Add an alert about average nodes memory utilization

[Aman1994]

What is missing?

Alert about average nodes memory utilization

Why do we need it?

Right now the existing kube-prometheus-stack doesn't have any alerts where average nodes memory utilization goes above a certain threshold value. I think this alert will be really helpful. In one of our clusters we do a manual scaling of nodes - so we have no way to know when we actually go ahead and scale the nodes. But, with the help of this alert, we can know that the time has come to scale the nodes.

Also, even if we do autoscaling of nodes - this alert will help us to verify the logic of autoscaling if that's happening correctly or not.

Apart for scaling point of view, i think adding this alert really makes sense.

Environment

• kube-prometheus version:

  v0.11.0

[philipgough]

@Aman1994, I wonder is it better to raise this issue in https://github.com/kubernetes-monitoring/kubernetes-mixin?