On 9/29/06, Kevin Closson <kevinc@xxxxxxxxxxxxx> wrote:
Back on thread. This post makes me ask, is it common out there to apply patchsets (not a "patch" from r1 to r2 as in th OP) to functional Oracle Homes? I should think a safer approach would be to install a new R1 home, apply R2 to it and if you are happy, switch over to it by dragging your TNS stuff over. Or is my views of the common Oracle Home as too simplistic? That is how I approach it here anyway. That is, until we finish implementing our filesystem snapshots in which case the patchset application process can be as junky as anyone could imagine and I wouldn't care. Just revert to a snapshot and make it writeable... not worries.
Today I patch the current oracle home.
What I would like to do in the future is clone the Oracle Home, patch it and and leave the old one behind for a few days as a failsafe if needed.
Sometimes it would be needed right away. :)
The only obstacles (I think) to doing so are managing the Oracle inventory.
Andy Rivenes has a great paper on managing Oracle apps from a DBA perspective, which includes managing the Oracle Inventory.
Following the practice outlined in that paper would make it a simple matter to rename the Oracle Home, cp -r to the oldOracle Home name, patch it and restart the database.
This is also one of the techniques Jeremiah Wilton espouses for minimizing downtime when patching.
These techiques assume a unix like platform. Cloning on Windows is not covered. It can be done, but requires different (more work) techniques.
It's the registry ya know.
-- Jared Still Certifiable Oracle DBA and Part Time Perl Evangelist