Re: Different plan for very similar queries

"Peter J. Holzer" <hjp-pgsql(at)hjp(dot)at> writes:
>>> Merge Semi Join (cost=316864.57..319975.79 rows=1 width=81) (actual time=7703.917..30948.271 rows=2 loops=1)
>>> Merge Cond: ((t.term)::text = (f.berechnungsart)::text)
>>> -> Index Scan using term_term_idx on term t (cost=0.00..319880.73 rows=636 width=81) (actual time=7703.809..7703.938 rows=3 loops=1)
>>> Filter: (((facttablename)::text = 'facttable_stat_fta4'::text) AND ((columnname)::text = 'berechnungsart'::text))

> Just noticed that this is a bit strange, too:

> This scans the whole index term_term_idx and for every row found it
> checks the table for the filter condition. So it has to read the whole
> index and the whole table, right? But the planner estimates that it will
> return only 636 rows (out of 6.1E6), so using
> term_facttablename_columnname_idx to extract those 636 and then sorting
> them should be quite a bit faster (even just a plain full table scan
> and then sorting should be faster).

Hm. I do not see that here with Tomas' sample data, neither on HEAD nor
9.1: I always get a scan using term_facttablename_columnname_idx. I agree
your plan looks strange. Can you create some sample data that reproduces
that particular misbehavior?

regards, tom lane