AWS Classic Load Balancer created for default ingresscontroller missing couple of AWS availability zones after OpenShift upgrade

Issue

• Availability zones of AWS classic load balancer in AWS gets changed after upgrade.
• Couple of worker nodes are missing from AWS classic load balancer endpoint list after cluster upgrade.
• All ingress traffic is being forwarded to a single router POD even though there are multiple replicas running and balance mechanism is set as roundrobin.
• After cluster upgrade all worker nodes are not added as endpoint to AWS classic load balancer created to forward traffic to router PODs created as part of default ingresscontroller deployment.