Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Shutdown Abort and Backup

Re: Shutdown Abort and Backup

From: Vinay Joshi <vjoshi_at_pinnacle.co.uk>
Date: 1998/03/02
Message-ID: <717BDE76D864D011870F00A0C9283F1E14A14E@exchange.pinnacle.co.uk>#1/1

My opinion:
When you do a clean shutdown,
a checkpoint occurs which results in
flushing the dirty buffers to the datafiles. This then results in synchronisation of all the file headers.

Where as in the shutdown abort situation, a checkpoint does not occur,
the dirty buffers donot get flushed out to the datafiles.

I am a bit doubtfull about logfiles however.

These are just my opinions/understanding. Feel free to correct me if I am wrong.
> -----Original Message-----
> From: Martin Haltmayer [SMTP:Martin_Haltmayer_at_KirchGruppe.de]
> Posted At: 02 March 1998 14:30
> Posted To: server
> Conversation: Shutdown Abort and Backup
> Subject: Re: Shutdown Abort and Backup
>
> Oracle do not insist on a clean backup. If you run your database in
> archivelog
> mode, you may also do a hot backup which is nearly never a clean
> backup.
>
> For cold backups I do not have an idea by now, perhaps later.
>
> Martin Haltmayer
>
> Doug Burns wrote:
> >
> > After years I still don't have a good technical answer to the
 following ....
> >
> > Assumptions
> > -----------------
> > 1. In the event of a power failure, the information in the online
 redo log
> > files
> > will allow automatic instance recovery to take place
> >
> > 2. Likewise, in the event of a SHUTDOWN ABORT, smon will perform
 instance
> > recovery on STARTUP
> >
> > (These two assumptions are based on every example I have seen, but I
 would
> > be interested in any experiences which contradict my own)
> >
> > Question
> > -------------
> > So, why do Oracle insist on a clean shutdown before a backup. Won't
 a backup
> > taken after a shutdown abort contain files in the same state as a
 database
> > after a power failure? Are Oracle suggesting that a database might
 not be
> > re-startable after a SHUTDOWN ABORT?
> >
> > I'm asking this question because I have never had a convincing
 answer before
> > now
> > and because there seems to be a flaw in the logic of the gospel ;-)
 I would
> > really love
> > to get a decent debate going on this and *all* contributions are
 welcome.
> > --
> > Doug Burns
> >
> > Certified Oracle DBA
Received on Mon Mar 02 1998 - 00:00:00 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US