Description

Helm charts do not set the ipFamilies attribute for the longhorn-frontend service. The longhorn-ui does not support IPv6. For dual-stack clusters that have IPv6 as their primary family, the ipFamilies for the longhorn-frontend ends up being automatically set to ["IPv6"], thus making the Longhorn UI inaccessible.

To Reproduce

• Create a K8S cluster with an IPv6 range as the first first value for kube-apiserver's --service-cluster-ip-range option.

• Install Longhorn with the Helm charts.

• Observe that the service automatically gets ipFamilies set to ["IPv6"] (kubectl get -o yaml -n longhorn-system service/longhorn-frontend).

• Try to access the UI via kubectl proxy at http://localhost:8001/api/v1/namespaces/longhorn-system/services/longhorn-frontend:http/proxy/.

• Get an error that looks like this:

  {
    "kind": "Status",
    "apiVersion": "v1",
    "metadata": {},
    "status": "Failure",
    "message": "error trying to reach service: proxy error from 127.0.0.1:6443 while dialing [R:E:D:A:C:T:E:D]:8000, code 502: 502 Bad Gateway",
    "reason": "ServiceUnavailable",
    "code": 503
  }

Expected behavior

The longhorn-frontend service have its ipFamilies attribute "manually" set to ["IPv4"] (until #2259 is solved) by the Helm charts and is thus accessible.

Support bundle for troubleshooting

N/A

Environment

• Longhorn version: v1.4.4
• Installation method (e.g. Rancher Catalog App/Helm/Kubectl): Helm
• Kubernetes distro and version: K3s v1.27.6+k3s1

Additional context