Re: Copying database instance from one server to another

David Fitzjarrell wrote:

>Why? He has the complete database from the source machine and the
>operating systems are compatible. Creating a new database is extra and
>unnecessary work. The only items he may need to recreate are a
>password file and an spfile.

I dunno, isn't 7.3 to 9 redhat a pretty big jump?

But maybe I'm antsy about the subject, since I just ran into problems applying the jan 2006 security patch for 9ir2 on hpux, simply because of an older patched linker. Then again, that just might be Oracle support yanking my flugelhorn. Yet again, it does explain weirdo umask and setuid issues in previous installations. Since I haven't worked all the way through it yet, I don't know. But I'd sure hate to have to add serious OS differences to it. And I'd _really_ hate to be unsupported doing it.

All-in-all, I think it is bad practice to clone db's beyond the documented methods, unless you have very tight controls to keep the OS's _identical_ - which is exactly not the OP. In fact, I wonder any time two linux installations exist if they are going to be identical, too many manual steps involved, especially when I see things like http://openacs.org/forums/message-view?message_id=188672

And I have never yet seen anyone test these things adequately for production. (Not saying no one does... they just haven't shown me. Also not saying it won't work - I used to script these things for repeated refreshes until I became comfortable with RMAN. Also recognize assuming there would be any database differences structurally between the same version of Oracle on similar OS's is a weak argument. Also not saying 8174 exp/imp and RMAN don't have problems too...)

Often the bigger argument for creating a new db is the window of opportunity to correct all the deferred maintenance.

jg

-- 
@home.com is bogus.
http://www.paulgraham.com/love.html