08/19/2022, 10:34 PM
When I try to install Rancher Server 2.6.6 on k3s > 1.24.0 it fails because the limit is set in the Helm Chart for that k3s version. However, if I install k3s <1.24.0, install Rancher Server 2.6.6, then upgrade the underlying k3s, I can use something > 1.24.0. How dangerous is this and does this suggest Rancher Server should always be upgraded prior to it’s underlying k3s cluster?
This makes me think I need to add the same Helm version check into my Ansible that manages k3s


08/20/2022, 3:34 AM
It’s somewhat dangerous, especially depending on what’s changed in the newer version of k8s
Typically I would upgrade rancher first


08/20/2022, 5:06 AM
It does not install on that version because it's not supported on that version. Do you want to use something unsupported?


08/22/2022, 8:45 PM
No I definitely don’t WANT to be able to run this unsupported. I like the Helm check, I was just pointing out a loophole in this flow and was considering if something could be done to prevent this unsupported state from occurring. Because if it was bad enough you’d either 1) have to rebuild the cluster and restore and/or 2) wait until a Rancher Server version came out that supported your underlying k3s version because the Helm check would prevent you from changing doing another install/upgrade until the condition was satisfied.