-
Notifications
You must be signed in to change notification settings - Fork 1.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[rancher desktop] v0.13.0 not starting in Mac (Intel) #2765
Comments
can you check the node image is the one mentioned in the release notes? https://github.com/kubernetes-sigs/kind/releases/tag/v0.13.0
if it is , try to create the cluster with the |
Just a data point, this is on macOS 12.3.1:
|
Thanks for the quick response. I cleaned up whatever I could clean in my local docker, and kind create cluster still fails with the same error. The downloaded image seems to coincide with the image from @aojea :
Please find attached the logs: |
👁️
alpine + lima rancher? |
kubelet fails to start because is missing the kubelet cgroup path cc: @BenTheElder
there is also this error in the serial.log
|
Yes, I'm using https://rancherdesktop.io/ with |
first time I heard about it, didn't know it worked, nice |
ahh, I just checked, this is limavm under the hood , @AkihiroSuda I think we broke something with latest changes moving to systemd cgroup driver in lima |
IIRC docker desktop (not rancher desktop) is on cgroupsv2 for some time now, so while we checked with mac user(s) for docker desktop it did not cover this. We have coverage of In the meantime I think you'll find that v0.13.0 works if you use a |
We should confirm that |
In #2766 the issue looks is isolated between v1.24 and v1.23.6. It should also be possible to use the following to force cgroupfs instead of systemd: kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
containerdConfigPatches:
- |
[plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc.options]
SystemdCgroup = false
[plugins."io.containerd.grpc.v1.cri".containerd.runtimes.test-handler.options]
SystemdCgroup = false
kubeadmConfigPatches:
- |
apiVersion: kubelet.config.k8s.io/v1beta1
kind: KubeletConfiguration
cgroupDriver: cgroupfs |
Interesting, in github actions docker / cgroupv1 we see in node serial:
Which is expected from #2709 but then in the logs here I see:
So this seems to be broken when systemd not running on the host on cgroups v1. (seems to imply rancher desktop uses openrc instead of systemd). Kubelet log as mentioned above:
|
So we're failing here because the So unlike our bad release note assertion, currently systemd is once again required on the host when running with kind + Kubernetes v1.24.0. We should roll forward with a fix, and update the release notes to reference this issue + workaround #2766 (comment) (the config above #2765 (comment)). |
|
I have similar issues on WSL 2 (docker installed as linux service) and below command works like a charm |
Hi folks, sorry for the delay. I've tested an updated fix in #2767 on colima xref #2778 which should be the same root issue (kind v0.13.0 / v1.24 imagse brokene on non-systemd hosts) Please let me know if this updated image works with kind v0.13.0 Hopefully that should be it, and we'll cut another release soon with this fix. |
Hi @BenTheElder ! I just verified that this image works with a fresh installation of Rancher Destkop. Thank you for addressing this issue! |
@BenTheElder Thanks you, works for me as well |
With v0.13.0 trying to create a cluster in Mac OS X fails due a timeout when trying to check the health of the control plane. I reverted back to v0.12.0 and I could create it without any problem.
See logs at the end of the message.
kind version
): v0.13.0kubectl version
): Client Version:docker info
): Using Rancher Desktop for Mac (Intel)OS (e.g. from
/etc/os-release
): Mac OS XLogs:
Logs:
The text was updated successfully, but these errors were encountered: