Re: Questions about applying archive logs

Richard Foote wrote:
>
> Hi Hub,
>
> Ahh well, let's try again ;)
>
> There's no way to automate this from an Oracle perspective as such.
>
> You need a mechanism to copy the archived redo logs across to the standby
> server (which you already do). You then need to have the standby (clone)
> database mounted (via the backed database files) and in recovery mode via a
> "recovery database until cancel using backup controlfile' command. It will
> then continually request redo logs to be applied. You then need a mechanism
> to apply these or execute this command automatically, again perhaps via a
> cron job.
>
> You can then 'cancel' recovery after the 'last' redo log and open the
> database with the resetlogs option when appropriate. I don't see how you can
> meaningfully open the database in read only mode and subsequently continue
> the recovery as resetlogs needs to be applied. Other than repeating the
> whole process with a fresh backup again. Perhaps someone else can help you
> with that one.
>
> Good Luck
>
> Richard
>
> "Hub" <hub_at_houston.rr.NOSPAM.com> wrote in message
> news:Ptzq9.127992$8o3.3785512_at_twister.austin.rr.com...
> >
> > Hi Richard,
> >
> > Thanks for the help.. However we're using Standard Edition, not EE.. so
> no
> > dataguard for us.
> >
> > So I'm back to my original main question.. What is the syntax/technique
> for
> > taking the latest archive and redo logs and applying against the cold
> backup
> > files I have on another server..and what is a good way to automatically
> > apply those logs as they come across.
> >
> > thx.. Allen
> >
> > ps I will read the dataguard stuff even if we can't use it..
> >
> > "Richard Foote" <richard.foote_at_bigpond.com> wrote in message
> > news:8zyq9.52488$g9.153346_at_newsfeeds.bigpond.com...
> > >
> > > "Hub" <hub_at_houston.rr.NOSPAM.com> wrote in message
> > > news:NEsq9.125210$8o3.3755846_at_twister.austin.rr.com...
> > > >
> > > > Howdy ya'll..
> > >
> > > Let me guess. You're from India ;)
> > >
> > > I have some archive log questions. Oracle 9i, linux,
> > > > standard edition.
> > > >
> > > > Ok our main prod server spits out archive logs. On a seperate standby
> > > > server I have copies of all the cold backup files.. tablespace, logs,
> > > > control, redo, etc.. Every 15 minutes I cron job rsync the archive
> logs
> > > to
> > > > the standby server. What should I do on the standby side to
> > automatically
> > > > apply these archive logs to the standby server as they arrive (or soon
> > > > thereafter)?
> > >
> > > OK. Recommendation 1. Lookup the Data Guard Concepts manual and see if
> > > Oracle's Data Guard (or Standby DB) is the way to go. Based on what
> you've
> > > requested to do, I would suggest it is as everything can be nicely
> > automated
> > > and protected.
> > >
> > > So you need to create the standby DB on your standby server (basically
> > > copying the database files across, a special standby control file that's
> > > initially created on the prod server and possibly creating a set of
> > standby
> > > redo log files). Then mount the database and put it in managed recovery
> > mode
> > > which will make the standby DB applied redo as and when it arrives.
> > >
> > > On the Prod DB, the log_archive_dest_n parameters have options that
> > enables
> > > Oracle to automatically transport the redo to the standby server via
> > either
> > > the ARCH or LGWR processes.
> > >
> > > > Then if for some reason we have to switch from the main prod
> > > > server to the standby one.. what is the appropriate technique/syntax
> to
> > > take
> > > > the latest archive/redo files, apply them to the standby box, bring
> the
> > > > database up on the standby server.. etc.. does the main prod server
> > have
> > > > to be down to do this or can it still be running?
> > >
> > > OK. Recommendation 2. Lookup the Data Guard Concepts manual and the
> > various
> > > options you have. You can either perform a graceful switchover which
> > enables
> > > you to swap the behaviours of the Prod/Standby database (and very easily
> > > back again) or you can simply activate the standby DB which converts it
> to
> > > the 'Mainman'.
> > >
> > > >
> > > > Also, how can i have this standby database be read-only, updated as
> the
> > > > latest archive logs arrive, used for a report server..
> > >
> > > OK. Recommendation 3. Lookup the Data Guard Concepts manual and see how
> > you
> > > can take the standby DB out of recovery mode and open it as read only.
> At
> > > this point, users can log on and execute their queries but of course the
> > > database is no longer being maintained as such. But that's fine, any
> > queries
> > > that only requires historical data and isn't too fussed about having
> > today's
> > > data will be as happy as Larry. Then you can kick everyone off and place
> > the
> > > standby db back in recovery mode. Note that a Temp LMT is crucial here
> but
> > > the Data Guard doco covers all of that.
> > >
> > > >
> > > > Also these archive logs, can I apply them to another database somehow?
> > > Just
> > > > pick an archive log file and say "try to apply all of these changes to
> > > this
> > > > database"?
> > >
> > > Yes. You can create a "clone" database (which you create from the
> backups
> > of
> > > the Prod DB) and use the archive logs created since the backup to get
> the
> > > clone as up to date as required. Note the database must be a clone
> > database,
> > > it can't just be any database for (I hope) obvious reasons.
> > >
> > > My last recommendation. Read up on Data Guard. There's a lot to Standby
> > DBs
> > > these days with many options that make what you appear to want to do
> very
> > > flexible and automated.
> > >
> > > Good Luck
> > >
> > > Richard
> > >
> > > >
> > > > thx a bunch.. Hub
> > > >
> > > > hub_at_houston.rr.NOSPAM.com
> > > >
> > > >
> > >
> > >
> >
> >

When you say no automated solution, is it the case that log_archive_dup_dest (or whatever the second archive dest parm in Std Edition is) cannot be a "SERVICE=" parm?

If it could, then can't the normal auto-transmission with the appropriate "managed recover" be done...

Cheers
Connor

-- 
==============================
Connor McDonald

http://www.oracledba.co.uk

"Some days you're the pigeon, some days you're the statue..."