Re: reinstate database without flashback/DB Restore/RMAN

From: Andrew Kerber <andrew.kerber_at_gmail.com>
Date: Thu, 4 Jun 2015 15:16:39 -0500
Message-ID: <CAJvnOJZTM+4vQNQpjSGMmz9Z+dz0jAAzvLsaza5MydLOUHYH3Q_at_mail.gmail.com>

I am referring to the original primary in that case. As I understand it, with flashback turned on, the reinstate command in dgmgrl will flash the original primary back to an earlier point, then apply the logs from the current primary to put the original primary into standby mode, ready for a clean switchover.

On Thu, Jun 4, 2015 at 2:57 PM, max scalf <oracle.blog3_at_gmail.com> wrote:

> Thanks Andrew, i did not realized that. Also looks like the log seq#
> resets back to 1 on the new primary(DC-B).
>
> Also when you mentioned "flashback allows you to flash the PRIMARY back to
> an instance before the divergence and start the apply"...are you talking in
> this situation the CURRENT PRIMARY (DC-B) which has been up for 4 hours now
> will be flashed back to noon followed by rollforward ?? Wouldnt that be on
> the standby ??
>
> On Thu, Jun 4, 2015 at 2:40 PM, Andrew Kerber <andrew.kerber_at_gmail.com>
> wrote:
>
>> Nope. At that point, if you have a done failover, the dbid on the
>> standby is different from the primary. In effect, at that point its a
>> completely different database. Flashback allows you to flash the primary
>> back to an instant before the divergence, and you can apply the logs from
>> that point, but you have to flash it back for that to work.
>>
>> On Thu, Jun 4, 2015 at 1:29 PM, max scalf <oracle.blog3_at_gmail.com> wrote:
>>
>>> Hello list,
>>>
>>> I am trying to get my head around reinstating a database in dataguard
>>> scenario, i am very new to DG and hence having issue trying to understand
>>> this.
>>>
>>> So lets say i have datacenter DC-A and Datacenter DC-B and i have setup
>>> DG between the two(Max Perf mode). Now lets say DC-A goes down
>>> unexpectedly (and lets assume that DC-A will come back up in 3-4 hours as
>>> know we know whats wrong and network team is trying to fix that), also lets
>>> assume this happened at noon and my log seq# at primary was 300. Obviously
>>> in the mean time i am going to FAILOVER to standby in DC-B and everything
>>> is happy. Time passes along and now its 4PM and lets assume now my primary
>>> in DC-B is at log seq# 320.
>>>
>>> Now at 4PM, my DC-A is back and i want to bring the former primary into
>>> standby mode. I can either use flashback(if enabled earlier) or recreate
>>> my standby again to do that. My question is, instead of using flashback or
>>> rman to recreate my standby(DC-A), cant i just mount my database in DC-A,
>>> copy over logs from seq# 300-320 from DC-B on to DC-A, roll forward DC-A
>>> and then start my redo apply instead of recreating standby(or using
>>> flashback)??
>>>
>>> I guess in short i am asking y do we have to use flashback or use
>>> RMAN/re-create to reinstate my standby database? is there any other way to
>>> reinstate my database(fromer primary).
>>>
>>
>>
>>
>> --
>> Andrew W. Kerber
>>
>> 'If at first you dont succeed, dont take up skydiving.'
>>
>
>

-- 
Andrew W. Kerber

'If at first you dont succeed, dont take up skydiving.'

--
http://www.freelists.org/webpage/oracle-l

Received on Thu Jun 04 2015 - 22:16:39 CEST