Note: I’ve created a detailed issue at cluster machines fail to initialize - containerd: "unknown service runtime.v1.RuntimeService" · Issue #220 · lxc/cluster-api-provider-incus · GitHub
If this is considered crossposting, I’d like to apologize beforehand. I merely hope to harness the forums creative energy 
When creating a k8s cluster on IncusOS using cluster API and the CAPN provider for Incus, the machines fail to initialize when they are not started in an VM, but in a container.
The error originated from containerd, which seems to start without error, but it’s not possible to communicate with it: crictl info
FATA[0000] validate service connection: validate CRI v1 runtime API for endpoint "unix:///run/containerd/containerd.sock": rpc error: code = Unimplemented desc = unknown service runtime.v1.RuntimeService
This is also the primary issue that prohibits the kubeadm init pre-flight checks to succeed.
The same thing works fine when deploying in VMs. Also, deploying a k8s cluster on IncusOS using kini works fine as well, so it doesn’t seem to be a ‘nested container’ type issue.
I’m grateful for any input
P.S.: amazing projects, thanks for all the effort 