importing to 11Gr2 from 9i - full import or schema's only?

One of them has only a couple of users, roles etc. so I pre-create all the tablespaces, users, etc. then import with 11G imp utility using fromuser=user1,user2 etc. in the .imp parfile.

Another has an application front end that creates a database user for all the application users, and assigns certain roles etc. So I have dozens of users on this database, along with various roles as well. I'm wondering what the best way to approach this one is, so I'm thinking:

Do I do a import full=y, and let the import create all the users roles etc. ? *I already pre-created all the application tablespaces when I created the empty database with DBCA

Do I generate the ddl for all the users and roles, run it on the new db first, then import with fromuser=user1,user2..? (doable, but is quite time consuming in this case)

Do I import full=y for trial run, then when ready to import final production data, drop users that have data with cascade option, recreate these users and import their schemas only?

Keep in mind that I am doing several "rehearsals" on this, importing from full exp files on the current production database, for users to test the app against and also to debug my import commands,scripts etc. so this will be a one step(batch) process when I am ready to do the final import.

I'm asking all of this because I'm wondering if its just ok to do full=y when upgrading, (I usually don't)

I'm assuming all the system etc. tables, views etc. in the 11G are just left alone when ignore=y parameter is used and only other user data is imported. Received on Wed May 05 2010 - 09:59:06 CDT