Error connecting cluster causing panic

[bigkevmcd]

Describe the bug

❯ gitops-ee connect cluster test
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x8 pc=0x209da22]

goroutine 1 [running]:
github.com/weaveworks/weave-gitops-enterprise/pkg/cluster/connector.getSecretNameFromCluster({0x3658790, 0xc001192f30}, {0x36383a0, 0xc0013f51d0}, 0xc0000c3950?, {{0x2f3b6d2, 0x7}, {0x7ffd1083b106, 0x4}})
    /home/runner/work/weave-gitops-enterprise/weave-gitops-enterprise/pkg/cluster/connector/secret.go:38 +0x3c2
github.com/weaveworks/weave-gitops-enterprise/pkg/cluster/connector.getSecretNameForConfig({0x3658790, 0xc001192f30}, 0xc0000c39a0?, 0xc00103fc28)
    /home/runner/work/weave-gitops-enterprise/weave-gitops-enterprise/pkg/cluster/connector/connector.go:66 +0x71
github.com/weaveworks/weave-gitops-enterprise/pkg/cluster/connector.ConnectCluster({0x3658790, 0xc001192f30}, 0xc00149fc28)
    /home/runner/work/weave-gitops-enterprise/weave-gitops-enterprise/pkg/cluster/connector/connector.go:87 +0x136
github.com/weaveworks/weave-gitops-enterprise/cmd/gitops/app/connect/clusters.connectClusterCmdRunE.func1(0xc000f81800, {0xc000d35610?, 0x1?, 0x1?})
    /home/runner/work/weave-gitops-enterprise/weave-gitops-enterprise/cmd/gitops/app/connect/clusters/cmd.go:61 +0x1bb
github.com/spf13/cobra.(*Command).execute(0xc000f81800, {0xc000d355e0, 0x1, 0x1})
    /home/runner/go/pkg/mod/github.com/spf13/cobra@v1.7.0/command.go:940 +0x862
github.com/spf13/cobra.(*Command).ExecuteC(0xc000004900)
    /home/runner/go/pkg/mod/github.com/spf13/cobra@v1.7.0/command.go:1068 +0x3bd
github.com/spf13/cobra.(*Command).Execute(0xc000062060?)
    /home/runner/go/pkg/mod/github.com/spf13/cobra@v1.7.0/command.go:992 +0x19
main.main()
    /home/runner/work/weave-gitops-enterprise/weave-gitops-enterprise/cmd/gitops/main.go:14 +0x6a

To Reproduce

Actual behaviour

Expected behaviour

Additional context

• Weave GitOps version:
• Kubernetes version:

[darrylweaver]

This was generated with a GitopsCluster manually added to the WGE management cluster:

apiVersion: gitops.weave.works/v1alpha1
kind: GitopsCluster
metadata:  
  name: test
  namespace: default

The spec is missing as well as any labels, which could be causing this issue.

[darrylweaver]

The in-cluster object was:

apiVersion: gitops.weave.works/v1alpha1
kind: GitopsCluster
metadata:
  creationTimestamp: '2023-11-23T13:56:27Z'
  generation: 1
  labels:
    kustomize.toolkit.fluxcd.io/name: flux-system
    kustomize.toolkit.fluxcd.io/namespace: flux-system
  managedFields:
    - apiVersion: gitops.weave.works/v1alpha1
      fieldsType: FieldsV1
      fieldsV1:
        f:metadata:
          f:labels:
            f:kustomize.toolkit.fluxcd.io/name: {}
            f:kustomize.toolkit.fluxcd.io/namespace: {}
      manager: kustomize-controller
      operation: Apply
      time: '2023-11-23T13:56:27Z'
  name: test
  namespace: default
  resourceVersion: '148517'
  uid: a52a145d-d278-4ecc-980c-0b24f32139a2
  selfLink: /apis/gitops.weave.works/v1alpha1/namespaces/default/gitopsclusters/test

which of course has labels, but still no spec.

[bigkevmcd]

Two things to fix then...

1. We need an admission controller that checks that the GitopsCluster has at least a capi reference, or secret reference.
2. Connecting a cluster should not error for invalid Clusters.

[darrylweaver]

I added the spec to the object and it didn't fix the issue, I still get the same error.
I created an empty secret for the kubeconfig reference, in case it needed to exist, but I still get the same error. I added data for a kubeconfig to the secret, same error. I added an empty secret for certs too, but the same error.

I haven't found a way to avoid it yet, so no clue what is wrong here. Something missing probably. I tried with debug, but I don't get much more information, just that it detects the GitopsCluster object exists, then I hit the error.

[bigkevmcd]

If it's the same error, it's failing here: https://github.com/weaveworks/weave-gitops-enterprise/blob/498fa43fe1a6e5515ba8a2e70c2ffbbe8a6a3301/pkg/cluster/connector/secret.go#L38

Which implies that the secretRef isn't configured?

[bigkevmcd]

Can you paste the YAML for your GitopsCluster?

[darrylweaver]

For reference, here is the GitopsCluster object with the reference added.

apiVersion: gitops.weave.works/v1alpha1
kind: GitopsCluster
metadata:
  creationTimestamp: '2023-11-23T13:56:27Z'
  generation: 2
  labels:
    kustomize.toolkit.fluxcd.io/name: flux-system
    kustomize.toolkit.fluxcd.io/namespace: flux-system
  managedFields:
    - apiVersion: gitops.weave.works/v1alpha1
      fieldsType: FieldsV1
      fieldsV1:
        f:metadata:
          f:labels:
            f:kustomize.toolkit.fluxcd.io/name: {}
            f:kustomize.toolkit.fluxcd.io/namespace: {}
        f:spec:
          f:capiClusterRef:
            f:name: {}
      manager: kustomize-controller
      operation: Apply
      time: '2023-11-23T17:07:15Z'
  name: test
  namespace: default
  resourceVersion: '193300'
  uid: a52a145d-d278-4ecc-980c-0b24f32139a2
  selfLink: /apis/gitops.weave.works/v1alpha1/namespaces/default/gitopsclusters/test
spec:
  capiClusterRef:
    name: test

[bigkevmcd]

Ahh...you don't need to connect a capi cluster...it just "works"?

But we can definitely detect this and improve the message.

[darrylweaver]

Of course it does, I copied it from another GitopsCluster object. I have now added the correct secretRef and removed the extra secrets I manually created. That has solved the problem, so the missing secretRef was the issue.

Output now:

gitops-ee connect cluster test
{"level":"info","ts":1700760829.72523,"logger":"gitops","caller":"connector/connector.go:117","msg":"Successfully connected cluster","cluster":{"name":"test","namespace":"default"}}

[bigkevmcd]

We'll definitely improve the errors around this.

[bigkevmcd]

this cluster does not reference a secret - can only connect clusters that reference a secret