Hello there everyone! I suppose some of you are us...
# harvester
f
Hello there everyone! I suppose some of you are using rancher + harvester and also using the rancher provider to spin up child rke2 clusters. Have you encountered an issue with the deployment, in which the kubeconfig does not contain the namespace where the child nodes are created, thus the harvester CCM fails to find the nodes and properly initialize them? In my case, the issue is solved by adding the namespace in the context within the kubeconfig file, but this is a manual step which I need to do on every node being created, which is something I'd like to avoid. Versions being used: Rancher: 2.12.0 Harvester: 1.5.0 Rancher provider: 8.0.0 OpenTofu: 1.10.something
m
from what i can see, it doesn't look like rancher would include the namespace in the context: https://github.com/rancher/rancher/blob/c4321ee15ff7bbe6345871765508cfceb46978d1/pkg/kubeconfig/templates.go#L36-L42
when i tried to emulate the
data.http.kubeconfig
data source in your issue with curl, the response didn't account for the namespace, service account name etc. included in the post body
f
Exactly @millions-microphone-3535, which then causes the Harvester CCM to not run properly.
m
might wanna update your issue with this finding
f
I added a comment in the issue, but this is definitively not a TFR2P issue, more of a rancher issue, so perhaps I will create a new issue in the rancher repo. At least I will check if someone else encountered the same behavior as me.