[BUG] WSL2: host.k3d.internal warning & failed kubectl connection at 0.0.0.0
#858
Comments
|
@jadler-goodrx looks like other people are seeing this as well. |
|
I'm also seeing this, @keyiis does using |
|
@kslacroix After I changed kubeconfig host to |
What do you mean by kubeconfig host? |
|
@kslacroix |
|
@keyiis Ahh yeah this makes sense, thank you! For reference, this is where it fails for me... // GetHostIP returns the routable IP address to be able to access services running on the host system from inside the cluster.
// This depends on the Operating System and the chosen Runtime.
func GetHostIP(ctx context.Context, runtime runtimes.Runtime, cluster *k3d.Cluster) (net.IP, error) {
rtimeInfo, err := runtime.Info()
if err != nil {
return nil, err
}
l.Log().Tracef("GOOS: %s / Runtime OS: %s (%s)", goruntime.GOOS, rtimeInfo.OSType, rtimeInfo.OS)
isDockerDesktop := func(os string) bool {
return strings.ToLower(os) == "docker desktop"
}
// Docker Runtime
if runtime == runtimes.Docker {
// Docker (for Desktop) on MacOS or Windows
if isDockerDesktop(rtimeInfo.OS) {
toolsNode, err := EnsureToolsNode(ctx, runtime, cluster)
if err != nil {
return nil, fmt.Errorf("failed to ensure that k3d-tools node is running to get host IP :%w", err)
}
ip, err := resolveHostnameFromInside(ctx, runtime, toolsNode, "host.docker.internal", ResolveHostCmdGetEnt)
if err == nil {
return ip, nil
}
l.Log().Warnf("failed to resolve 'host.docker.internal' from inside the k3d-tools node: %v", err)
}
l.Log().Infof("HostIP: using network gateway...")
ip, err := runtime.GetHostIP(ctx, cluster.Network.Name)
if err != nil {
return nil, fmt.Errorf("runtime failed to get host IP: %w", err)
}
return ip, nil
}
// Catch all other runtime selections
return nil, fmt.Errorf("GetHostIP only implemented for the docker runtime")
}The line |
|
The way I fixed the issue on my end (without modifying kubeconfig) was to modify the coredns config-map. Which returns the hostIP as seen from the docker VM. The IP was |
|
humm.... I'm having the same problem. It worked fine before... Once I edited kubeconfig directly, I also confirmed that it works. |
|
@heesuk-ahn what do you have to do to fix this? When I try to use an image I have loaded into the cluster I get the following error |
|
same issue: WARN[0001] failed to resolve 'host.docker.internal' from inside the k3d-tools node: Failed to read address for 'host.docker.internal' from command output |
|
Up to version 5.0.1, the above code normally receives the value However, strangely in the later versions |
|
In my case, since k3d is executed inside the container, after changing the server host information in kubeconfig to |
|
Hi @keyiis , thanks for opening this issue and thanks to the others for providing more input on this! However, all the fuss about I think this issue actually covers two different problems, which I will investigate today 👍 |
|
@keyiis (and others!) are you running your commands in the WSL2 terminal or in PowerShell?
|
|
Another few things to note here... Docker for Desktop has at least three possible variations:
Testing with WSL2 backend:
Other cases to consider:
Potential solution: making use of the |
iwilltry42
changed the title
host.k3d.internal warning & failed kubectl connection at 0.0.0.0
|
@iwilltry42 hi, My development environment is used by installing k3d inside the alpine container and connecting to the docker daemon of the host (macOs). In this environment, if I change And my docker version is as below. |
|
@iwilltry42 You're right, I misunderstood the issue at first. My problem is different from @keyiis especially since I'm on Macos. I was seeing these logs also which made me think we had the same issue. Should I open a new issue or will your fix address my issue also? |
Your docker client (CLI) and kubectl are also running inside that alpine container?
It will be addressed by the linked PR 👍 |
|
thanks your advise :) I tried How can I possibly solve this error? 🤔 |
|
@heesuk-ahn ah yes, didn't think of that... Instead of relying on the kubeconfig output by k3d upon k3d kubeconfig get mycluster | sed 's/0\.0\.0\.0/host.k3d.internal/' > /tmp/k3d-mycluster.yaml && \
KUBECONFIG=$KUBECONFIG:/tmp/k3d-mycluster.yaml kubectl config view --merge --flatten > $HOME/.kube/configNote: I did not test this, so better backup your kubeconfig first! |
What did you do
What did you expect to happen
kubectl cluster-infocan work.Screenshots or terminal output
Which OS & Architecture
My OS is win10,docker run in wsl2,I use dind mode to install k3d,I runing k3d container by
docker run ... -v /var/run/docker.sock:/var/run/docker.sock ...,Therefore, K3D and nodes(k3d cluster node) are running at the same level.Which version of
k3dWhich version of docker
The text was updated successfully, but these errors were encountered: