-
Notifications
You must be signed in to change notification settings - Fork 30
Unable to run kind node in k8s pod on CoreOS #2646
Comments
Thanks for the report. I'm a bit confused here as I don't think we ship any |
Sorry for the confusion. The unit is generated by kubeadm and looks something like this:
This and also the containerd service inside the kind container are killed/crashing every 50s and I am not sure what to look for here... |
hi @lucab (kind developer here), nice to meet you again :) |
Thanks, for clarifying @aojea |
So, if I understood correctly, there is a dedicated systemd session running in a pod, and the I would probably have a look at |
Yes, exactly. That is the problem, which I am currently stuck at. I attached some journal logs from the Here is some output from
|
Issue Report
Bug
I am currently trying to run a kind node inside a Kubernetes Pod running on a CoreOS Node.
The Pod manifest looks like this:
I am able to bootstrap the node, join the cluster and so on...
But about every 50 seconds, the systemd services for
kubelet
andcontainerd
inside the Pod are restarted after messages from the kubelet looking like this:Therefore the Node can't stay ready and can't run any Pods.
Container Linux Version
Environment
What hardware/cloud provider/hypervisor is being used to run Container Linux?
aws
Expected Behavior
systemd services should keep running inside a docker container.
Actual Behavior
systemd services are killed about every 50s.
Reproduction Steps
kindest/node
image on a CoreOS NodeOther Information
I already read through the issues over at kind (mainly kubernetes-sigs/kind#303 and kubernetes-sigs/kind#890) and tried all the suggestions, but none of them seem to fix the problem on CoreOS.
/cc @BenTheElder
/cc @aojea
The text was updated successfully, but these errors were encountered: