We fixed #4759 but there are still some edge cases if we need to restart more than 1 container in the same pod. We need to improve our handling of these cases. Any time we have to restart the POD infra container, we must restart all containers in the pod. This may run against restart settings.