Svein Erling Tysvaer schrieb:

> Maybe something like
>
> select pk_field, count(*)
> from table
> group by pk_field
> having count(*) > 1
>
> to identify the duplicate records to remove/modify before recreating the
> key?
>
> Set

which will only help as long as you've got the current databasefile and
won't help at all with the database that's recreated without indexes
from a backupfile...

If you have to recreate it from the backupfile you might think about
creating one database without indexes, one just with the metadata,
locate and remove the damaged records as Ser suggested and then pump the
data from one database to the other.

Luc.