Failed to read cluster ca certificate when joining worker node

Question

I am attempting to setup a multi-node k8s cluster as per this kOS Setup Link, but I face the error below when I try to join one of the nodes to the master node:

    k0s token create --role=worker
    WARN[2022-01-12 13:55:31] no config file given, using defaults
    Error: failed to read cluster ca certificate from /var/lib/k0s/pki/ca.crt: open 
    /var/lib/k0s/pki/ca.crt: no such file or directory. check if the control plane is 
    initialized on this node

I verified from the control node that this file does, exist however:

cd /var/lib/k0s/pki
ls

I am new to k8s setup from scratch, and a bit unsure which configuration item I need to fix (and where between master and worker nodes?). My research shows that A token is required when joining a new worker node to the Kubernetes cluster -- This token is generated from the control node. It also says that When you bootstrap a cluster with kubeadm, a token is generated which expires after 24 hours

When I try to check for the existence of a token on master node I get:

 kubeadm tokens list
-bash: kubeadm: command not found

I am unsure however if this is correct ( Is k0s even bootstrapped with kubeadm??).

However using k0s CLI syntax I can see that they are no tokens on the master:

k0s token list
No k0s join tokens found

My question:

1. What do I need to do for the file that is reporting as missing.
2. Is this error (possibly) related to the issue of token above and if so do I first need to regenerate the token at the master node end.
3. I have encountered multiple fixes at this Github Issue but I am not sure if any of them applies to my issue. Last thing I need is to break the current setup before I have even finished the cluster setup.

Environment Master node : Debian 10 Buster Worker node : Debian 10 Buster

Answer 1

I've got the same error as you when I tried to run k0s token create --role=worker on the worker node.

You need to run this command on the master node:

Next, you need to create a join token that the worker node will use to join the cluster. This token is generated from the control node.

First you need to run k0s token create --role=worker on the master node to get a token and later use this token on the worker node:

On the worker node, issue the command below.

k0s worker <login-token> 

So:

• generate a token on the master using k0s token create --role=worker
• use this token on the worker using k0s worker <login-token>

In my case I also needed to add sudo before both commands, so they looked like sudo k0s token create --role=worker and sudo k0s worker <login-token>

You wrote:

I am unsure however if this is correct ( Is k0s even bootstrapped with kubeadm ?? ).

No, they are two different and independent solutions.