On Tue, Jul 14, 2020 at 03:38:49PM +0530, Bharath Rupireddy wrote:
> Approach #4:
> A postgres_fdw foreign server level option: connection idle time, the
> amount of idle time for that server cached entry, after which the
> cached entry goes away. Probably the backend, before itself going to
> idle, has to be checking the cached entries and see if any of the
> entries has timed out. One problem is that, if the backend just did it
> before going idle, then what about sessions that haven't reached the
> timeout at the point when we go idle, but do reach the timeout later?
Imagine implementing idle_in_session_timeout (which is useful on its
own), and then, when you connect to a foreign postgres_fdw server, you
set idle_in_session_timeout on the foreign side, and it just
disconnects/exits after an idle timeout. There is nothing the sending
side has to do.
--
Bruce Momjian <bruce@momjian.us> https://momjian.us
EnterpriseDB https://enterprisedb.com
The usefulness of a cup is in its emptiness, Bruce Lee