Hello! I see a minor glich, doesn't hurt, just would like to know what causes it.
I am using Postfix + Cyrus, using unixsocket-LMTP transport. I am using address verification for local destinations too (Postfix doesn't know about the cyrus users, so it uses the LMTP to discover a user, and reacts to the client accordingly). I wrote a simple script to check i configured an e-mail address properly (it's deliverable), it simply issues a MAIL FROM:<mym...@mydomain.com> then an RCPT TO:<testm...@testdomain.com> then a QUIT. And I see the response for the RCPT command. "mym...@mydomain.com" is a local e-mail address. "testm...@testdomain.com" too. I am asking the master domain name server for both domains + the IP i am using at client side has forward and reverse too on that server, and they're connected with a crossover cable, so name resolution is fast. I am using native IPv6 (both SMTP connection and NS lookup). When i first try a new e-mail address, it takes around 10 seconds to check it exists or not. Second time it takes less than 1 second (the address comes back from the verify cache). So my conclusion is: The slow point could be the reject_unverified_recipient check, which causes a verify lookup, which makes an LMTP lookup, so it's either Cyrus problem, or might be something before that. The strange things: The server is rather frequently used, but i wouldn't say the load is high. Load is around 0.02, half of the 1GB ram is free (used by cache and buffers), and nothing starts to eat processor time when i make the first e-mail address check. Also the LMTP protocol itself seems to be fast when it comes to delivery (like the second mail with success verification, containing DATA command, which also traverses through SpamAssassin).What piece of configuration would you like to see, cause sending all is rejected by the mailing list.
smime.p7s
Description: S/MIME Cryptographic Signature
