Tomas Vondra <tv@xxxxxxxx> writes: > Anyway the output is a bit strange. It's writing a lot of temp files > that are significantly smaller (about 3MB) than work_mem (128MB). The most obvious guess about what's doing that is a hash join that has a drastic overestimate of how many rows it has to hash, so that it's batching the join unnecessarily. AFAIR, sorts don't spill to disk until they've actually reached work_mem, but hash join works differently. regards, tom lane -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
