The remedy to that is to let it fail miserably or at
least take lots of downtime in order to convince the PHBs that there are better
The best part is that there's nothing nefarious about
it because it'll happen wether you want it to or not.
Luckily the only CPU (Jan'06) I applied to the
Windohs/Oracle box was before it went production. And the 4-way Xeon
had no problems with it, which I can only explain as pure luck. And yes,
Windohs locks down open Oracle files, including log and trace files
Thanks Jeremiah, I will take a look at that.
10 minutes is
definitely a big improvement over an hour.
The problems I have run into
however have been
on Windows machines - no relinking.
Things may go
smoothly on a test system, but
not quite so smoothly on production, taking
more time. It has happened more than once on
Windows (for me at
least) that some steps had to
be repeated as they would fail on the first
(can't recall at the moment just which ones)
that the backout scripts generated by Opatch
on Windows do not work as is
does not help.
(The CALL command has be inserted where the BAT
script calls other scripts )
I don't think you can replace the
binaries while in use either,
as they will be locked by Windows while in