"=?iso-8859-2?B?U1rbQ1MgR+Fib3I=?=" <surrano@mailbox.hu> writes:
> ---------------------------- 7.2.1 PLAN ---------------------------------
> -> Seq Scan on valuta (cost=0.00..1.06 rows=6 width=4) (actual time=0.02..0.11 rows=6 loops=2)
>
> ---------------------------- 7.3.2 PLAN ---------------------------------
> -> Seq Scan on valuta (cost=0.00..20.00 rows=1000 width=4) (actual
time=0.02..0.06rows=6 loops=2)
Ah, there's the problem. You never vacuumed or analyzed "valuta", so
the 7.3 planner didn't know it had only six rows, and chose a plan that
was more appropriate for a larger table. The thousand-row estimate is
the tipoff, because that's the default assumption when there are no
stats.
regards, tom lane