"Stigers, Greg [And]" wrote:
>
> X-no-Archive:yes
> Upgrading to stay the same?

Don't we all "sort of" do this from time to time, other than those heavily dependent and sacred applications?

> This is something I have wondered about, and have not been able to get a
> clear answer on from those who should know. When upgrading an OS (and
> perhaps ANY OS?) running certain versions of Oracle, it is then necessary to
> "upgrade" Oracle to the same version it already is, but a flavor?
> subversion? (pun intended) of THAT SAME VERSION?

We've been on HPUX 10.20 and only a couple variants of Oracle 7.3.x.x
since 1997.  The impetus of the change is a large new application that
asks for HPUX 11.0 specifically.  Since this requires recompilation of
all the applications, libraries, stored procedures, and forms (that
must sound ludicrous to our MPE people, but it's true) it is just as
easy to update to the latest sub-release of the current Oracle 7.3
release as it would be to stay where we are; you reinstall in either
case.

> Due to some very strange reasons, I have been told not to put time into
> finding my own answers. But we have a production 3000 stranded on 55p7, and
> Oracle 7.2.x. Another party attempted the upgrade to what was then the most
> recent version of 7.3.4.x, but failed, primarily due to a lack of
> preparation (among other deficiencies). As I understood things, this would
> have allowed us to upgrade to MPE 60, while also requiring another Oracle
> "upgrade" to the 60-compatible subversion of 7.3.4.x.

I don't know about Oracle/MPE dependencies, but for HPUX it is horrible.
We did a "virgin install" of Oracle, and ran into several show-stoppers
where we had to tweak a dozen or two kernel parms to get it to run.
So you can't reliably (if ever) update one without at least tweaking
the other.

we aren't gluttons for punishment either, thus we didn't go to Oracle 8
or 8i.  Those are still in the MPE 6.5 category of "let's wait and see".
Richard didn't mention it specifically in his post, but we *did* do a
minor version of Oracle in addition to the HPUX update.  But fortunately
most of the quirks there was due to the "virgin" HPUX kernel before we
rolled forward our kernel tweaks.  But no, we didn't update to 8.

> I find this approach to "upgrades" fascinating. Is it just me, or is an
> upgrade to stay on the same version number just a bit peculiar? Those who
> think it's just me that's a bit peculiar should probably reply off list, in
> light of recent events ;-).

Hope that clarifies a bit.

Jeff Kell <[log in to unmask]>   (*yawn*)