Welcome to the Linux Foundation Forum!

Lab2.2 - Worker node not starting

mikelmaas
mikelmaas Posts: 5
edited November 2024 in LFD259 Class Forum

I'm using AWS EC2 instances with ubuntu version 24.04 for cp and worker nodes. CP node is started properly. Note: I can't select ubuntu 20.x on AWS, 22.x is the oldest.

The join command on the worker node doesn't work. When I check the node (kubectl get nodes) I get the following message:

E1118 16:20:38.157848 10649 memcache.go:265] "Unhandled Error" err="couldn't get current server API group list: Get \"http://localhost:8080/api?timeout=32s\": dial tcp 127.0.0.1:8080: connect: connection refused"

Please advise what todo. There are no files in the $HOME/.kube directory.

Answers

  • mikelmaas
    mikelmaas Posts: 5

    Can I use minikube to perform all the labs as an alternative?

  • chrispokorni
    chrispokorni Posts: 2,606

    Hi @mikelmaas,

    The join command on the worker node doesn't work

    Do you see any errors when running the join command? Please share them to help with troubleshooting.

    When I check the node (kubectl get nodes) ... There are no files in the $HOME/.kube directory.

    Are you running the kubectl command on the cp node? The $HOME/.kube directory and its content are missing on the cp node?

    Due to its automation and abstraction layers Minikube is not a recommended lab environment for this course. Several course exercises explore features that cannot be reproduced on Minikube.

    Regards,
    -Chris

  • mikelmaas
    mikelmaas Posts: 5

    Hi Chris,

    Thank you for your response. Attached are the logging of the worker script, and the output of some commands on the worker.

    The cp node has a correct .kube dir and contents. Kubectl get nodes on the cp node shows a running cp node.

    Best regards,
    Mikel

  • mikelmaas
    mikelmaas Posts: 5

    Note: the output of the worker logging is now based on the 22.04 Ubuntu version AWS offers. This is the oldest version which can be selected.

  • chrispokorni
    chrispokorni Posts: 2,606

    Hi @mikelmaas,

    With Ubuntu 20.04 LTS image only available as an AWS Marketplace AMI, picking the next available release from the Quick Start AMIs list makes sense. While Ubuntu 22.04 LTS has not been fully tested, it seems that 24.04 LTS works well for lab VMs.

    The errors you see are the expected behavior when running the kubectl get nodes command on the worker node. The missing .kube is also expected. Perhaps reading though the exercise description will help to clarify this.

    What happened after you executed the sudo kubeadm join command on the worker node? Does the node join the cluster? When provisioning the two EC2 instances, did you make sure to follow the video guide from the introductory chapter for VM configuration and SG rules?

    Regards,
    -Chris

  • mikelmaas
    mikelmaas Posts: 5

    Hi Chris,

    Thank you for your support, it is now working properly on 24.04. I didn't configure the security group properly, my bad. Apologies for wasting your time. I'm glad I can continue with the course now.

    Best regards,
    Mikel

  • humanbeing
    humanbeing Posts: 2

    And what was your problem there?
    I have the same issue.

    And which instruction have you followed? :)

  • chrispokorni
    chrispokorni Posts: 2,606

    Hi @humanbeing,

    Please share details about your setup, in order to determine what exactly is causing the "same issue". Similar misbehavior may be caused by different components that could be improperly configured for the lab environment.

    What hypervisor/cloud is provisioning your VMs? What is the guest OS? What is the VM's size: CPU, RAM, disk, how many and the type of network interfaces per VM, what are the VMs' private IP addresses? Any firewalls in place at hypervisor/cloud level - if yes, how is inbound traffic handled?

    Regards,
    -Chris

  • humanbeing
    humanbeing Posts: 2

    Hi, Chris,
    Thanks for the response.

    I've followed these instructions:
    AWS setup: https://trainingportal.linuxfoundation.org/learn/course/kubernetes-for-developers-lfd259/course-introduction/course-information?page=7
    Kubernetes setup: https://trainingportal.linuxfoundation.org/learn/course/kubernetes-for-developers-lfd259/files?page=7

    So I'm running on AWS with 2 instances of t2.large (2 CPU, 8Gb Ram), 30 Gb of storage.

    Firewall is setup as in instruction:
    All traffic for all protocols on all ports with source 0.0.0.0/0

    VMs private IPs:
    172.31.31.57
    172.31.26.51

    Entire setup and installation of the kubernetes worked fine.
    It fails on kubectl get nodes on worker node with error:

    "Unhandled Error" err="couldn't get current server API group list: Get \"http://localhost:8080/api?timeout=32s\": dial tcp 127.0.0.1:8080: connect: connection refused"
    The connection to the server localhost:8080 was refused - did you specify the right host or port?

    Any suggestions?

  • chrispokorni
    chrispokorni Posts: 2,606

    Hi @humanbeing,

    The lab guide has no such step, to run a kubectl command successfully from the worker node. The error you see is expected behavior, since the worker node is not configured with the credentials and context data required by the kubectl CLI.

    Regards,
    -Chris

Categories

Upcoming Training