Before: ambe.x.yz -> py1:8080|py2:8080|py3:8080|py4:8080|py5:8080|py6:8080 #production,
When changing this to:
After: ambe.x.yz -> py1:8080 #production,
... we noticed the service responded with a 502.
It seems https-portal does not recognise that it goes from a load balanced scenario for a domain to a singular container, when hot-reloading with docker compose up -d.
The dirty solution is to stop the service entirely and start it up again from 0.
I realise this is a bit of an edge case scenario that normally never happens but thought it might be good to have it documented. We do this to debug more easily as we will then know on which container activity will happen.