>This means that nginx ignores the source port in the proxy protocol. >Is that documented somewhere? It does not ignore it, the variable exists. My configuration doesn´t use it for outbound, as plenty of ports are in used, and dynamic is ok for the use case. Does postfix have a dependency on the port chosen? If one uses a dedicated server or reserves a port range to be used one could probably prescribe the port to be used outbound.
>How does the client find out that a destination server is down? >Is that documented somewhere? I am not aware of documentation but haven´t tried hard to find documentation. In my software I am experiencing timeouts and (unexpected) connection closes, which are imho no different than they occur with direct connections or NAT. My test service retries three times and then gives up, usually on the typical high prio honey pots. >This case is very different from the inbound case where the backends are >monitored for availability. One cannot do monitor the whole Internet >that way. No, but why the entire internet? Postfix would retry anyway on broken connections, or not? I do monitor my postfix queue and get notifications. I am actually not monitoring backends from nginx, as the most likely culprit is a broken internet connection in the data centers or to my home network, and especially for the latter there is nothing I can do with reasonable cost. If one is using proxy protocol within a data center of course monitoring is useful. If you want me to test specific connection scenarios please let me know. Joachim _______________________________________________ Postfix-users mailing list -- postfix-users@postfix.org To unsubscribe send an email to postfix-users-le...@postfix.org
