I have a subnet of IPs that I allocate to a dozen different domains, and right now I use a combination of master.cf rules and sender_dependent_default_transport_maps = lmdb:/etc/postfix/sender_transport to send out mail from different customer virtual domains via different IPs. If I put PMG in front of this mailserver on its own VPS with a high risk UCEPROTECT-Level1 allocated IP, and if all mail goes out this single IP, I will be in a world of pain and lose segregation of customer domains per IP. Incoming mail would go to the MX of the PMG VPS and get forwarded to the mail server for customers to pick up, all good, but would outgoing from the mailserver need to go through PMG (for scanning and logging) and then get routed back to the mailserver so that the mailserver can then on send that message out via its current sender_transport rules?
Will this work anyway?
Is there anyway I could avoid the re-send back to the mailserver from PMG to take advantage of outgoing IP per domain and avoid that extra outgoing hop back to the main mailserver?
How to avoid the sender_transport lookup when first sending outgoing mail to PMG, but respect sender_transport when the message comes back to the mailserver and then needs to go out again directly from the mailserver (per subnet IP) via sender_transport?
Will this work anyway?
Is there anyway I could avoid the re-send back to the mailserver from PMG to take advantage of outgoing IP per domain and avoid that extra outgoing hop back to the main mailserver?
How to avoid the sender_transport lookup when first sending outgoing mail to PMG, but respect sender_transport when the message comes back to the mailserver and then needs to go out again directly from the mailserver (per subnet IP) via sender_transport?
Last edited: