Ok. So - makes sense. Have you thought about solving this problem via DNS? So - your collectors use a DNS name instead of an IP Address to connect to your servers? then you can control via DNS to which server the collectors are connecting?
The Docker images on github are mainly for evaluation and testing. While you can run a server or collector in docker we do not officially support it for production environments.It mainly comes down to making sure we get the resources (cpu, mem, ...) that our deployment guide suggests. If you are not configuring docker correctly (and that is true for other virtualized environments) we might not get these resources which can impact dynatrace performance
I forwarded this thread to our product management team to get an official response. Maybe we should clarify the "deprecation of multiple server instances" and what we officially suggest to support multiple servers on a single server instance
Interesting, i am exploring this area too but just completed reverse proxy browsers -> rp -> Web AppMon, Client -> rp -> AppMon. I am working in another PoC which would like to test this it too collector -> rp -> reverse proxy.
My exploration is a bit different from yours. my intent to use nginx rpoxy to direct customer to their respective appmon web and/or client. Each customer would have a subdomain for web and client. But client don't support https tunnel as I did try and it only works for http tunnel. next best alternative is to use nginx stream but then each customer would have a different port to connect to.