Adopted local kind cluster doesn't work

[Josca]

Describe the bug
I tried to adopt local "kind" cluster as adopted cluster for k0rdent testing. It should be possible according to Adopted cluster guide.
Local setup:

• kind k0rdent management cluster created using demo setup.
• a second kind cluster to be used as a managed cluster.

To Reproduce
Steps to reproduce the behavior:

1. Create secret and credential resource:

apiVersion: v1
data:
  value: YXNkc2RmYXNkZgo= # invalid kubeconfig - but it's still accepted
kind: Secret
metadata:
  name: adopted-cluster-kubeconf
  namespace: k0rdent
  labels:
    k0rdent.mirantis.com/component: "kcm"
type: Opaque
---
apiVersion: k0rdent.mirantis.com/v1alpha1
kind: Credential
metadata:
  name: adopted-cluster-cred
  namespace: k0rdent
spec:
  description: Adopted Credentials
  identityRef:
    apiVersion: v1
    kind: Secret
    name: adopted-cluster-kubeconf
    namespace: k0rdent

2. Create adopted cluster ClusterDeployment based on adopted-cluster-0-0-2 template:

apiVersion: k0rdent.mirantis.com/v1alpha1
kind: ClusterDeployment
metadata:
  name: kind-adopted-cluster
  namespace: k0rdent
spec:
  template: adopted-cluster-0-0-2
  credential: adopted-cluster-cred
  # dryRun: <"true" or "false": defaults to "false">
  config:
    clusterLabels:
      k0rdent: demo
  serviceSpec:
    services:
      - template: ingress-nginx-4-11-0
        name: ingress-nginx
        namespace: ingress-nginx

4. Check clusterdeployment cluster status, it shows cluster ready:

k get clusterdeployments.k0rdent.mirantis.com 
# NAME                   READY   STATUS
# kind-adopted-cluster   True    ClusterDeployment is ready
k get clustersummaries.config.projectsveltos.io -A # ... but no cluster summaries
# No resources found

• Services are not installed
• sveltos agent is not being installed to manged cluster
• It behaves the same way no matter whether adopted-cluster-kubeconf.data.value is valid kubeconfig or not. I tried to use just encoded blablabla and it worked the same way.

Expected behavior
Adopted cluster really works.

Additional context
Add any other context about the problem here.

[kylewuolle]

I think the issue is the kubeconfig provided for the kind cluster created is probably pointing at localhost (127.0.0.1) which is the case if it's obtained using kind get kubeconfig -n my-test-cluster. If you instead use a kubeconfig obtained using the "internal" flag you will have a kubeconfig that will work within the docker network of the kind cluster. So for example you could create the credential using the result of kind get kubeconfig --internal -n my-test-cluster| base64 -w 0.

Alternatively, you can create a kind config file similar to this:?

kind: Cluster apiVersion: kind.x-k8s.io/v1alpha4 networking: apiServerAddress: "0.0.0.0" apiServerPort: <choose a port>

And then create a kind cluster like this kind create cluster -n my-test-cluster --config <kind config>. This will bind the api port to all interfaces and then you could use your host ip in the kubeconfig file.

[Josca]

Hey @kylewuolle , thank you. It really worked when I created a kubeconfig using --internal flag. So the major issue is resolved. But anyway, it would be great to make it more visible when the adopted cluster is not "adopted" properly.

• Credentials and ClusterDeployment objects should describe the issue in their status to simplify troubleshooting.

[kylewuolle]

Hey @kylewuolle , thank you. It really worked when I created a kubeconfig using --internal flag. So the major issue is resolved. But anyway, it would be great to make it more visible when the adopted cluster is not "adopted" properly.

• Credentials and ClusterDeployment objects should describe the issue in their status to simplify troubleshooting.

I've made a pull request that adds the sveltos cluster status to the clusterdeployment that should help with these situations.