Switching Platforms

Have you reviewed the online docs on these topics?

These methods can save you time if dealing with large amounts of data.  I don't think you will find anything that says when you should use one method over the other.

I'm in the process now.  Someone on a listserve stated they were able to make the platform change using transportable tablespaces.  Looking at the documentation it appears to me that using transportable tablespaces you still need to run expdp anyhow so I'm not seeing any benefit.  The database is about 350 gb.  Exported the .dmp file comes out to about 80 gb.
Of course there is a bug 8984274 where Oracle is incorrectly having a user grant privileges to themselves when using expdp/impdp which is currently preventing my attempt to import.  However if Oracle supplies us with a patch for this bug i'm just wondering if a straight expdp/impdp[ will be sufficient for the platform change or if for some reason i don't know i will be compelled to do the transportable tablespace exercise?

That is probably correct since the actual data will be the tablespaces having gone through the RMAN CONVERT process.  This is what I'm inquiring about.  If i create the database on linux, should I not be able to just then import from a plain .dmp file?  When I tried it did seem to be working, but then ran into that bug where it was trying GRANT INSERT ON tablex TO usera, then generated that ORA-01749 error saying usera cannot grant privileges to itself.

Did you do a FULL import?  If so, I'm not a big fan.  I normally import the specific schemas I need after I have pre-created them (using IGNORE=Y).

I've tried it both ways.  After the full failed on the ORA- error, I tried just doing the schema.  The problem is the schema has the same incorrect GRANT INSERT ON tablex TO usery and generated the same error saying usery  can't grant itself privileges.   With no patch the only thing i can see is impdp system/pass
SQLFILE=import.sql ........., then go into the SQLFILE and not only extract sql that will create the tablespaces but also create the users, then extract all the GRANT(s) into some script and run that as system or sys instead of as the users as Oracle is trying to do.  I don't know about that IGNORE.  There may be something i do NOT want to IGNORE happening :-)

I'm not an impdp expert.  I've only used it a couple of times.  In the old imp version you needed to set IGNORE=Y if the objects already existed in the database because imp tried to create them.

I'm sure the same sort of thing is around in impdp.  The docs will have the info.

Yes, it exists with impdp also.  Basically it will ignore an error from a  CREATE TABLE where the table exists and proceed to load the existing table.

Actually its not IGNORE=Y with impdp but something like EXISTS_ACTION=skip.
11.2.2 fixed the problem, but now I back on a 10.2.0.4 and I'll probably have to follow your technique for the 10 databases.