Create 11g Standby (single node or RAC primary, single node standby)

Example node names: prmdb[ab] as primary, stbdb as standby. Assume both have +DATA and +FRA diskgroups.


1. Make sure primary can connect to standby as sys
	sqlplus sys/password@oracpsb as sysdba
and standby can connect to primary as sys
	sqlplus sys/password@oracp as sysdba

To do that, on primary, add
ORACP10SB =
  (DESCRIPTION =
    (ADDRESS = (PROTOCOL = TCP)(HOST = stbdb.example.com)(PORT = 1521))
    (CONNECT_DATA =
      (SID = oracpsb)
    )
  )
to <DB home>/network/admin/admin/tnsnames.ora, on all nodes if RAC. On standby, add
ORACP10 =
  (DESCRIPTION =
    (ADDRESS = (PROTOCOL = TCP)(HOST = d1prvscan10.example.com)(PORT = 1521))
    (CONNECT_DATA =
      (SERVICE_NAME = oracp)
    )
  )
both ASM (i.e. grid) home and DB home network/admin/tnsnames.ora. On standby, add
SID_LIST_LISTENER =
 (SID_LIST =
   (SID_DESC =
     (ORACLE_HOME = /u01/app/oracle/product/11.2.0/db)
     (SID_NAME = oracpsb)
   )
  )
to <ASM home>/network/admin/listener.ora and bounce listener (using +ASM, not DB, as environment).

On primary, scp <DB home>/dbs/orapw$ORACLE_SID <standby>:<standby DB home>/dbs
On standby, mv the password file to match standby SID, e.g., mv orapworacp1 orapworacpsb.
(If standby is RAC, you can choose to have the same SID as primary and no need to mv.)


2. Prepare primary

alter system set fal_client = 'oracp';
alter system set fal_server = 'oracpsb';
alter system set log_archive_config = 'dg_config=(oracp,oracpsb)';
alter system set log_archive_dest_2 = 'service=oracpsb valid_for=(online_logfiles,primary_role) db_unique_name=oracpsb';
alter system set log_archive_dest_state_2 = defer;  <-- just for now
alter system set standby_file_management = auto;
show parameter spfile
create pfile='/tmp/init.ora' from spfile='<spfile path shown above>';
scp /tmp/init.ora <standby DB home/dbs>


3. Prepare standby

cd $ORACLE_HOME/dbs
mv init.ora init<standby SID>.ora
Edit the file (prepend *. to param names if RAC):
remove cluster_database=true (unless it's RAC)
reduce SGA and db_cache_size
add db_unique_name='oracpsb'
change fal_client and fal_server, e.g.:
	fal_client='oracpsb', fal_server='oracp'
change log_archive_dest_[12], e.g.:
	log_archive_dest_1='LOCATION=+FRA'
	log_archive_dest_2='service=oracp async valid_for=(online_logfile,primary_role) db_unique_name=oracp'
sqlplus / as sysdba
startup nomount


4. Create standby

On primary
rman target / auxiliary sys/password@oracpsb
RMAN> duplicate target database for standby from active database nofilenamecheck;

If it complains about tape channel, force to use disk channel. If it complains about missing a regular channel, add a disk channel:

run {
allocate channel disk1 device type disk;
allocate auxiliary channel disk2 device type disk;
duplicate target database for standby from active database nofilenamecheck;
}

Can add more channels if the server has many CPUs.

If there's error not resolving the TNS name or invalid username/password, after checking anything obvious, try
* changing the single / to sys/password@<connect identifier>
* explicitly specifying nocatalog (even though it's default)
* making sure sys account is open.


5. Finish up

On standby: alter database recover managed standby database disconnect;
On primary:
	alter system set log_archive_dest_state_2 = enable;
	alter system switch logfile; <-- if RAC, do it on the instance the standby is waiting for, shown in standby alert.log or v$managed_standby
Standby alert.log should show a line beginning like 'Media Recovery Log +FRA/oracpsb/archivelog/...'

On primary and standby:
alter database add standby logfile ('+DATA','+FRA') size <size of current logfile viewde from v$log>;
Run it multiple times so the count matches regular logfile groups

On primary:
alter database force logging;