[Windows] Unable to connect to clusters using kubeconfig

[yanrez]

Description

When I pick cluster on headlamp startup, I'm getting error: Failed to get authentication information: Bad Gateway If I use kubectl manually, it is able to connect.

Impact

Can't use headlamp since it can't connect to my clusters

Environment and steps to reproduce

Headlamp: 0.6.0.0 OS: Windows 11

1. I launch headlamp C:\Users\yanrez\AppData\Local\Programs\headlamp\Headlamp.exe see censored stdout below
2. Pick cluster
3. Get error 'Failed to get authentication information: Bad Gateway'

Expected behavior

Able to connect and load UI

Additional information

❯ 13:48:30.614 > App starting... 13:48:31.359 > server process stderr: 2022/01/12 13:48:31 plugins-dir: C:\Users\yanrez\AppData\Roaming/Headlamp/plugins

13:48:31.364 > server process stdout: Headlamp Server API Routers:

13:48:31.364 > server process stdout: localhost:4466/clusters/miniaks/{api...} -> https://.hcp.eastus.azmk8s.io:443 localhost:4466/clusters/kind-kind-linux-cluster/{api...} -> https://127.0.0.1:61042 localhost:4466/clusters/x055-admin/{api...} -> https://.hcp.northcentralus.azmk8s.io:443 localhost:4466/clusters/x057-admin/{api...} -> https://.hcp.northcentralus.azmk8s.io:443

13:48:31.365 > server process stdout: localhost:4466/clusters/docker-desktop/{api...} -> https://kubernetes.docker.internal:6443 localhost:4466/clusters/kind-kind/{api...} -> https://127.0.0.1:55972

13:48:31.367 > server process stderr: 2022/01/12 13:48:31 listen tcp :4466: bind: Only one usage of each socket address (protocol/network address/port) is normally permitted.

13:48:31.373 > server process process exited with code:1 signal:null

[yanrez]

If i manually downgrade to 0.5.2 everything works fine. When downgrading, I had to also kill server.exe (v0.6.0) manually - it was running in background and prevented older UI from working

[illume]

Hi @yanrez

Thanks for the info!

In 0.6.0 we tried to handle Kubernetes removing and deprecating some URLs. Seems like it's probably related to that. I wonder what distributions and versions of k8s you are using (kubectl version)?

It is supposed to kill the server process on app exit, so that seems like something we need to fix too.

[yanrez]

❯ kubectl version Client Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.5", GitCommit:"aea7bbadd2fc0cd689de94a54e5b7b758869d691", GitTreeState:"clean", BuildDate:"2021-09-15T21:10:45Z", GoVersion:"go1.16.8", Compiler:"gc", Platform:"windows/amd64"} Server Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.2", GitCommit:"802eff1fe87ad2dd737ebbe891f30500b88beb00", GitTreeState:"clean", BuildDate:"2021-11-15T08:35:41Z", GoVersion:"go1.16.5", Compiler:"gc", Platform:"linux/amd64"}

[illume]

Thanks.

I wonder if there are any errors in the developer console?

ctrl+i (or view->Toggle Developer Tools) -> console tab

[yanrez]

That's odd. I've upgraded to 0.6.0 again, and everything works fine. I've killed server.exe manually to make sure I get latest version when launching headlamp 0.6.0, and it still works. I lost repro :(

[illume]

Oh! At least it works for you now :)

I wonder if the 0.5.2 server process was hanging around and the 0.6.0 app tried to talk to it.

[yanrez]

Maybe. One way or another, killing server.exe on app exit would be good thing.

[joaquimrocha]

Since this seems to indeed have been related to a running (unkilled) version of the server and we've addressed it in #395 , I think we should close this one down.

[gaalw]

Please reopen the issue.

Steps to reproduce.

Running `brew update --auto-update`...
==> Auto-updated Homebrew!
Updated 5 taps (hashicorp/tap, weaveworks/tap, minio/stable, homebrew/core and homebrew/cask).
==> New Formulae
cotp                 git-split-diffs      hashicorp/tap/hcp    tartufo              whisperkit-cli       yo
==> New Casks
nperf                           posture-pal                     proton-mail                     steinberg-activation-manager

You have 79 outdated formulae and 3 outdated casks installed.

==> Downloading https://github.com/kinvolk/headlamp/releases/download/v0.23.0/Headlamp-0.23.0-mac-x64.dmg
==> Downloading from https://objects.githubusercontent.com/github-production-release-asset-2e65be/220546758/eb639fc6-b9ab-4da6-bb
########################################################################################################################## 100.0%
==> Installing Cask headlamp
==> Moving App 'Headlamp.app' to '/Applications/Headlamp.app'

Run Headlamp.

The clusters from ~/.kube/config are loaded properly.

If I click on "Load kubeconfig" button:

So the function of loading external kubeconfig is not working at all. No idea why.

[illume]

Thanks for the detailed report. I opened a new issue with your post in case someone misses it here. See #1831