Got a weird one. When I deploy rke2 on vms the harvester node locks up. I am thinking it is from overlapping CIDRs. I have tried 1.5.1, 1.6.0-rc1. Ubuntu and Rocky10 in the vms. Next I am going to try k3s. Has anyone else seen this?

Recently we provisioned our cluster using openSUSE-Leap-Micro - since we figured using a SUSE image might give less friction and so far the nodes seem to go up faster and with fewer issues. Rocky we had a few issues with, and Debian was also problematic... I did try Ubuntu too but wanted to try something lighter weight and so far so good.

There my be an issue with rocky10 and Ubuntu 24.04 Plucky.. I am testing now with rocky9..

Good luck. We tried LTS and 9 from memory but I think they are a bit heavy for a k8s node - I think ideally you want something super light really, less overhead can’t be a bad thing

in the end, I think I don’t have enough CPU and networking…. 😞

Leap-micro has smaller footprint, but you need at least 3 nodes for your cluster for the control plane alone and you should have workers in addition to that. If your running locally for testing try rancher desktop - you can run a cluster on one machine and get quite a lot done.

I just spent the whole weekend installing/reinstalling/troubleshooting different distros network connectivity and it was all because of double NAT... morale - avoid double NAT at all cost :)