Upgrading Oracle in clusterenvironment with standby
edited Sep 10, 2009 10:51PM in High Availability Data Guard, Sharding and Global Data Services (MOSC) 4 commentsAnswered
Hi,
Our customer wants to upgrade the production-environment from Oracle 10.2.0.3 to 10.2.0.4.
The downtime should be as short as possible.
It is a clusterenvironment with physical standby.
The architecture looks like (clustermembers) A01, A02, A03, A04. A01 and A02 are used for the primary database, and A03 and A04 for the secundary. It's all Windows 2003.
To be able to get back we did the following:
Did a fresh installation of 10.2.0.1 in another directory, ran the 10.2.0.4-patchset over it. So the systems now have 2 Oracle-homes (1 10.2.0.3 and 1 10.2.0.4)
The clusterenvironment was modified so that during failover Oracle shouldn't be restarted. Brought Oracle down on A01 and A03 and started the database on A02 in upgrademode with the new oracle-home. Ran the upgradescript, which lasted 40 minutes. The following step is recreating the standby database with the new oracle-home. As the total downtime is approx 1 hour, the customer is wondering if there is a faster method. Starting up the standby as production isn't an option I guess. Or isn't there a faster method. If there isn't the customer has to live with a downtime of 1 hour.
Our customer wants to upgrade the production-environment from Oracle 10.2.0.3 to 10.2.0.4.
The downtime should be as short as possible.
It is a clusterenvironment with physical standby.
The architecture looks like (clustermembers) A01, A02, A03, A04. A01 and A02 are used for the primary database, and A03 and A04 for the secundary. It's all Windows 2003.
To be able to get back we did the following:
Did a fresh installation of 10.2.0.1 in another directory, ran the 10.2.0.4-patchset over it. So the systems now have 2 Oracle-homes (1 10.2.0.3 and 1 10.2.0.4)
The clusterenvironment was modified so that during failover Oracle shouldn't be restarted. Brought Oracle down on A01 and A03 and started the database on A02 in upgrademode with the new oracle-home. Ran the upgradescript, which lasted 40 minutes. The following step is recreating the standby database with the new oracle-home. As the total downtime is approx 1 hour, the customer is wondering if there is a faster method. Starting up the standby as production isn't an option I guess. Or isn't there a faster method. If there isn't the customer has to live with a downtime of 1 hour.
0