Open Tejeev opened 1 month ago
When configuring dual stack, ipv6 first then ipv4 does not work.
This is supposed to work. Possibly fixed by https://github.com/k3s-io/k3s/pull/10511
Fix IPv6 primary node-ip handling The loadbalancer does not properly bind to the ipv6 loopback when the node has an ipv6 primary node ip, as the comma-separated flag value containing both IPs cannot be parsed as a valid ipv6 address.
Using ipv6 only is where the challenge is
This should also work as far as I know. What specifically is not working for you?
RKE2 supports IPv6, but Rancher does not.
Same question... what bits of rancher are you seeing that don't support ipv6? There were some bugs around ipv6 address handling in the past (specifically around properly encoding ipv6 address literals) but I thought they were resolved.
I was able to mock-up most of the questions above in my lab. However, the dual-stack support relied heavily on things like the ingress controller and CNI. In regards to the documentation, perhaps we can add a blurb about the versions where that first bullet point was fixed in the 10511 issue.
I think situations where customers have ipv6-only networks are typically specialty networks, so factors like Rancher become moot, because everything Rancher touches would need to be dual-stack or ipv6-only (like Github). It might be moot because these specialty networks are probably air-gapped and/or have local private repos, etc, so those issues may be resolved by the environmental context.
IPV6 is becoming more and more important. Our customers are asking more questions about it and we should flesh out our docs on the subject. Here are some very loose findings from someone on our support team which were provided to a customer in an attempt to help. This should be doublechecked and fleshed out for a doc.