Re: 9.2.0.5 is available for win32

"Daniel Morgan" <damorgan_at_x.washington.edu> wrote in message news:1081006651.655924_at_yasure...
> I will give it to you as specifically as I can ... my chief
> technical contact at Oracle ... sorry can't name names ...
> when asked about installing the 9.2.0.5 patch said in two words,
> and I quote: "Don't ... Wait". That is all I know. In the next
> breath he said the 9.2.0.4 CD set just released is fine.

Not so much the 2 words as the pause between them then. Don't wait is of course an unhesitating recommendation!

> I'd suggest watching metalink for awhile and not applying it
> unless it specifically fixes a known problem.

I'm not sure I like the look of users must use 10.1 OUI to ensure that future patches can be applied. There is some interesting stuff in there about what happens if you install 9.2.0.4 before creating a database, which has been my practice for new databases. Possibly it is wiser to install the base release create your new db and then migrate it.

There is as always an interesting list of bugs that are fixed. At least some of the following might be worth an upgrade alone

1. Some types of redo corruption are not detected when they should be
2. SEQUENCES not properly maintained at Logical Standby
3. Archivelog records in the CONTROLFILE may have wrong length data
4. Higher CPU use if SQL_TRACE=TRUE -- the one I've been waiting for.
5. Slow network may hang primary in DataGuard configuration (2 similar ones for this).
6. An auto extensible undo tablespace may grow before reusing expired extents.

There is a heap of RAC stuff as well :(

Overall though, I do somewhat agree with your advice, we test patches in the lab and monitor metalink for new and interesting problems (something that metalink search makes oh so annoying), but still aim to apply within a reasonable timeframe of release.

-- 
Niall Litchfield
Oracle DBA
Audit Commission UK