Has anyone upgraded an existing k3s to v1.33 and a...
# k3s
s
Has anyone upgraded an existing k3s to v1.33 and added IPv6 cluster-cidr and service-cidr ? Referring to this page, I have a couple of questions. I have both publicly routeable IPv6 address space and ULA IPv6 address space. Does it make sense to use ULA IPv6 for one of the networks and publicly routable IPv6 for the other?
c
theres generally not any point in using routable address space for pod or service CIDRs. Traffic on these ranges is never meant to leave the cluster without being NATed or encapsulated.
If you’re doing a CNI that doesn’t encapsulate, and you are trying to enable direct routing between clusters or between clients and pods/services you might consider it, but doing so is generally considered a bad idea.
s
Thanks Brad. I'm using the default Flannel CNI. I think that means you advise using ULA for both cluster-cidr and service-cidr and only using routable addresses for the LoadBalancer in Services, after adding
--flannel-ipv6-masq
. > generally not any point in using routable address space for pod or service CIDRs That sounds like excellent extra hints to enhance the paragraph on using ULA range addresses.