While working with a customer on the cross cluster live migration feature, it was discovered that they will have trouble allocating contiguous blocks of ip addresses for virt-handler/virt-synchronization-controller to allow cross cluster live migration. Instead they mentioned that they would have enough ip addresses but they would not be in a block.

There is currently no mechanism available to specify a list of ip addresses that will get automatically assigned to pods (on a secondary interface, not the pod network) from a daemonset/deployment such as the virt-handler/virt-synchronization controller. It is not important which pod gets which ip address. Just that the ip addresses are assigned to the pods.

It is not viable to select a large cidr range, and then excluding everything but the IP addresses we have access to, that will be hard to manage. Also no DHCP is available on the network.