In https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/periodic-ci-openshift-release-master-ci-4.8-e2e-gcp-upgrade/1375890420118065152

during node upgrades the oauth-openshift-...-vrmzm pod is marked as deleted at 20:59:42 and is not fully deleted 50m later. This causes TargetDown to fire because the pod is unready and is never completed.

Mar 27 20:59:42.790 W ns/openshift-authentication pod/oauth-openshift-b88f6b558-vrmzm node/**************-17f95-gdtt6-master-1 reason/GracefulDelete in 40s

Marking high because this is a bad state to get in and effectively wedged the whole upgrade process. May be similar to other failures we are seeing in CI (such as the pod submit test still failing occasionally.

This is a release blocking bug and is potentially high impact to CI (although I have not quantified impact yet)