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: Restoring & recovering in NOARCHIVELOG

Re: Restoring & recovering in NOARCHIVELOG

From: Mark D Powell <Mark.Powell_at_eds.com>
Date: 25 Jul 2003 06:05:45 -0700
Message-ID: <2687bb95.0307250505.194c130e@posting.google.com>


miotromailcarlos_at_netscape.net (Carlos) wrote in message news:<1de5ebe7.0307250226.22aa87e7_at_posting.google.com>...
> Hi all.
>
> We are working with Oracle 8.1.6 on Win NT in a production environment
> (don't blame me, It's heritage). The Database archiving mode is
> NOARCHIVELOG (again don't blame me). We do whole database backups
> daily al the end of the day (It's not a 24/7 DB). Due to requisites,
> we must be able to restore the status that the DB had at the beginning
> of the day (start) in order to reprocess (by external processes) all
> the activity of the day if necessary (this is easily done by restoring
> the last-night backup). But I'm wondering about recovering the DB if a
> HW problem occurs. Here is where my question comes out: If we use redo
> log files large enough to hold all the redo activities of a whole day
> we could not only restore the 'last-night cold backup' in order to
> reprocess, but recover the DB status to the instant before a HW
> failure as well by using the redo log files to roll forward.
> Am I right?
>
> Thanks in advance.
>
> Carlos.

Carlos, no, you cannot roll forward in noarchivelog mode. In the event of a media failure that results in the loss of any db data file you will have to recover all the database data files and open the database resetting the logs. This means there is nothing in the online redo logs to apply.

If you switch the database to archivelog mode then yes, you can forward recovery the database using the archived redo logs. Should the online log be large enough to hold a full day's activity then Oracle will not need to read any archived logs. However, it would probably be better to make the online logs large enough to only hold a couple of hours activity so that in the event of a crash the time to recover is less. That is, on instance crash recovery, the online log file would have to read and reapplied as necessary. The bigger the log the longer it takes to read it.

I would suggest you retain at least enough archived redo logs to go back two backups. This way if the last backup was corrupted or otherwise damaged you would have the ability to go back to the prior backup set and forward recover to the present.

HTH -- Mark D Powell -- Received on Fri Jul 25 2003 - 08:05:45 CDT

Original text of this message

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