[Mailman-Users] Strange Wiki entry for Postfix Tuning

Ralf Hildebrandt Ralf.Hildebrandt at charite.de
Fri Aug 7 09:53:10 CEST 2009

* Brad Knowles <brad at shub-internet.org>:

> The way I read this, it has nothing to do with the number of MXes you
> have.  It has to do with how many SMTP delivery sessions you'll
> attempt over the same connection before you drop the connection and
> re-connect (if you have more than this number of deliveries left),
> and that re-connection may well end up going to a different MX.

No. That's controlled elsewhere, based on time:

smtp_connection_cache_destinations = 
smtp_connection_cache_on_demand = yes

smtp_connection_cache_time_limit = 2s
When SMTP connection caching is enabled, the amount of time that an
unused SMTP client socket is kept open before it is closed. Do not
specify larger values without permission from the remote sites. 

smtp_connection_reuse_time_limit = 300s
The amount of time during which Postfix will use an SMTP connection
repeatedly. The timer starts when the connection is initiated (i.e. it
includes the connect, greeting and helo latency, in addition to the
latencies of subsequent mail delivery transactions). 

> This helps avoid conditions where you get locked into a particular MX
> that is slow, and that slows down all your delivery to that site, for
> as long as you have mail for that site.

smtp_connection_reuse_time_limit does that:

The problem starts when one of a set of MX hosts becomes slower than
the rest. Even though SMTP clients connect to fast and slow MX hosts
with equal probability, the slow MX host ends up with more
simultaneous inbound connections than the faster MX hosts, because the
slow MX host needs more time to serve each client request.

Ralf Hildebrandt
  Geschäftsbereich IT | Abteilung Netzwerk
  Charité - Universitätsmedizin Berlin
  Campus Benjamin Franklin
  Hindenburgdamm 30 | D-12203 Berlin
  Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962
  ralf.hildebrandt at charite.de | http://www.charite.de

More information about the Mailman-Users mailing list