AUTOMATIC DATABASE" (IE, RESTORING OLD DATAFILES FROM A BACKUP)?
>
> Let's face it, folks, Oracle backups are a black art. What is stated in
> the books is nothing like what happens in the real world (and please don't
> get me started on this). While a lot of self-appointed "experts" will
> offer "authoritative" advice, the best work that happens in this group is
> trial and error, and the report of the findings.
>
> While Oracle remains closed-source, this is the best we can do.
>
> ------------------------------------------------------------------------
---Received on Fri Aug 24 2001 - 21:11:01 CDT
>
 |
 |