05/19/2023, 6:36 PM
Hello! Following my initial message on #general, I'm redirecting my RKE2 specific questions here in a thread
RKE2: • while RKE2 is CNCF compliant, I was surprised to see that, contrary to K3s, it isn't considered a CNCF project per se. Does this mean anything governance wise? • Is it possible to do away with the additional features on top of the vanilla experience like the helm controller? In particular, I'd rather follow an upstream cilium chart instead of the bundled one • Deployment wise, we're evaluating cluster api. We've seen, and rancher manager docs mention using CAPI to deploy RKE2. Is this what is used by rancher? Basically, we're already familiar with vanilla K8S, and are mainly interested in RKE2 for the hardening and compliance parts, not so much the comfort features. In this situation, is RKE2 a good fit for us?


05/19/2023, 8:57 PM
1. It's owned by suse, not cncf. Only k3s was donated to cncf. At the moment it means very little, as we have not had anyone outside suse step up to help maintain k3s so both projects are maintained by the same team, and neither have required much action by way of governance. Note that distros do not have to be owned by cncf to be certified as compliant, in fact very few are. 2. Yes, you can disable the bits you don't like. 3. No, rancher has its own provisioning framework based on capi that it uses to provision nodes. That standalone capi provisioner is experimental and not currently used by any of our supported products. This is more of a rancher question than a rke2 question though.


05/19/2023, 9:23 PM
thank you very much for your responses! Indeed, CAPI might be a tad young to be used industrially. So the recommended ways to install for now are either through rancher of through the tarball/script? I'll patiently wait for answers from the rancher team on the other channel 🙂