@ldez Could you please also look at the whitelists at the same time? They used to be on entry points, but now it's a middleware, which means, that in kubernetes scenarios, it needs to be attached to every single Ingress / Ingress Route.
I used to use whitelist on entry points to make sure that only authorized traffic can reach the cluster ingress, that is I had a hardware load balancer in front of the cluster, that would forward incoming request to each node where traefik listens to.
The incoming IP of the load balancer was whitelisted, so that no one else could access traefik end points.
With version 2 of traefik this is now problematic. People who author kubernetes manifests (that is developers) do not really care about white-listing, so it would be unreasonable to ask them to include the middleware in each ingress route manifest.
Of course, depending on your kubernetes deployment pipeline, it could be possible to include some manifest transformations before they are applied to cluster to account for the mandatory middleware.
However, from security perspective it means that if somehow by accident a middleware is missing from a particular service this service is less secure. By applying whitelisting globally, we make sure that any request does not get past the entry point, thus making the whole integration more secure.
Do you think this could be possible?