Stefan Kaltenbrunner <stefan(at)kaltenbrunner(dot)cc> writes:
> ok - the planner switches to a different plan at about 2.5GB of
> effective_cache_size resulting in the following [ much better ] plan:
OK, so it sounds like it'd be a good idea to try to pro-rate
effective_cache_size among all the tables in the query. I'll see
if I can get that in without kluging the code up too much.
regards, tom lane