Databases Reference
In-Depth Information
Failover to a Standby Database
Since the primary database can't use flashback because all of its control files are gone, you must restore and recover.
Failing over to a standby database incurs the least downtime of all three methods we discuss here. You can flash back
to the failover SCN if the primary database is intact, this being the SCN at which an old standby database became
the new primary database. Once you flash back, the standby will catch up automatically with the redo from the new
primary database.
■
if the old primary database isn't intact, you need to re-create it from a backup of the new primary database and
bring it back as a new standby database.
Note
Create a New Control File
You can execute the following SQL statement on the standby database to generate a trace file that will help you
re-create the control file on the primary database:
SQL> alter database backup controlfile to trace noresetlogs;
Once you generate the trace file, edit the SQL script contained in the trace file and run it on the primary database,
after starting the database in the nomount state, as shown here:
SQL> startup nomount
SQL> @create_ctl.sql
Recovering with a Backup Control File
Finally, you can use a backup control file from the primary database and recover the database, as shown here:
RMAN> connect target sys@primary
RMAN> restore controlfile;
RMAN> alter database mount;
RMAN> recover database;
RMAN> alter database open resetlogs;
How It Works
Recovering with a backup control file takes the most time of all the options we've reviewed here, but it's the only
option if a failover or the re-creation of the control file from standby proves impossible. The
resetlogs
(
alter
database open resetlogs
) option creates a new redo branch (new incarnation). When the archived redo log from
this new redo branch reaches the standby database, it registers the new redo branch and stops the managed recovery
process (MRP). When you restart MRP at the standby database later on, the database follows the new redo branch.
Configuring RMAN on a Primary Database
Problem
You want to configure RMAN optimally on the primary database in an Oracle Data Guard environment.
