Closed neo3matrix closed 10 months ago
Thanks @neo3matrix
To add a worker node, we can follow the docs : https://github.com/kubernetes-sigs/kubespray/blob/master/docs/nodes.md#addingreplacing-a-worker-node
If there are errors, it may be a bug.
If you are glad to, it's welcome to investigate and provide a PR to fix it.
@yankay That's the exactly same doc I followed as I mentioned above in my problem statement.
After some discussion with @nicolas-goudry on Kubespray slack channel, we found what may be the root cause:
I am adding a worker node which has dual stack networking setup (IPv4 & IPv6) to my regular IPv4 based k8s cluster.
Due to that, kubelet on this new node not able to connect with api-server due to invalid certificate
as the SAN on api-servers does not include [::1]
- no IPv6 support in default config.
My questions are:
enable_dual_stack_networks
variable to true? Or I need to change some other variables too?I tried setting enable_dual_stack_networks
to true on a fresh cluster setup but still not seeing any IPv6 support in my cluster SANs.
Someone please help. There is not enough info or steps out there about dual stack network nodes with Kubespray.
I tried setting
enable_dual_stack_networks
to true on a fresh cluster setup but still not seeing any IPv6 support in my cluster SANs.Someone please help. There is not enough info or steps out there about dual stack network nodes with Kubespray.
Could you open a new issue outlining the problem in a more detailed way ?
Regarding migrating from single to dual-stack, I don't think that's currently possible (though I haven't tried)
/close Please open a new issue regarding the dual stack problem if necessary
@VannTen: Closing this issue.
Hi everyone,
I used scale.yml with --limit="new-worker-node-hostname". The ansible command worked fine without any errors. But, on master node, when I run
kubectl get nodes
, I can't see the new worker node listed there. I even rebooted the worker node but no use.Environment: My setup is on-prem (Bare metal). I have one master and 7 worker nodes in my cluster.
printf "$(uname -srm)\n$(cat /etc/os-release)\n"
): All 8 nodes have Centos 7 running on them. The new worker node (that's supposed to be added to cluster) has Ubuntu 22.04 running on it.Kubespray version (commit) (
git rev-parse --short HEAD
): My Kubespray git version is locked to commit#18efdc2
Command used to invoke ansible:
On some forums, I read that instead of scale.yml, you should run cluster.yml Check NOTE section on this blog. But, as per official documentation on Kubespray git repo, cluster.yml is only used to add control plane node and scale.yml is used for adding worker node.
So, can someone please guide me on the same.