I'm not a programmer, but with scripts your worries are likely to become true. Once the scripts are done and disconnected from the DB server, TCP connections will remain in TIME_WAIT. I've noticed that on my systems and solve it by modifying the TCP parameters (reuse, recycle and timeout).
I would not expect this behavior from a daemon policy service keeping the connections alive for reuse. -----Original Message----- From: owner-postfix-us...@postfix.org [mailto:owner-postfix-us...@postfix.org] On Behalf Of E.B. Sent: Thursday, July 16, 2015 10:22 AM To: postfix-users@postfix.org Subject: Proxy database connections to a policy service? Another question about policy service run under postfix spawn--- If using a database to store policy service state, on a busy site I understand this could mean possibly very large number of policy servers along with each smtpd service. if each policy service has a DB connection this could be too much. Is it advisable to connect to postfix proxymap service to solve this problem? If so is there any sample perl code floating around for a policy server to connect to proxyread/proxywrite default services? Is it better to start a separate proxymap service for the policy server? Are there any restrictions that would make this impossible or too difficult? Is there a better solution? Again, sample code much apprecaited :) (or link to open source policy server that uses this technicque) Thnanks!
