Master nodes are not ready on an interrupted 3.10-->3.11 upgrade

Issue

• A 3.10-->3.11 cluster upgrade has been interrupted in a moment such that:

  • atomic-openshift-node has been upgraded to a 3.11 as per rpm -qa | grep openshift | sort
  • Master components have not been upgraded yet, as per: grep image: /etc/origin/node/pods/* | grep -v etcd output on master

• On a 3.10 cluster, an atomic-openshift-node package has been accidentally upgraded to 3.11 while masters stay at 3.10.

• Error in "diagnostic steps" is displayed in atomic-openshift-node logs.