Yesterday we've added additional LB IP range, yet the change never propagated to /etc/vmware/anyip-routes.cfg :
Once some K8s service was published (with IP chosen sequentially from new range) and a bind entry was added to haproxy.cfg, haproxy.service crashed and couldn't start anymore: "cannot bind socket [10.20.30.128:port]".
Appending new range 10.20.30.128/25 via SSH manually restored normal operation.
While there is a VMware support ticket raised, a GitHub issue won't hurt.