Yes postmaster as recipient of delivery status notifications. I have set > notify_classes = bounce, 2bounce, delay, policy, protocol, resource, software
so postmaster receives delay warnings in general on my system. But it also receives warnings for messages where the one and only RCPT TO of the message gets NOTIFY=NEVER via smtp command filter. That's why I guessed that the NOTIFY=NEVER maybe only affect delay warning to the sender but not the one that is send directly to postmaster due to notify_classes. Cheers tobi On Fri, 2022-01-07 at 10:57 -0500, Wietse Venema wrote: > tobs...@brain-force.ch: > > Wietse > > > > seems the ugly hack does not work for delay messages to postmaster. > > postmaster as the RECIPIENT of copies of delivery status > notifications? > > By default, postmaster does not receive copies of delivery > notifications. So that is one way to fix this. > > If you turn on postmaster copies of delivery status notifications, > and if some recipients of a message have NOTIFY=NEVER and other > recipients do not, then postmaster should receive copies of delivery > status notifications only for the "other" recipients. > > If that is not the case then that would be a bug in Postfix's > notification code. > > Wietse > > > Is > > it possible it only works for delay messages to sender and not for > > postmaster delay notifications? I still got delay warnings in my > > postmaster box using the ugly hack. > > > > Anyway changed back to graveyard-mode ;-) > > > > Cheers > > > > tobi > > On Fri, 2022-01-07 at 08:00 +0100, tobs...@brain-force.ch wrote: > > > Wietse, > > > > > > > With an ugly hack we could make these a "notify-none" recipient > > > > > > I always like ugly hacks ;-) > > > > > > This > > > > > > > /^(RCPT\s+TO:\s<corpus-(ham|spam)@MYDOMAIN.*)/$1 NOTIFY=NEVER > > > > > > in command_filter seems to do the job. Just wonder could a time- > > > unit > > > be > > > specified instead of NEVER? > > > > > > But I still keep my graveyard postfix which I already setup based > > > on > > > Viktors reply. Always good to have a graveyard behind the house > > > ;-) > > > > > > Have a good one and happy 2022 to all > > > > > > tobi > > > > > > On Thu, 2022-01-06 at 11:30 -0500, Wietse Venema wrote: > > > > Viktor Dukhovni: > > > > > On Thu, Jan 06, 2022 at 01:08:33PM +0100, > > > > > tobs...@brain-force.ch?wrote: > > > > > > > > > > > Is somehow possible to use other delay notification > > > > > > settings > > > > > > for > > > > > > a > > > > > > particular recipient address? > > > > > > > > > > No, this is a message-level property, same for all delayed > > > > > recipients > > > > > of the message. > > > > > > > > With an ugly hack we could make these a "notify-none" recipient > > > > (see example in > > > > http://www.postfix.org/postconf.5.html#smtpd_command_filter). > > > > > > > > > > My global setting is 30min which is fine except for two > > > > > > addresses. > > > > > > Those addresses are on a remote system which is not always > > > > > > up. > > > > > > It > > > > > > gets > > > > > > the mail on boot by ETRN command to the postfix server. So > > > > > > it > > > > > > queues on > > > > > > postfix server until that box is up again the next day > > > > > > (starts > > > > > > once a > > > > > > day for about 3hrs) > > > > > > > > > > > > Now I get a shipload of delay message warnings for this two > > > > > > addresses. > > > > > > So I wonder if it would be possible to configure other > > > > > > delay > > > > > > warning > > > > > > setting (ex 24h) for these two particular addresses. > > > > > > > > > > > > Thanks for any ideas and have a good one > > > > > > > > > > The only way to handle this is to relay the mail in question > > > > > to a > > > > > separate "slow" Postfix instance (not a transport, but a > > > > > completely > > > > > separate Postfix with its own main.cf, queue manager, ...). > > > > > > > > > > In that Postfix instance you can have a longer or no delay > > > > > warning. > > > > > > > > I agree, a proper solution requires a Postfix 'graveyard' > > > > instance. > > > > > > > > ????????Wietse > > > >
