Recently, I've stumbled upon a situation, where my server rejected an e-mail sent to me from blacklisted DSL via smarthost (beyond my control).
Sender was from domain hosted on his MSP's server. His mail was sent from blacklisted DSL via his MSP's smarthost (which of course requires SMTP AUTH to relay from individual customers) to recipient on my server. This smarthost relayed e-mail to my server, where I run SpamAssassin via Exim's DATA ACL (exiscan-acl). Two of rules that hit (among others, but I'm wondering about those two) were: RCVD_IN_BL_SPAMCOP_NET RCVD_IN_SORBS_WEB and they were referencing original sender's DSL IP address, not smarthost which delivered message to my server. So, mail routing looked like this: [remote_sender_sends_mail_from_blacklisted_DSL]--> [mail_is_accepted_by_his_MSP's_smarthost]--> [this_smarthost_relays_to_my_server_with_SA]--> [my_server_rejects_mail] I'd like to "convince" my SpamAssassin only to do DNSBL checks on last "untrusted" address (or addresses, if there are forged Received: headers, impersonating my server) - that deliver directly to my server, to avoid such situations in the future. What is the preferred way to deal with situations like described above? -- Jacek Politowski
