Based on the issue described here: https://github.com/kubernetes/kubernetes/issues/133442

Setting the priorityClassName field on static pod definitions have no impact on the shutdown order as kubelet ignores the field for static pods. This causes the static pods to be terminated in the first round of terminations rather than at the time specified by the priorityClassName.

This renders the gracefulShutdown order almost useless for Single Node Openshift because we lose kube-apiserver and kube-etcd right away.

I have opened the PRs below to fix this issue. Could we please merge and immediately backport this fix ASAP as our customers are facing very long shutdown times, shutdown hangs, and forced shutdowns impacting the storage layer for SNO environments because on this issue.

https://github.com/openshift/cluster-kube-apiserver-operator/pull/1915
https://github.com/openshift/cluster-etcd-operator/pull/1476
https://github.com/openshift/cluster-kube-controller-manager-operator/pull/865
https://github.com/openshift/cluster-kube-scheduler-operator/pull/572

Describe the impact to you or the business
Long shutdown times and storage layer problems caused by forceful termination from graceful Termination not being respected. 

In what environment are you experiencing this behavior?
All SNO environments

How frequently does this behavior occur? Does it occur repeatedly or at certain times?
Every shutdown 

    1.
    2.
    3.