os is windows 2003 sp 2
I have practised patch apply on copy production many times and have
confidence this will work.
The general order of events will be:-
1) Stop production
2) Stop Standy
3) apply binaries part of patch on production
4) apply binarries part of patch on Standby
5) Start standby
6) Start production
7) Apply database scripts to production
I will be unable to test until the week-end
My main concern is that standby can apply archivelogs that were
generated from redo before patch / database upgrade as well as logs
resulting from upgrade and after
/ especially in rac environment wher application of standby logs
generated from independant nodes is not independant
If anyone can allay my fears or point to hints/documentation I would
be gratefull
cheers
Chris B
Hi Chris,
We have done this several times without any issues. When we were
particularly nervous, after patching the primary software and
database, we would allow the the standby to accumulate the transaction
logs. We didn't apply them until we had patched the database software
on the standby database. We would then apply the logs, through which
the new patched structures would be applied by the correct version of
the database software on the standby.
However, as noted, we never had any issues when we just let them be
applied even with the "wrong" version (i.e., the standby database
software that didn't have the CPU patch applied).
HTH,
Steve
Harry