Error starting kubernetes on first start on Linux (after manual cleanup)

[thehejik]

Actual Behavior

First initial start of RD on Linux openSUSE 15.3 is failing when using default values (containerd, default k8s, ...). Subsequent start is successfull.

When validating rpm package rancher-desktop-9.main.1654131014.19ab90f-346.1.x86_64.rpm after uninstalling previous RD version and manual cleanup of its leftovers by:

rm -r ~/.local/share/rancher-desktop ~/.config/rancher-desktop ~/.cache/rancher-desktop ~/.kube* ~/.rd

and starting as normal user the following Error is thrown in Waiting for service phase:

 at ChildProcess.emit (node:events:390:28)
    at Process.ChildProcess._handle.onexit (node:internal/child_process:290:12) {
  command: [
    '/opt/rancher-desktop/resources/resources/linux/bin/kubectl',
    '--context',
    'rancher-desktop',
    'cluster-info'
  ]
}

Steps to Reproduce

On system when it was installed before

$ sudo zypper rm rancher-desktop
$ rm -r ~/.local/share/rancher-desktop ~/.config/rancher-desktop ~/.cache/rancher-desktop ~/.kube* ~/.rd
$ sudo zypper in rancher-desktop

• start RD from gnome and observer if the error occures

On fresh system:

• install RD from rpm (main branch)
• start it from gnome
• the error is thrown

Result

The error is described above.

Expected Behavior

It should start successfully on first attempt

Additional Information

No response

Rancher Desktop Version

rancher-desktop-9.main.1654131014.19ab90f-346.1.x86_64.rpm
rancher-desktop-9.main.1654131014.19ab90f-346.2.x86_64.rpm

Rancher Desktop K8s Version

v1.23.6 (stable,latest)

Which container runtime are you using?

containerd (nerdctl)

What operating system are you using?

Other (specify below)

Operating System / Build Version

opensuse 15.3

What CPU architecture are you using?

x64

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

rpm

Windows User Only

No response

[thehejik]

I was able to reproduce the same issue on fresh opensuse tumbleweed installation with gnome.

[thehejik]

The problem is with kuberlr which returns non-zero return code after downloading and moving kubectl binary from /tmp to /home. The problem is fixed and PR flavio/kuberlr#23 merged in kuberlr's master branch but not built yet.

So for fixing this issue we just need to:

• rebuild kuberlr binary by creating a new tag v0.4.2 in its repo https://github.com/flavio/kuberlr. EDIT: v0.4.2 has been released
• do a kuberlr version bump here

  rancher-desktop/scripts/download/tools.mjs

  Line 49 in 9eacdeb

  const kuberlrVersion = '0.4.1';