This is a clone of issue OCPBUGS-22689. The following is the description of the original issue:
—
This is a clone of issue OCPBUGS-22210. The following is the description of the original issue:
—
This is a clone of issue OCPBUGS-14010. The following is the description of the original issue:
—

Extended api-server disruption periods during upgrades on Azure, detected by TRT analysis, led to the finding that non-fatal etcd delays were causing an api-server 503s.

https://github.com/openshift/cluster-openshift-apiserver-operator/blob/00f7e4cc95063ba5aba1992568088d924cfbf516/bindata/v3.11.0/openshift-apiserver/deploy.yaml#L137 shows the current readiness check permits only one failure after 1 second. Suggesting we backport OCPBUGS-14010 to make this 10 seconds and more forgiving of temporary etcd issues.

This prowjob shows the behavior:

https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/origin-ci-test/logs/periodic-ci-openshift-release-master-ci-4.14-upgrade-from-stable-4.13-e2e-azure-sdn-upgrade/1706399977418264576/artifacts/e2e-azure-sdn-upgrade/openshift-e2e-test/artifacts/junit/e2e-timelines_everything_20230925-210342.html