When running our azure functions http triggers in consumption plan everything works as expected, the functions instances that are not buzy gets the requests and when checking the metrics for those all request are done within expected time.
After moving the same azure functions with http trigger to an app service plan some of the instances freezes and stays froozen even after the functionTimeout. Also new still tries to do requests against those froozen instances which makes the requests timeout.
So wonder why those freezes in the first place (which they dont do in consumption plan), and why not the "load balancer" this send requests to the froozen?