In certain cases, such as when PostgreSQL isn't available for UPS, it won't start properly. The Wildfly/EAP server process won't crash though, so the Pod will remain alive indefinitely, but the UPS container will never get to a "Ready" state (there is a Readiness probe already, but that failing won't cause anything to get restarted).

Adding a liveness probe will get k8s to restart the container (or Pod?) if the liveness probe doesn't return successfully after the given conditions. This means that manual intervention to delete the Pod won't be necessary.