Particular attention must be given to the operation of PRIMECLUSTER Wizard for Oracle in the Oracle Data Guard environment.
Such as Oracle Data Guard Broker, operation that automatically executes failover from the primary database to the standby database is not supported.
You can change the apply services using ALTER statements while a cluster system is running, but the attribute "REDOApply" of the Oracle instance resource is not changed. For changing the attribute "REDOApply" of the Oracle instance resource, stop RMS on all nodes and use RMS Wizard to change it.
When switch the database role, start maintenance mode of PRIMECLUSTER and stop monitoring the Oracle resources. For details, see "G.4 Overview of switch procedure of database role".
There is no problem though the following messages might be output to syslog when stopping and switching of userApplication on the standby site.
Message including ORA-01109
FSP_PCLW-ORACLE_FJSVclora: ERROR: 3403: clorainstance detected Oracle error! (ORA-01109: database not open ) Please check the Oracle alert log file.
The above message is output to syslog when stopping and switching of userApplication on the physical standby database in MOUNTED state.
Message including ORA-16136
FSP_PCLW-ORACLE_FJSVclora: ERROR: 3403: clorainstance detected Oracle error! (ORA-16136: Managed Standby Recovery not active ) Please check the Oracle alert log file.
The above message is output to syslog when stopping and switching of userApplication on the physical standby database with apply services is not active.
There is no problem though the following messages might be output to an alert log of the Oracle database when the start and the switch of userApplication are done on the standby site when the communication has not been established between the primary site and the standby site after database is converted from the snapshot standby to the physical standby.
There are cases that the primary database has stopped, the listener has stopped, and network is abnormal as the communication has not been established.
Sun Jul 24 01:14:28 2011 ALTER DATABASE RECOVER AUTOMATIC DATABASE Media Recovery Start started logmerger process Sun Jul 24 01:14:28 2011 Managed Standby Recovery not using Real Time Apply Media Recovery failed with error 1666 Recovery Slave PR00 previously exited with exception 283 ORA-283 signalled during: ALTER DATABASE RECOVER AUTOMATIC DATABASE...
If you stop Oracle instance by maintenance on the standby site, become the same state in which the log apply services is specified for attribute "REDOApply" of the Oracle instance resource when you restart Oracle instance.
The parameter passed to user script mount* under the /opt/FJSVclora/usr is as follows. For details about user script mount* under the /opt/FJSVclora/usr, see "2.3.1 PRIMECLUSTER Installation and Configuration".
$1 : $ORACLE_HOME
$2 : $ORACLE_SID
$3 : Oracle user name
$4 : Database role (DATABASE_ROLE column in V$DATABASE view)
In the scalable operation with Oracle RAC, when the Oracle instance failure occurs on the standby database and the Oracle instance is recovered by Oracle Clusterware, the Oracle instance becomes the state of OPEN and the REDO Apply is not restarted, regardless of the setting of REDO Apply of PRIMECLUSTER Wizard for Oracle.
In the scalable operation with Oracle RAC, you must set and operate apply services of the standby database to start on one of the nodes.