Can not access Juju Dashboard k8s

[aataibi]

Describe the bug
Following this guide, juju dashboard command returns an error and fails to proxy the dashboard.
I think the issue is because it's trying to access the apiserver using the Cluster IP.
I tried to change the svc type from ClusterIP to LoadBalancer but that didn't help as the external IP is not picked up by juju.

juju dashboard --debug
11:58:27 INFO  juju.cmd supercommand.go:56 running juju [3.6.0 6dec77a01480916689430d38ef3e032cb1e06b78 gc go1.23.3]
11:58:27 DEBUG juju.cmd supercommand.go:57   args: []string{"/snap/juju/29130/bin/juju", "dashboard", "--debug"}
11:58:27 INFO  juju.juju api.go:86 connecting to API addresses: [10.152.183.225:17070]
11:58:27 INFO  juju.kubernetes.klog klog.go:113 Use tokens from the TokenRequest API or manually created secret-based tokens instead of auto-generated secret-based tokens.%!(EXTRA []interface {}=[])
11:58:27 DEBUG juju.api apiclient.go:509 starting proxier for connection
11:58:27 DEBUG juju.api apiclient.go:513 tunnel proxy in use at localhost on port 38855
11:58:27 DEBUG juju.api apiclient.go:1036 successfully dialed "wss://localhost:38855/api"
11:58:27 INFO  juju.api apiclient.go:938 cannot resolve "localhost": lookup localhost: operation was canceled
11:58:27 INFO  juju.api apiclient.go:571 connection established to "wss://localhost:38855/api"
11:58:57 DEBUG juju.api monitor.go:35 RPC connection died
ERROR running connection runner: starting tunnel proxy: connecting k8s proxy: Get "https://10.152.183.1:443/api/v1/namespaces/controller-microk8s-controller/services/dashboard": dial tcp 10.152.183.1:443: i/o timeout
11:58:57 DEBUG cmd supercommand.go:549 error stack: 
Get "https://10.152.183.1:443/api/v1/namespaces/controller-microk8s-controller/services/dashboard": dial tcp 10.152.183.1:443: i/o timeout
github.com/juju/juju/caas/kubernetes/provider/proxy.(*Proxier).Start.func1:117: connecting k8s proxy
github.com/juju/juju/cmd/juju/dashboard.(*dashboardCommand).Run.tunnelProxyRunner.func4:261: starting tunnel proxy
github.com/juju/juju/cmd/juju/dashboard.(*dashboardCommand).Run.func3:177: running connection runner
github.com/juju/juju/cmd/juju/dashboard.(*dashboardCommand).Run:194: 

kubectl get svc 
NAME         TYPE           CLUSTER-IP     EXTERNAL-IP      PORT(S)         AGE
kubernetes   LoadBalancer   10.152.183.1   10.124.203.244   443:31265/TCP   6d18h

To Reproduce

juju add-model microk8s
juju deploy -m microk8s microk8s --channel 1.28/stable
juju exec --unit microk8s/leader -- microk8s config > ~/.kube/config
juju add-k8s microk8s-cloud
juju bootstrap microk8s-cloud microk8s-controller
juju switch microk8s-controller:admin/controller
juju deploy juju-dashboard-k8s dashboard
juju integrate dashboard controller
juju config dashboard juju-external-hostname=k8s-dashboard
juju expose dashboard
juju dashboard

Expected behavior
The juju dashboard command should open a connection to the dashboard and provides the dashboard address and credentials.

Workaround

• run the command in the microk8s unit.

 juju ssh microk8s/leader
 juju dashboard
 Dashboard for controller "microk8s-controller" is enabled at:
  http://localhost:39869
Your login credential is:
  username: admin
  password: e18c93d04bdf0e39f0eb2f5f32f2e4e8

• The service is bound to 127.0.0.1, so will need an ssh tunnel to access it.
  LISTEN 0 4096 127.0.0.1:39869 0.0.0.0:* users:(("juju",pid=457853,fd=13))
  

[tlm]

FYI this isn't a "bug" on the dashboard project. This is should be a bug on the Juju projector Launchpad. Have responded to @aataibi in Ubuntu Matrix about this as I think this is just a networking issue at the moment due to the complex nature of the testing steps.

[huwshimi]

Thanks for investigating @tlm. Let me know if there are any issues on this side.