This message was deleted.

I can't see the servers being configured anywhere in the Rancher Desktop code. When are they being accessed? When you start the app, when you pull an image, or when you run a container?

The DNS requests to 8.8.8.8 are made when the rancher-desktop app is started and the queries stop when we stop the application. Is there any way to check in logs or rancher-desktop WSL distro to check or debug ? There are running containers in the kubernetes environment. The /etc/resolv.conf in rancher-desktop WSL points to the local ip address of the WSL instance

Are you sure the requests are not coming from inside your containers? Do you have a chance to do a factory reset, and start the app without any workloads deployed (and without Kubernetes)? To see if this already starts the queries? Or if they only start once you run k8s. Or once you deploy your containers.

Yeah I can do a factory reset and check if without workloads queries are also executed. The network tunnel option is not enabled yet. Will try this in the office on Friday! Will let you know the outcome

Ok, thanks! The networking tunnel will become the default in the next release, and the legacy networking stack will be removed sometime after that. The tunnel solves many DNS problems, especially when using a VPN, as all traffic will originate from the host and not inside WSL. So it is possible that this might already solve your issue, if it indeed comes from Rancher Desktop and not the container itself.

Thanks for the chat and tips, will let you know asap

I'm fairly certain that switching to the networking tunnel mode will solve the issue for you; the only place those addresses appear is as fallback servers in the host-resolver code that is only used by the legacy networking stack.

Hi Jan, switching to network tunnel stop the requests to 8.8.8.8 but now a few of my host file entries could not being resolved from a pod. What happens in tunnel mode with the windows host file entries? Especially the hostnames in the hosts file which ends with the domain *.docker.internal. These entries could not being resolved when the tunnel option is active. When I changed them to antoher domain for example *.example.lab everything works fine. Seems that the docker.internal domain receives some special treatment in terms of resolving. So switching to tunnel mode and changing domain names will do the job!

Can you please file a Github issue about this?