Retry for endpoint detection

[erikpetzold]

When endpoint detection fails (e.g. due to network issues), then SBA will not know about the endpoints of that instance and there is no option to trigger the detection again. The current solution would be restarting the instance.

Endpoints should not change while the application is running, so we do not need a periodic trigger or a button in the ui. Only in error case there should be a retry. See ProbeEndpointsStrategy.

[pasolid]

Maybe some management button with "Forget instance" functionality. I have similar issue in Kubernetes. I use client discovery to locate all services and their endpoints, but if fo some reason service is redeployed then SBA keeps trying to connect the old one.
I get the rationale behind this behavior, but I would still prefer to have an option/manual control to tell SBA to forget current instance and try to locate another (to trigger rediscovery of service)