Re: Postgres optimizer choosing wrong index

Jack Orenstein <jack.orenstein@hds.com> writes:
> EXPLAIN says that the correct index is being used -- it didn't used
> to. However, pg_stat* says otherwise. In my test, I have exactly one
> dh value. Running EXPLAIN with this value produces a plan using idx_dh
> (the correct index), but pg_stats says that idx_dn is being used (see
> psql session below).

Yeah, if you are using cached plans (via PREPARE or plpgsql functions)
then the plan stays the same for the life of the session ... pre 8.3
that is.

            regards, tom lane