> On Fri, May 8, 2009 at 10:00 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: >> Thanks. Could I trouble you for one other data point --- about how many >> rows are in each of these tables?
> Not a problem:
As best I can tell, the selectivity numbers are about what they should be --- for instance, using these stats I get a selectivity of 0.0000074 for the join clause fkr.submission_id = tr.submission_id. Over the entire relations (646484 and 142698 rows) that's predicting a join size of 683551, which seems to be in the right ballpark (it looks like actually it's one join row per canvas_foreignkeyresponse row, correct?).
I took the time to load this data into an 8.4beta2 install, and the same query runs in a much more reasonable timeframe (~3s as opposed to ~50s). I set the statistics target to 500, and got this explain [1].