Upgrade Physical Standby 8i to 9i

we are currently upgrading our oracle software from 8.1.7.4 to 9.2.0.8
( politics are keeping us from 10 at the moment )

I have a physical standby database in another location. it is running in mount state. we have OS scripts that copy archive logs to the location and apply to logs to the mounted database - continuously.

when we upgrade production - how to do the upgrade on the standby without a restore and recover and without opening the database ( which would lead to a restore and recover )

I have searched metalink and read some notes that deal with standby
databases. It seems that these standby databases are set up with RMAN ...
the one i am dealing with was not ...

any advice or direction would be appreciated ...

It does not matter how standby was setup.

When you are ready to upgrade, make a note of your latest log file and make sure it applied to the standby database. Proceed with the normal upgrade steps, after you done with the primary upgrade standby server (only software portion of course) When everything completed your standby database should be able to accept archived logs from primary.
Also look into dataguard configuration, you do not need to use shell scripts to move archived redo logs , arch or lgwr will take care of this.

after upgrade of standby ( software only ) to 9i, on production site, we are
planning to drop rollback segments and rollback tablespace and implement the an undo tablespace and let oracle manage the undo.

question: when rollback segments are offlined and dropped on production, how to make that happen on the standby.

will the archive log contain the instructions and when applied remove rollbacks from data dictionary ( and update controlfile ) ?

will i have to remove the rollback logs manally on standby ?

I know data guard is a better solution for standby, but we aren't there yet.

Quote:

---

Originally Posted by gojack

after upgrade of standby ( software only ) to 9i, on production site, we are
planning to drop rollback segments and rollback tablespace and implement the an undo tablespace and let oracle manage the undo.

question: when rollback segments are offlined and dropped on production, how to make that happen on the standby.

will the archive log contain the instructions and when applied remove rollbacks from data dictionary ( and update controlfile ) ?

will i have to remove the rollback logs manally on standby ?

I know data guard is a better solution for standby, but we aren't there yet.

---

How big is your DB?
How much downtime can you afford?

If the DB is small enough and you can afford some amount of downtime, best way is after you finish the UNDO thing on the primary instance. Take a cold backup and go for Physical Standby configuration using standby controlfile in oracle 9i.

Put it in managed recovery and don't worry about manual shipping of Archive logs.