On Mon, Dec 16, 2013 at 4:14 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
Joe Van Dyk <joe@tanga.com> writes: > Hm, setting set join_collapse_limit = 9 seemed to fix the problem. Is that > my best/only option?
Yup, that's what I was just about to suggest. You might want to use 10 or 12 in case some of your queries are a bit more complex than this one --- but don't go overboard, or you may find yourself with unreasonable planning time.
Is there a way to measure the planning time? It's not reported in 'explain analyze' or 'explain analyze verbose', right?
