Currently, OpenShift's egress router lacks a built-in high availability (HA) feature. The EgressRouter Custom Resource (CR) doesn't allow for specifying replicas, meaning you can't create multiple egress router Pods for HA. This is because the EgressRouter CR generates a Network Attachment Definition (NAD) object that provides a single IP address, which can only be bound to one Pod at a time. We've received numerous customer requests for HA with multiple egress IP addresses for the OpenShift EgressRouter.

Technically, this is feasible. The EgressRouter CR can accept a pool of IP addresses for the egress address pool, defining it via Whereabouts IPAM. With this approach, each egress router Pod would use a dedicated IP address from the Whereabouts pool.

For customers impacted by this, the OpenShift egressIP feature can not serve as an alternative until the RFE Allow non-default routes on the OVN-Kubernetes gateway logical router is implemented.