On Thu, 31 Mar 2011 12:39:20 -0400, Victor Duchovni <victor.ducho...@morganstanley.com> wrote:
> The receiving sites policies are stupid if they don't implement > them sensibly by just returning 4XX responses without penalizing > subsequent transactions. I am sorry to hijack this thread but we have what seems to be the same problem. While using the default Postfix settings (v.2.8.1 on Ubuntu 10.10), we do have trouble to connect with several MTA's (usually smtp1.min-saude.pt and smtp2.min-saude.pt, but sometimes others at .min-saude.pt). The server at smtp3.min-saude.pt never complains, nor do any of the other email MTA at .min-saude.pt whose name do not start with smtpNN. When they refuse our connections, they seem to start shutting down at 25 to 30 RCPT commands, with: "...mx postfix-slow/smtp[4907]: 36BB7818B: to=<some_subscri...@subdomain.min-saude.pt>, relay=smtp1.min-saude.pt[194.65.151.38]:25, delay=415, delays=414/0.25/0.41/0, dsn=4.0.0, status=deferred (host smtp1.min-saude.pt[194.65.151.38] refused to talk to me: 421 #4.4.5 Too many connections from your host.) " To deal with this we are currently using: /etc/postfix/transport .min-saude.pt slow: /etc/postfix/master.cf slow unix - - - - - smtp -o syslog_name=postfix-slow -o smtp_connection_cache_on_demand=no EOT /etc/postfix/main.cf slow_destination_concurrency_failed_cohort_limit = 3 # we give up after getting three 421 slow_destination_recipient_limit = 20 # keep it bellow 25 slow_destination_rate_delay = 1 # do not know if we really need this > Have you considered the less aggressive > concurrency feedback controls in Postfix 2.5? Do you think that the following would be a more elegant approach than the above described setting? /etc/postfix/master.cf slow unix - - - - - smtp -o syslog_name=postfix-slow -o smtp_connection_reuse_time_limit=30s EOT /etc/postfix/main.cf slow_initial_destination_concurrency = 2 slow_destination_concurrency_limit = 15 slow_destination_concurrency_failed_cohort_limit = 5 slow_destination_concurrency_positive_feedback = 1/5 slow_destination_concurrency_negative_feedback = 1/8 Thank you, M.