"Peter Darley" <pdarley@Kinesis-CEM.com> writes:
> Something that caught me up at first is that PostgreSQL won't shut down
> until all it's backends close, which means that if you have an application
> connected that uses pooled or cached connections you will probably have to
> shut it down first.
This depends on how the postmaster is commanded to shut down. I think
what Troy is looking for is the various stop-mode options to pg_ctl:
http://developer.postgresql.org/docs/postgres/app-pg-ctl.html
or perhaps he wants to know about the underlying signals that can be
sent to the postmaster:
http://developer.postgresql.org/docs/postgres/postmaster-shutdown.html
regards, tom lane