On Tue, 2009-05-12 at 21:24 +0100, Simon Riggs wrote:
> On Tue, 2009-05-12 at 15:52 -0400, Robert Haas wrote:
> > On Tue, May 12, 2009 at 12:49 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> > > 1. There is no (portable) way to pass the connection from the postmaster
> > > to another pre-existing process.
> >
> > [Googles.] It's not obvious to me that SCM_RIGHTS is non-portable,
> > and Windows has an API call WSADuplicateSocket() specifically for this
> > purpose.
>
> Robert, Greg,
>
> Tom's main point is it isn't worth doing. We have connection pooling
> software that works well, very well. Why do we want to bring it into
> core? (Think of the bugs we'd hit...) If we did, who would care?
I would.
Not to knock poolers but they are limited and not very efficient. Heck
the best one I have used is pgbouncer and it has problems too under
heavy load (due to libevent issues). It also doesn't support all of our
auth methods.
Apache solved this problem back when it was still called NSCA HTTPD. Why
aren't we preforking again?
Joshua D. Drake
--
PostgreSQL - XMPP: jdrake@jabber.postgresql.org
Consulting, Development, Support, Training
503-667-4564 - http://www.commandprompt.com/
The PostgreSQL Company, serving since 1997