"list access denied" when switching contexts within k9s since 0.32.0

[tibeer]

Describe the bug
Switching between multiple kube contexts that belong to the same cluster trows an error since 0.32.0

To Reproduce
Steps to reproduce the behavior:

1. Create two kube config files that contain each one context for the same cluster but for different namespaces (e.g. context_one: namespace a, context_two: namespace b)
2. export the environment variable like this: export KUBECONFIG=/Users/foo/.kube/context_one.yaml:/Users/foo/.kube/context_two.yaml
3. select with kubectl config use-context context_one
4. start k9s and view some resources of context_one
5. use ctxinside k9s to switch contexts
6. try to view resources
7. see the error

Historical Documents
N/A

Expected behavior
Context switchting from within k9s is possible without any errors

Screenshots
see above

Versions (please complete the following information):

• OS: OSX
• K9s: 0.32.0
• K8s: 1.20 (customers are scared of updates :)

Additional context
N/A

[lennan-norlys]

I see almost the same issue after updating to k9s 0.32.0, and the same behavior on 0.32.1

In my use case, I have multiple contexts for different clusters, but the namespace is namespace_a in every context.

I'm able to switch contexts using kubectl config use-context xyz, making me confident about this being an issue relating to k9s.

I noticed the error code indicating that k9s is trying to use default namespace, which makes this "expected behavior" because my user is limited to a single namespace.
This makes me think k9s cant use/read the namespace from kubeconfig when switching context thus resorting to default.

These are my observations. Unfortunately, I'm not familiar with the k9s code base, so I really don't have any suggestions for a solution.

[derailed]

@lennan-norlys Thank you for the additional details!
Could you post your logs k9s -l trace. So we can fish this out?

[@tibeer] what does this yield: k auth can-i list pods -n default --context ctxXXX using the context from your repro

[tibeer]

As expected: "no" 😄
But as far as I am aware, we do not even have the default namespace any more. I guess it was deleted right after cluster creation, but I do not have insight into this info.

[lennan-norlys]

@derailed This should be the file you're asking for.
k9s.log

[tibeer]

Awesome! Thanks a lot 🚀

[lennan-norlys]

The latest release solved my issue. Thanks alot!