Personally, I've always found the top-level folder name sufficient for
this (db_1, db_2, etc) - because ultimately it's very difficult to rely
on the directory name to completely identify what software is in the
home. Even if you do add forth or fifth digits, that still won't
accommodate for multiple homes at the same PSU level with different
combinations of one-off patches applied. Also, this doesn't work for
any systems where you only do CPUs since they don't increment the
I was once in a position where we repackaged oracle binaries with
various patchset & one-off combinations for internal distribution and I
was looking for a good version numbering scheme for identifying these
packages. You're trying to deal with the same core problem: coming up
with a "name" (you're using directory instead of my version numbering)
to intelligently and more easily identify the software installed on a
system. I wrote a blog post with some thoughts; doesn't answer your
question, but I think you might find it to be an interesting read -
mainly the requirement list at the bottom of the post for a robust
On 10/28/2011 9:30 AM, [email protected] wrote:
Metalink Note: 1189783.1 speaks of out-of-place upgrading and patching.
I'm wonder if it might be a good idea to change the oracle home naming conventions to use four or five digits under product directory, out-of-place patching being a driving factor.
I was wondering what a sampling of opinions would produce from this list.
Five digits would represent PSU patching. In my case, once the oratab file is updated, everything else (scripts, etc), will set environment correctly, so no issues there.