Thanks!
|>ouglas
P.S. Actually, for this particular problem, just deleting the statistics would be fine. I've tried doing:
alter table maindb_astobject alter column survey_id set statistics 0;
And then analyzing the column, but when "statistics" for a column are set to 0, Postgres seems to leave the current statistics in place, which is not the right thing for me at all. I can successfully set "statistics" to 1, but that turns out to be one statistic too many.I've tried settings the statistics via the table "pg_stats", but that turns out to be a view, and Postgres won't allow to me to alter it.
Perhaps I can achieve the end by altering the "pg_statistic" table instead, but that table is more than a bit opaque to me.
P.P.S The Seq Scan is 2-4 orders of magnitude slower than the Index Scan.