changing from load balanced configuration for domain to singular container creates 502
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.
Hmm .. In your first case with load balancing, I'm not entirely sure whether HTTPS-PORTAL was able to pick up the load balancing string.
If you have DEBUG=true in environment variables, does the debug info actually tell you the load balancing is properly parsed?