Welcome to the Linux Foundation Forum!

Kubeadm failed to join

jasonseah
jasonseah Posts: 3
in LFS258 Class Forum

I realise this might be the same issue with the other post with the same title, but the solution was updated firewall rules to allow all ports.

I already did that but this issue consist the same, I tried restart the VM but it didn't do anything. I don't mind to redo the whole thing but I wish to understand the issue here and hopefully this will help someone there is new like me.

Here is the spec for both cp and worker (e2-medium)

The screenshot of the error
the screenshot of the error

The screenshot of the CP VM & Firewall settings

The screenshot of the VPC network

The screenshot of the Firewall on VPC network

Answers

  • chrispokorni
    chrispokorni Posts: 2,606

    Hi @jasonseah,

    Thank you for providing infra and CLI details. While the VPC and firewall look ok, However, I am concerned about the hardware profile of your instances. e2-medium is a 2vCPU/4GB mem, and the video guide from Chapter 1 is calling for 2vCPU/8GB mem which would be an e2-standard-2. Sometimes Kubernetes components panic when they do not have enough resources.

    Also, the worker VM should be on the same network as the cp VM, meaning in the same VPC, under the same firewall. This was not provided above, but it is worth mentioning as it may be another reason for initial failures.

    Configuring multiple runtimes, as in both docker and cri-o, will also cause Kubernetes agents to panic. Please make sure you only use the recommended runtime for this course release - the docker runtime.

    When bootstrapping the cluster, please follow the Kubernetes version found in the Chapter 3. A misconfiguration of the docker/daemon.json file may produce errors as well.

    The /etc/hosts files of both VMs are expected to include the private IP of the cp node and the k8scp alias.

    Hope this helps to get the issues resolved.

    Regards,
    -Chris

  • jasonseah
    jasonseah Posts: 3

    Hi @chrispokorni,

    Thanks for the reply, and you have my thanks! It solve the problem but for the record if someone fell for the same mistake as I am I am going to write down what I went through.

    for the VM instance, yes it was my mistake for using a smaller machine, but I do have a separate worker VM running the same VPC network as the CP, the reason I did not mention is is because since both are having the similar setting which created by "create similar button".

    Configuring multiple runtimes is not the problem happen for me, both are using docker.

    the /etc/hosts for both VMs are using the same private IP of the cp node and with k8scp alias

    For bootstrapping the cluster, this is what cause the issue, when I pasted the configuration that I copied from the PDF, it was a one liner json, so I didn't think that much I just saved it and restart the docker so apparently one liner json won't work for this case, I have to retype it and restart the docker, then suddenly it works!

    Thank you so much Chris! you have just made my day!

    Regards,

    • Jason Seah
  • chrispokorni
    chrispokorni Posts: 2,606

    Hi @jasonseah,

    Glad it all works now!

    However, keep in mind that you have the SOLUTIONS tarball, where most config files are available for you as .yaml, .sh, .json, .cfg, etc... So there is no need to copy file content from the PDF - this process typically introduces formatting issues such as the one you reported above.

    Regards,
    -Chris

  • jasonseah
    jasonseah Posts: 3

    Hi @chrispokorni

    Thanks for the reminder! I will use those from now on!

    Regards,
    Jason Seah

Categories

Upcoming Training