So rounding it up, we figured it out (thanks to Abhilash).
- After an update from docker-compose v1 to v2, the core-components of my setup suddenly started with a different IP-subnet (192.168.x.y instead of 172.16.0.0/12 ... probably because there were not enough left or so)
- The mailman-web-container always got an extra IP within core to talk to the outside world, so always had 2 IPs (1 fixed within mailman subnet, 1 drawn from the external network)
- This second IP was before within the 12 subnet and being approved in Postfix.
- Probably it was luck, that the correct one was chosen before :-D
The solution now was to set the variable SMTP_HOST to the gateway of the mailman subnet within mailman-web, and NOT to the domain of the host: SMTP_HOST=172.22.0.1 (DB, core and web are then 2,3 and 4)
I also set this within core-container so that IPs are fixed now.
Regards, Jens.
Am 28.08.23 um 14:42 schrieb Jens Günther:
Am 28.08.23 um 10:21 schrieb Stephen J. Turnbull:
Jens Günther writes:
> 2023-08-26T08:21:59.522238+02:00 postfix/smtpd[2843414]: NOQUEUE: > reject: RCPT from unknown[192.168.176.6]: 450 4.7.25 Client host > rejected: cannot find your hostname, [192.168.176.6]; > from=<$admin@mydomain.de> to=<$recipient> proto=ESMTP helo=<mailman-web>
There are a bunch of $myxxxx variables in Postfix, I think $myhosts might be the right one. If you add the IP 192.168.176.6 to that variable, it will be whitelisted and this problem should go away. There may be better ways to handle this, but don't mess with hostnames. Find the "right" way to permit that IP should do the job.
Thank you, adding the IP/network under mynetworks, as Abhilash writes, the error is gone.
(this is so dumb, somehow docker started with a different IP and I have to figure out why ... and there is another Django-error -> shall I open a new thread? "OSError: [Errno 99] Address not available")