feature request - please provide an option to use a cidr for the whole cluster

[kube-vip]

feature request - please provide an option to use a cidr for the whole cluster

[thebsdbox]

This is certainly something that could be done:

The replicaset for kube-vip has a bunch of anti-affinity rules to ensure placement for a "few" instances whilst always ensuring only a single instance per underlying node. So I'm not sure what benefit moving to a daemonset would bring other than more instances for higher availability?

The "global cidr" is a nice idea, I wanted to provide "multi-tenant" VIP ranges per namespace so that applications could sit in different ranges OR use VIPs on a different interface to other namespaces. But a fall-back CIDR range shouldn't be too complicated to implement.

[vrabbi]

Any update on this feature? A global cidr would be awesome and is currently a blocker for our environment for moving from metallb to kube-vip

[thebsdbox]

No worries I'll look into it this weekend. 🙂

[vrabbi]

Awesome! Thank you! This is a great project and really looking forward to using it