The IGNORE_DUP_KEY will send back a message that the DataPump thinks is
an error.

Personally I would be looking to see how I managed to allow multiple
rows in the source with the same key attribute OR why a key on the
source is not defined the same as in the destination.
I would also be looking at which row in the source for each duplicated
key val is the correct one. Is there a distinguishing
feature/attribute i.e. Date. If there is then I would probably bring
across the rows into a staging table an use TSQL to only pick out the
real rows in which I am interested.


Allan Mitchell (Microsoft SQL Server MVP)
MCSE,MCDBA
www.SQLDTS.com
www.konesans.com - for all your consultancy needs