If there is a change with in the config map configuration in between the release and pod(s) get evaded or deleted during new Operator version start up it can corrupt the cluster upgrade process an old cluster nodes will start attempting to read new config map configuration meant for new cluster nodes.