We have an AKS cluster on Azure and an endpoint on Azure App Service.

Calls are being initiated from a service (exposing a deployed app) on AKS to an App on Azure App Service (Public URL)

When deploying to Azure App Service, the application goes down in order to deploy, meanwhile calls are still being initiated from the AKS micro-service.

The AKS service starts throwing exceptions that the endpoint is not available which is normal. but when the Azure App service is up again the AKS service still throws the same error even when the app is online again and serving other services successfully. it is like the AKS is caching the error.

In order to fix it I have to kill all the living pods of the faulty micro-service and then scaling it back out again.

Can someone please guide me on why is this happening and how can I prevent this in future deployments?