Today we enabled SSL on our production TC server using Apache as the middle server to handle HTTPS.
All we did was to apply ssl settings and restart httpd service on the server side.
The "plan" was to wait for all agents to finish their builds and then restart them so they could connect to the secure https:teamcity.
However the problem was that all agents running builds seem to hung after the SSL enabling on the server (some had overtime of more than 1h). Canceling the builds did not work from the UI.. Probably this makes sense since the 'cancel' request would be sent to the non secure teamcity server URL which was not valid anymore. I cannot explain though why the builds hung and did not finish.
I thought that agents are almost 'autonomous' and running builds should NOT be affected even if server is been stopped. When server comes up they can connect again and update TC Server with any data if necessary.
Maybe someone has an explanation on why this happened? If someone can also please clarify how agents and server communicate? Are there ping alive msgs every now and then? This could explain the problem...