Stephen Friedrich wrote:
Richard Huxton wrote:
I don't suppose changing it to:
ORDER BY this_.cdr_id, this_.id
does anything for you?
Nope. Same cost :-(
Maybe I'm barking up the wrong tree then.
Maybe the costs for the index just look wrong.
Hmm - looking at it, this seems the case. Directly querying the
partition it uses the index but estimates a cost of:
cost=0.00..23510.68 rows=575532 width=8907)
The seq-scan says it has a cost of:
cost=0.00..20737.15 rows=575532 width=8907
Could you issue "set enable_seqscan=off" and then show explain analyse
on the second query again?
--
Richard Huxton
Archonet Ltd
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
