TLDR; There is a test which checks kubelet probes work well for UDN-enabled pods by ensuring the no. of pod restarts is 0 which means the startup/readiness/failure probes are all healthy.... unfortunately given rhel8 workers are using cgroupsv1 where UDNs are not supported, pod's will indeed be restarting and probes are expected to fail here.
So the test has to be modified to take into account whether this is an environment with cgroupsv1 or v2 
however this is a test update bug, i can confirm there is no actual bug in the core code. This bug has been opened so that the tests can be fixed.

See https://prow.ci.openshift.org/view/gs/test-platform-results/logs/periodic-ci-openshift-release-master-nightly-4.18-e2e-aws-ovn-workers-rhel8/1889497310941941760 as example.