Configure TAF on Data Guard

Question

Hi,
I've created a data guard setup consisting of single-node primary and standby databases using 11.2.0.1 AIX 6.1
Now I would like to configure the TAF, so that the application is automatically and transparently failed-over to the new primary database during switchover. I've read several links, but they seem to be pointing to some documents which are no-longer accessible on the Oracle website.
I'd a appreciate a simple walk-through on the process, or a pointer to some useful documentation on this.
Regards,
dula

Answer 1

Not sure how far you want to go with this. If far enough is to have the users re-connect then what follows works.
I assume you have flashback database configured and an observer in place.
Given that' true then:
A trigger will be created in the database so that whichever database opens as “primary”, the service will start. The service will be defined in tnsnames so that applications connect to that service. Since only the current primary will have the service running, applications will only connect to the “primary”.
1.     Create the service current primary.
SQL> Exec dbms_service.create_service (service_name => 'testserv',network_name = > 'testserv');
PL/SQL procedure successfully completed.
2.     Now create the trigger to start the service.
create or replace
trigger start_testserv
after startup on database
declare
role varchar (30);
begin
select database_role into role from v$database;
If role= 'PRIMARY' then
dbms_service.start_service('testserv');
end if;
end;
3.     Manually start the service (fires at database startup normally).
SQL> execute dbms_service.start_service('testserv');
4.     Verify the service is started.
Lsnrctl status
And you should see:
Service "testserv.nsight.com" has 1 instance(s).
Instance "standdb", status READY, has 1 handler(s) for this service...
5.     Create a tnsnames entry for this service that references both databases – but only the primary will have the service running so when a user connects, they get the current primary and, should a failover occur, when they login they connect to what is then the primary.
testserv =
(description=
(SDU=32767)
(address_list=
(address=(protocol=tcp)(host=ora-vm-db2.nsight.com)(port=1521))
(address=(protocol=tcp)(host=ora-vm-db3.nsight.com)(port=1521))
(connect_data=
(service_name = testserv.nsight.com)
(server=dedicated)
(FAILOVER_MODE =
(TYPE=select)
(METHOD=BASIC)
(RETRIES=30)
(DELAY=10)
Note: included in the address list are both servers
6.     Test the connection
Create a connection to testserv from sql developer.
select db_unique_name from v$database;
DB_UNIQUE_NAME
standdb
Now shutdown abort the current primary (standdb);
SQL> shutdown abort;
ORACLE instance shut down.
Now connect to the broker and look at status.
DGMGRL> show configuration verbose;
Configuration - drconfig
Protection Mode: MaxAvailability
Databases:
testdb - Primary database
Warning: ORA-16817: unsynchronized fast-start failover configuration
standdb - (*) Physical standby database (disabled)
ORA-16661: the standby database needs to be reinstated
(*) Fast-Start Failover target
Fast-Start Failover: ENABLED
Threshold: 300 seconds
Target: standdb
Observer: ora-vm-db3.nsight.com
Lag Limit: 30 seconds (not in use)
Shutdown Primary: TRUE
Auto-reinstate: TRUE
Configuration Status:
WARNING
The former standby is now the primary.
Try again from sql developer. You will need to reconnect.
select db_unique_name from v$database;
DB_UNIQUE_NAME
testdb
And that is basically how the applications will operate – reconnect if there is a failure and they will be connected to the new primary with no data loss and with only a momentary loss of service while the failover occurs.