Initial setup stuck on 'Updating kubeconfig' (WSL)

[Heap0017]

Actual Behavior

The initial setup gets stuck at 'Updating kubeonfig'.

Steps to Reproduce

Open Rancher Desktop, choose K8s Version + engine as below.

Result

The initial setup gets stuck at 'Updating kubeonfig'.

Expected Behavior

The initial setup is succesfull.

Additional Information

No diagnostic errors.
I tried reinstalling Rancher Desktop.

• rancher-desktop-guestagent.log repeated message: kubernetes: failed to read kubeconfig [config-path=/etc/rancher/k3s/k3s.yaml][error=could not load Kubernetes client config from /etc/rancher/k3s/k3s.yaml: stat /etc/rancher/k3s/k3s.yaml: no such file or directory]
• wsl-exec.log repeated message: Running: wsl.exe --distribution rancher-desktop --exec /bin/sh -c if test -r /etc/rancher/k3s/k3s.yaml; then echo yes; else echo no; fi
• k3s.log repeated message: level=fatal msg="no default routes found in \"/proc/net/route\" or \"/proc/net/ipv6_route\"". Potentially related issue Error "no default routes found" when using Air-gap installation k3s-io/k3s#1144. However, there is a default route on my machine.

See also https://rancher-users.slack.com/archives/C0200L1N1MM/p1718399080544969

Rancher Desktop Version

1.15.1

Rancher Desktop K8s Version

1.30.4

Which container engine are you using?

containerd (nerdctl)

What operating system are you using?

Windows

Operating System / Build Version

Windows 10 Enterprise 19045

What CPU architecture are you using?

x64

Linux only: what package format did you use to install Rancher Desktop?

None

Windows User Only

Using VPN

[Heap0017]

A workaround for me is to disable my VPN. Likely, the underlying issue is outside the scope of Rancher Desktop. However, it would be great to improve the error handling given that k3s reports a fatal error that seems to stay unnoticed by Rancher.

[mook-as]

Could you enable debug logging and reproduce the problem, and then attach all of the logs to the bug? That would help track down issues.

For technical detail: we set up a network namespace in WSL, so the /proc/net/route that k3s sees should be ours (instead of your defaults).

[Heap0017]

Since working around the issue once, I am unable to reproduce it. Rancher Desktop seems to preserve some configuration across factory resets, reinstallations, and system reboots. How can I reset to a fresh installation?

[mook-as]

Odd; the only thing we may attempt to persist across factory resets are:

• The cached Kubernetes images (which we don't modify locally); if you factory reset using the UI, you need to tick the checkbox to keep it (and for the CLI, adding --remove-kubernetes-cache deletes it).
• Snapshots; these are only used when you explicitly interact with them.
• On Windows only: we don't uninstall WSL for you, because you may have non-Rancher Desktop distributions installed.

Neither should cause the issues you're seeing; so if we're keeping anything else, that would be a bug…

[jandubois]

I'm closing this bug since the issue seems not to be reproducible anymore