I am using the sqlldr to load data with direct=true and parellel=true.
now my question is i have some column with unique indexes and primary keys also. if at all any record is duplicate then index becomes unusable. so what is the solution to delete those records which are violating unique indexes.
2) can any one provide me the query which making use of hints which does not make use of the index which is in un usable state and delete the records.
You could create a temporary table which looks exactly like your original table, but without any indexes / unique / primary keys.
Load data into the temporary table.
Insert data into original table, eliminating duplicate rows.
I'd say you'll finish it much faster that way than trying to make Loader do a job itself.