Honor KUBECONFIG

[Kyle Tarplee]

It appears that the environment variable KUBECONFIG is not honored. In fact I have both KUBECONFIG set and ~/.kube/config exists and that seems to break it. I had to unset KUBECONFIG for this awesome tool to work. KUBECONFIG should take precedence so the behavior match kubectl, I believe.

$ kubectl view-allocations
[2020-04-22T14:35:21Z ERROR kubectl_view_allocations] failed 
    cli: CliOpts { namespace: None, show_zero: false, resource_name: [], group_by: [resource, node, pod] }
    error: failed to load kubectl config
    
    Caused by:
        0: failed to load kubeconfig
        1: Error loading kube config: No such file or directory (os error 2)
✔ ~/code/devsecops/ace-hub-api [master|✔] (⎈ |meerkat:kyle)
$ unset KUBECONFIG
✔ ~/code/devsecops/ace-hub-api [master|✔] (⎈ |meerkat:pachyderm)
$ kubectl view-allocations
 Resource                                                         Requested  %Requested      Limit  %Limit  Allocatable  Free 
  cpu                                                                19170m         68%     57960m    207%           28     0 
  ├─ meerkat1                                                         1100m         28%      6800m    170%            4     0 

Of course kubectl works fine with my KUBECONFIG set. It is set to multiple files, colon delimited.

kubectl-view-allocations is version 0.7.0

[Kyle Tarplee]

This fails on 0.7.2 as well.

[Kyle Tarplee]

I am also having issues without having KUBECONFIG set. Without KUBECONFIG set this worked on 0.7.0 but it no longer works on 0.7.2

I build master b9b15b9 and it gives me a different error (same config file).

2020-05-22T12:35:20Z ERROR kubectl_view_allocations] failed 
    cli: CliOpts { namespace: None, show_zero: false, resource_name: [], group_by: [resource, node, pod] }
    error: failed to collect info from nodes
    
    Caused by:
        0: Failed to list nodes via k8s api
        1: ApiError: "404 page not found\n": Failed to parse error data (ErrorResponse { status: "404 Not Found", message: "\"404 page not found\\n\"", reason: "Failed to parse error data", code: 404 })

How do I see the URL being used to make the request? Is there a way to increase logging?

[Kyle Tarplee]

Using a debugger I was able to track this issue down to an issue in "kube". My cluster uri includes more than the hostname (i.e. https://hostname/foo/bar) because I am using rancher.

let uri_str = self.cluster_url.join(pandq.as_str())?;

pandq is "/api/v1/nodes?". The problem is that the resultant uri_str is "https://hostname/api/v1/nodes?" and not what it should be, namely "https://hostname/foo/bar/api/v1/nodes?"

I created an issue in "kube" kube-rs/kube#244

[David Bernard]

Thanks for tracking the issue and reporting the root cause.
Sorry for being quiet.

[Kyle Tarplee]

This now fixed upstream in kube. You need to update the dependency to 0.35.0

[David Bernard]

Thanks, I'll do a release after work.

[David Bernard]

0.7.3 released

Thanks for your contribution, feel free to reopen if the fix doesn't work

[Kyle Tarplee]

It worked. Thanks.