Hi Patrick, Not sure this is helpful but I searched the archives and found this.
http://marc.info/?l=postfix-users&m=148063002508453&w=2 maybe brush up on the doc of transport config file options ? Good luck. -ALF -Angelo Fazzina Operating Systems Programmer / Analyst University of Connecticut, UITS, SSG, Server Systems 860-486-9075 -----Original Message----- From: owner-postfix-us...@postfix.org [mailto:owner-postfix-us...@postfix.org] On Behalf Of Patrick Landolt Sent: Thursday, June 22, 2017 2:29 PM To: Postfix users <postfix-users@postfix.org> Subject: Re: determine transport based on sender and receiver Thank you taking the time to answer my question. Unfortunately I have no idea what you mean. May you point me to some documentation that I can follow and understand your answer? Regards, Patrick > On 22 Jun 2017, at 17:28, Wietse Venema <wie...@porcupine.org> wrote: > > Patrick Landolt: >> I try to get a setup like to following ready: >> >> 1. I have multiple ip postfix should be able to send mails to other MTA >> 2. I setup different transport in master.cf each with a smtp_bind_address >> corresponding to a single ip >> so far, so good! that setup already works >> >> But now I need to handle slow ISPs like yahoo etc. My first try is to add >> transport_maps with like >> yahoo.com slow: > > This requires support for multiple criteria, in this case sender > and recipient. The only place that currently does this in Postfix > is the SMTP server policy protocol, which is implemented before > mail is queued. A sane implementation of what you want should be > done post-queue. > > Strawman transport_policy: > - The client is implemented in the trivial-rewrite service. > - The protocol format is like the SMTP policy protocol: one line > per name=value attribute. This easy to parse in any language. > - The request contains one sender, one recipient, and perhaps other > information. > - The service replies with one transport, one nexthop, and one > recipient. > - Open question: how are the client requests placed in time relative > to transport_maps, sender_dependent_mumble_maps, and other routing > features? > > Wietse
