Cluster Operators using https://localhost:6443 leads to Alerts and KubePodCrashLooping

Issue

• When new API server revisions are generated, we see certain operators failing to connect to the API server causing alerts and KubePodCrashLooping
• During upgrades we see Alerts for openshift-kube-controller-manager, openshift-kube-controller-manager-operator, openshift-cluster-version-operator, openshift-kube-schedueler-operator and openshift-cluster-machine-approver
• In our operator logs, we see leaderelection issue connecting to the API server on localhost leaderelection.go:320] error retrieving resource lock openshift-kube-controller-manager/cert-recovery-controller-lock: Get https://localhost:6443/api/v1/namespaces/openshift-kube-controller-manager/configmaps/cert-recovery-controller-lock?timeout=35s: dial tcp [::1]:6443: connect: connection refused