Inaccessible Local Development Cluster in WSL2

[SchahinRohani]

When starting the local development cluster in WSL2, it is not reachable. This appears to be due to missing networking modules for Cilium in the WSL2 Linux kernel.

A potential solution could involve OS detection and automatically switching to the default CNI when these modules are unavailable.

Feedback and discussion on an implementation approach would be greatly appreciated.

FYI: Modifying the WSL2 kernel for Cilium is possible, as outlined here: WSL Cilium Kernel Modifications, but this approach is not particularly user-friendly as it requires kernel modifications.

[MarcusSorealheis]

We should avoid kernel modifications whenever possible.

[SchahinRohani]

@MarcusSorealheis i completely agree!

That’s exactly why I wanted to open up the discussion. A lot of Windows devs rely on WSL, and having a kind dev cluster that works seamlessly in WSL would really speed up testing and development.

[aaronmondal]

@JannisFengler IIRC you mentioned solving a similar issue before. Do you remember how you solved this?

[JannisFengler]

I think i solved it by building the 6.1 linux kernel(
https://learn.microsoft.com/en-us/community/content/wsl-user-msft-kernel-v6 )
and enabling mirrored mode in the wsl config, which enables IPv6.

[SchahinRohani]

@JannisFengler
I built the 6.1 kernel and enabled mirrored mode, but it doesn't seem to work.
I also built the WSL2 Linux 6.6 kernel, and it still does not work. Therefore, the issue might be not related to missing kernel modules.

@aaronmondal
Various Flux controllers, the Tekton feature flags triggers, and the tekton-triggers-webhook are also failing when updating the local development cluster.

The issue could lie elsewhere entirely...