1. Proposed title of this feature request
   Cluster Autoscaler should take into account Node limits like EgressIPs
2.  What is the nature and description of the request?
   Currently, the autoscaler only has awareness for resources like cpu, memory, gpu, etc.  However there are some other resources which are node bound, one of them is for instance EgressIP on cloud providers.
   See https://docs.openshift.com/container-platform/4.17/networking/ovn_kubernetes_network_provider/configuring-egress-ips-ovn.html#nw-egress-ips-public-cloud-platform-considerations_configuring-egress-ips-ovn

3. Why would a customer need this ?
EgressIP are commonly used to create firewall access control lists. On large deployments, where a cluster can host thousands of namespaces, when each namespace has its own EgressIP, the number of EgressIP assignable to all nodes can be reached, and it has case it would make sense that the cluster autoscaler create a new node.

4. List any affected packages or components.
Cluster Autoscaler / OVN Kubernetes.

See also https://issues.redhat.com/browse/RFE-5050 as another approach for this problem.