We had an incident where the PRIMARY node was considered healthy internally and could communicate with other nodes but lost connectivity with other external services. In these type of cases it would be great to have external connectivity monitoring that can detect whether the PRIMARY is accessible from outside of the cluster.