I have a number of servers running Postfix, each of which host several different domain names. They've all been set up using sender_dependent_default_transport_maps to ensure that mail from/to each domain is coming in and exiting the server from it's own IP address.
A few days back I had a crash on an older server so I did a quick migration of one user's domain onto another server. I've been monitoring the logs and everything seemed to be running ok but after they did a month-end invoice mailing I noticed a minor issue in the logs. It's entirely possible that this has been happening for years and I just never noticed... but now that I've seen, it's going to annoy me until I fix it
Unfortunately because of the circumstances I don't have the old server to be able to go back and look through the logs and configs to look for differences.
If I send an email from a test user on the server, everything goes ok and I can see in the message headers that everything is processing with the correct hostnames and IP addresses, etc. They have one user, however, that is used for sending out their monthly accounting mailings. Incoming mail for that user (which honestly is about 90% bounces and 10% of people replying to the "do not reply to this unmonitored email address") is forwarded to their accounting department using the following mail filter receipe...
[pre]redirect "otheruser@theirmaindomain.com";
discard;
stop;[/pre]
The recipe attempts to do the forward but the mail provider for their other domain has been bouncing the forwarded messages as too much traffic from an unknown IP (they have so many random restrictions on sending and receiving... not even going there ). In any case it was from those bounces that I noticed that the forwarded messages are being sent out from the main IP of the server, and not the IP specific to that client.
How can I poke at the configuration of the server to resolve this such that those redirected messages hit the right transport and go out the correct IP address?
Mike
A few days back I had a crash on an older server so I did a quick migration of one user's domain onto another server. I've been monitoring the logs and everything seemed to be running ok but after they did a month-end invoice mailing I noticed a minor issue in the logs. It's entirely possible that this has been happening for years and I just never noticed... but now that I've seen, it's going to annoy me until I fix it
Unfortunately because of the circumstances I don't have the old server to be able to go back and look through the logs and configs to look for differences.If I send an email from a test user on the server, everything goes ok and I can see in the message headers that everything is processing with the correct hostnames and IP addresses, etc. They have one user, however, that is used for sending out their monthly accounting mailings. Incoming mail for that user (which honestly is about 90% bounces and 10% of people replying to the "do not reply to this unmonitored email address") is forwarded to their accounting department using the following mail filter receipe...
[pre]redirect "otheruser@theirmaindomain.com";
discard;
stop;[/pre]
The recipe attempts to do the forward but the mail provider for their other domain has been bouncing the forwarded messages as too much traffic from an unknown IP (they have so many random restrictions on sending and receiving... not even going there
). In any case it was from those bounces that I noticed that the forwarded messages are being sent out from the main IP of the server, and not the IP specific to that client. How can I poke at the configuration of the server to resolve this such that those redirected messages hit the right transport and go out the correct IP address?
Mike
