"Luke Chambers" <lukechambers7@gmail.com> writes:
> The following query plans both result from the very same query run on
> different servers. They obviously differ drastically, but I don't why
> as one db is a slonied copy of the other with identical postgresql.conf
> files.
There's an order-of-magnitude difference in the estimated row counts for
some of the joins, so it's hardly surprising that different plans would
be chosen. Assuming that these are exactly the same Postgres version,
the only explanation would be considerably different ANALYZE statistics
stored in the two databases.
> Both databases are vacuum analyzed nightly.
Maybe you should double-check that.
regards, tom lane