There appears to be a bug in 2013 (inc. CU1) where the transport stops receiving connections with this error. In my case I have the FE role on the mailbox server, and also have a custom (hub transport) receive connector for apps (tighter scope), so have 6 receive connectors:
- Client proxy | Bindings :::465, 0.0.0.0:465 | RemoteIPRanges: All
- Default | Bindings :::2525,0.0.0.0:2525 | RemoteIPRanges: All
- Default Frontend | Bindings :::25,0.0.0.0:25 | RemoteIPRanges: All
- Outbound Proxy Frontend | Bindings :::717,0.0.0.0:717 | RemoteIPRanges: All
- Client Frontend | Bindings :::587,0.0.0.0:587 | RemoteIPRanges: All
- Custom (anonymous allowed) | Bindings 0.0.0.0:25 | RemoteIPRanges: <specific IP of app servers
There are various discussions on the forums, many with snake-tonic solutions. Currently trying out one with setting the binding on the default receive connector (for internal relay on 2525) to the specific IP address of the server.
Problem is that the problem can occur anything from a few hours to days / weeks, with no reliable repro. Will try to remember to update this post if I see a reliable fix.