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
Fail to create cluster locally with linux-lqx
as kernel on Garuda (~Arch) Linux
#2852
Comments
@caniko can you provide a little more detail or context? Kind should work fine with cgroupv2. There were some older releases that did not. Can you provide the version ( |
cgroupv2 is supported, but Kubernetes does not support it before v1.19 https://kind.sigs.k8s.io/docs/user/known-issues/#failure-to-create-cluster-with-cgroups-v2 KIND has has cgroupv2 since v0.10.0 / January 2021, not long after Kubernetes, with some fixes since then https://github.com/kubernetes-sigs/kind/releases/tag/v0.10.0 We test cgroupsv2 on every PR. |
Oh, it says cgroup v2 issues on the website. I understand that I interpreted it wrongly now. kind version: kind v0.14.0 go1.18.2 linux/amd64 I am on Arch Linux, and behind a proxy. Docker and K8 runs fine. I succesfully ran minikube in kvm2 mode. Could the proxy be blocking the health check URL? Stack trace:
|
Can you share the logs from |
It's possible it's the proxy but we should be setting sufficient no_proxy, we can check what the component logs show. |
The stack trace is too long. Should I have saved it to a file? Here is the tail:
|
The stack trace does not contain all the node logs, just kind / kubeadm but #2852 (comment) would export many more log files to inspect |
The error / trace tells us kubelet didn't become healthy, but to find out why kubelet didn't become healthy we need to dig deeper into the system. |
There were no log files, could you be more explicit on what I should provide? I made sure to run the command you provided. |
Maybe you should break out the commands to one at a time.
|
I just noticed one line where it said exported to /tmp/ Root of logs: Control-plane logs: |
could you upload a zip or tarball with the full directory? there's some other useful files and github will accept a zip or tarball. |
Two things jump out from what we have so far:
And then there's this in the kubelet logs
(similar logs in containerd) Does not look proxy related. |
I did not know that the kernel was an important component for K8! I am using linux-lqx-bmq, which is downstream from zen-kernel. Switching back to zen fixed the issue. Garuda is Arch with some desktop environment sugar. I made the switch from zen to lqx, it is my fault. |
linux-lqx
as kernel on Garuda (~Arch) Linux
It's possible the custom kernel lacks some cgroups support (CPU weight?), IIRC previous problems with Arch kernels were that sort of thing. Since the kind "nodes" are running on the shared host kernel, it must meet Kubernetes's requirements in terms of e.g. supported cgroup controllers. |
This is a known limitation in the kernel damentz/liquorix-package#101 (comment) |
Is Groups v2 a hard requirement? Could we create the kind cluster with cgroups v1? |
It isn't yet, some parts of the ecosystem may eventually drop support. Cgroups of either v1 or v2 are. |
does it mean, some systems which support only cgroups v1 can use kind? |
yes, but has to supports all the cgroups "groups" required by kubelet, some arch like old raspberry models doesn't support it per example |
This issue is now resolved with Liquorix by implementing the missing stubs that Relevant links: |
Why is this needed
cgroupv2 is supported by k8, should also be supported by kind.Revised issue:
I am having an issue setting up a kind cluster
The text was updated successfully, but these errors were encountered: