Path to k8s-ca.crt is relative

[quarckster]

kubent fails to start if it's not executed in ${HOME}/.kube directory:

kubent -x <SOME CONTEXT> --helm3
11:12AM INF >>> Kube No Trouble `kubent` <<<
11:12AM INF version 0.7.0 (git sha d1bb4e5fd6550b533b2013671aa8419d923ee042)
11:12AM INF Initializing collectors and retrieving data
11:12AM ERR Failed to initialize collector: <nil> error="failed to assemble client config: invalid configuration: unable to read certificate-authority certs/<SOME CLUSTER>/k8s-ca.crt for <SOME CLUSTER> due to open certs/<SOME CLUSTER>/k8s-ca.crt: no such file or directory"
11:12AM ERR Failed to initialize collector: <nil> error="failed to assemble client config: invalid configuration: unable to read certificate-authority certs/<SOME CLUSTER>/k8s-ca.crt for <SOME CLUSTER> due to open certs/<SOME CLUSTER>/k8s-ca.crt: no such file or directory"
11:12AM INF Loaded ruleset name=custom.rego.tmpl
11:12AM INF Loaded ruleset name=deprecated-1-16.rego
11:12AM INF Loaded ruleset name=deprecated-1-22.rego
11:12AM INF Loaded ruleset name=deprecated-1-25.rego
11:12AM INF Loaded ruleset name=deprecated-1-26.rego
11:12AM INF Loaded ruleset name=deprecated-future.rego

I tried both a binary and a container the error is the same. Only when I change the current working directory to ${HOME}/.kube kubent works.

[github-actions]

This issue has not seen any activity in last 60 days, and has been marked as stale.

[stepanstipl]

Can we see the relevant part of your .kubeconfig? I assume that perhaps the certs there are referenced as relative to ${HOME}/.kube?

[quarckster]

Can we see the relevant part of your .kubeconfig? I assume that perhaps the certs there are referenced as relative to ${HOME}/.kube?

...
clusters:
- cluster:
    certificate-authority: certs/<REDACTED>/k8s-ca.crt
    server: https://<REDACTED>
  name: <REDACTED>
...

Good catch, but anyway, this is something not obvious.

[stepanstipl]

Thanks for confirming this.

I did check how kubectl behaves, and it seems it will resolve paths relative to the kubeconfig location. We should probably try to do the same.

[femrtnz]

Hi quarckster, Are you still facing this issue? I wasn't able to simulate, please check below running from any directory:

$ k config get-contexts
CURRENT   NAME               CLUSTER            AUTHINFO           NAMESPACE
*         main_gke_cluster   main_gke_cluster   main_gke_cluster   

$ kubent -x main_gke_cluster  --helm3=true
3:15PM INF >>> Kube No Trouble `kubent` <<<
3:15PM INF version 0.7.0 (git sha d1bb4e5fd6550b533b2013671aa8419d923ee042)
3:15PM INF Initializing collectors and retrieving data
3:15PM INF Target K8s version is 1.31.1-gke.1846000
3:15PM INF Retrieved 0 resources from collector name=Cluster
3:15PM INF Retrieved 0 resources from collector name="Helm v3"
3:15PM INF Loaded ruleset name=custom.rego.tmpl
3:15PM INF Loaded ruleset name=deprecated-1-16.rego
3:15PM INF Loaded ruleset name=deprecated-1-22.rego
3:15PM INF Loaded ruleset name=deprecated-1-25.rego
3:15PM INF Loaded ruleset name=deprecated-1-26.rego
3:15PM INF Loaded ruleset name=deprecated-future.rego