Why do we have to flashback the failed primary DB to make it the new standby ?

Question

RDBMS Version: 11.2.0.3
Platform : RHEL 5.8
Standby Type : Physical Standby
Primary DB name        : berlin
Standby DB Unique name : lisbon
Lets say berlin DB goes down due to a hardware failure so lisbon is made the Primary DB. Now lisbon is running without a standby DB.
berlin crashed --------> at SCN 4000
lisbon takes over ------> from SCN 4000 and proceeds
In the below url, some steps involving flashback are mentioned on how to make the failed primary (berlin) the new physical standby.
http://docs.oracle.com/cd/E11882_01/server.112/e25608/scenarios.htm#i1050055
berlin DB crashed at SCN 4000. So, to make berline the new physical standby , can't we just run
SQL> ALTER DATABASE CONVERT TO PHYSICAL STANDBY;
and mount the DB and start the Redo apply ?
Why do we need to flashback the old primary DB to SCN 4000. Wasn't berlin DB already at SCN 4000 when it crashed?

Answer 1

Hello;
The key is the SCN must be at exactly the same point the Standby became Primary.
Generally you want to use Flashback so the "Old primary" can be at a point where it can become the new Standby. Once you are there you can use Switchover to return the Data Guard setup.
To get the correct SCN I use :
SELECT TO_CHAR(STANDBY_BECAME_PRIMARY_SCN) FAILOVER_SCN FROM V$DATABASE;
Example
SQL> FLASHBACK DATABASE TO SCN 1011295;
Flashback complete.
SQL>ALTER DATABASE CONVERT TO PHYSICAL STANDBY;
Database altered
If you don't have flashback you can use RMAN to perform the same thing.
Best Regards
mseberg