>
> I'd still recommend manually running ANALYZE on any involved tables,
> just in case.
>
Just out of curiosity....
Say I did run it. Would it make the query run a few orders of magnitude faster?
If you read my original post you'd see that I let the query run for
over an hour before I killed it. It seems to me that a query joining
two tables on primary keys, updating one field on one table with
another field on another table on less than 100K records ought to take
less than three minutes.
Do the database statistics go that far out of whack especially if
autovacuum is on?