Difference between revisions of "(Re)create user/schema using DataPump"

Revision as of 22:57, 26 December 2012

If the user does not yet exist on the destination database, just import it using the procedure outlined here.

If however, the user(schema) already exists, it will have to be dropped and re-created before the data can be imported.
But, before dropping the schema, run this script to generate the correct DDL beforehand!
After this DDL has been run and the results stored, the following command can be executed:

drop user <schema name> cascade;

Now, recreate the schema with the above-mentioned DDL
The dumpfile (or a part thereof) can now be imported to fill up the currently empty schema:

impdp <usr>/<pwd> parfile=schema.par

where schema.par would look like this:

directory=DP_IMPORT
dumpfile=<db_name>_DP.DMP
schemas=(<schema name>)

and/or maybe one or more of the following:

exclude=grants
tables=(<schema name>.<table name>)
remap_tablespace=old:new

Finally, recompile all invalid objects:

exec utl_recomp.recomp_serial();
or
exec utl_recomp.recomp_parallel(4);
or
@?\rdbms\admin\utlrp.sql
or
@?\rdbms\admin\utlprp.sql(4)

and the grants are different on the destination database than they were on the source database, they should be retained. This can be accomplished by excluding grants when importing.

The schema should ideally be checked beforehand that it does not own objects like 'type's that may exist elsewhere...