Re: 9i Standby Database & Production Online Redo

From: Pete's <empete2000_at_yahoo.com>
Date: 17 Sep 2003 06:39:41 -0700
Message-ID: <6724a51f.0309170539.77566020@posting.google.com>

tomk_at_projectx.com.au (Tom K) wrote in message news:<b6b81b29.0309162032.7df0dfd5_at_posting.google.com>...
> I have a quick question on Physical Standby Databases (the 9i
> version). I have configured a physical standby database (STDBY) in
> Maximum Performance mode. All the archive logs are being transported
> from production (PROD) to STDBY.
>
> In the event of a catastrophic failure on PROD, given that I can save
> the online redo logs from PROD, and I copy them to the STDBY node, how
> can I "feed" then into STDBY?
>
> I have tried to register the redo by renaming them to arc_985.log, but
> this has not worked.
>
> Any ideas? Has anyone been able to do this?
>
> Cheers
> Tom K

All you should have to do is just copy them over to the standby and perform an 'Recover Standby database until cancel;'(Don't recall the exact syntax. Do this before opening the standby for write, if you open it for write before you do this, the archives will be useless. Also, instead of maximum performance, have you thought about 'Instant Protection'? I use it with LGWR transmitting the changes on a 24x7 DB and really do not see much of a performance hit, about 20 to 40 connections at any one time. I use instant and not guaranteed for several reasons, one is the scenario you are testing. With instant, I do not have to transport any archive logs to the standby as long as they are in sync.

A couple more notes with instant and guaranteed protection. With guaranteed, you have to have at least one standby available at all times, otherwise, the production/primary db will stop until one becomes available. With instant protection, the primary db will not stop if your only standby becomes unavailable for whatever reason, for instance, making a backup. However, when you bring up the standby, you either have to perform a manual recovery to resync it or set the REOPEN=36000 or set the time high enough(on the primary) for archives to get re-transmitted to the standby. Here's the procedure I use to bring my standby up and up to date after a backup:

1.  start standby into mount stage
2.  recover managed standby database disconnect;
3.  Now on the primary db
4.  alter system archive log stop;
5.  alter system archive log start;
6.  alter system switch logfile;-- you may need to do it a couple more

times.
The last 3 steps will basicall tell the primary to re-transmitting logs that are up to 36000 secs old. Before each of the last 3 steps, perform a select * on v$archive_dest and you'll see what I mean.

HTH,
Pete's

The opinions stated above are mine and not that of my employer. Use the advice at your own risk. Received on Wed Sep 17 2003 - 08:39:41 CDT