I'm just trying to install the same version we have in production.

Oh, I've had a look at the logs directly on the controlplane node (that I'm trying to make to join the cluster), and I get this:

ERROR: The environment variable CATTLE_CA_CHECKSUM is set but there is no CA certificate configured at <https://rancher.example.com/v3/settings/cacerts>

So I went to the control plane node itself (the one I'm trying to register, as I said), and looked at the pods status and came across this

But I have a hunch about why this might not work. 🙂

I thought I could get away with it. The problem is that I'm using the dns name that I've added in /etc/hosts, because it still points to a different cluster altogether (and there's not much I can do about it right now, but I do need to use that name) So this worked for the initial registration (the command which was run, of course, on the host directly), but this pod is definitely using coredns, which doesn't check /etc/hosts of the host itself. So it must be going to the other IP... eh

And, yes, I've just confirmed that with a tcpdump. Eh... I'll have to wait.

Funny, I've add a hosts directive in the coredns configmap and now it works.

And the worker node also joined automatically (there was no pod there that required accessing the management server directly, I guess)

Ah, I'm silly. The configmap will work for all nodes, of course, 'cause it's the same. I wondered why all the other nodes, including the control plane ones, worked without issues.