[Feature Request] detect possible down time because all the replicas are running in the same minion

[ryuheechul]

Check each deployments’s pod’s physical locations (minion or node’s ip)
And if the location is only one then we can say it’s a possible danger and report to the alert system.

Why this feature?
A web service on our k8s cluster was down for few minutes and I found out it’s because all three pods were in the same minion and that minion wasn’t available for few minutes

see also: https://kubernetes.io/docs/concepts/configuration/assign-pod-node/#affinity-and-anti-affinity