1. Proposed title of this feature request

Readiness probe on openshift-dns should test upstream DNS servers

2. What is the nature and description of the request?

As of now, the readiness probe on the openshift-dns pods only takes into consideration if CoreDNS itself is ready, but doesn't check if the upstream servers configured in the cluster are reachable.

Hence we are requesting that the readiness probe of openshift-dns pods also marks the pod as not ready if it cannot access any of the upstream DNS servers, so that pods on the node remain able to perform external DNS resolution via the openshift-dns pods of the other nodes.

3. Why does the customer need this? (List the business requirements here)

Customer had an incident and a very relevant part of it involved a concrete node not being able to reach the upstream DNS servers. If the openshift-dns pod had marked itself as not ready during the incident, the pods would have remained able to do DNS resolution through the other openshift-dns pods of the cluster and the impact of the incident would have been much lower. However, the openshift-dns pod remained ready, so we could not leverage openshift-dns high availability to mitigate the issue.

4. List any affected packages or components.

openshift-dns