Lab 13.3 version of helm chart of kubernetes-dashboard

m.taniguchi

Hi I've been really stacked at Lab 13.3. and Finally solved and understand what was wrong.

[Problem]
Kubernetes dashboard can't displayed properly.

[Brief steps]
1, fetch kubernetes-dashboard charts from https://kubernetes.github.io/dashboard/ based on artifacthub
2. cd to untarred directory modifyed values.yaml and install dashboard
$helm install dashboard .
3. Error message came up and stacked.

Error: INSTALLATION FAILED: failed post-install: warning: Hook post-install kubernetes-dashboard/templates/networking/post-install-ingress-issuer.yaml failed: Internal error occurred: failed calling webhook "webhook.cert-manager.io": failed to call webhook: Post "https://kdash-cert-manager-webhook.default.svc:443/mutate?timeout=10s": dial tcp 10.102.188.86:443: connect: connection refused

4. services were up so I accessed to the dashboard but can't displayed anything

I tried various thing including deletion of service,deploymen,pods in previous tasks, but nothing changes. and I found descriptions/explanations on this course are really different from what I'm seeing.

The problem is "Helm chart version" of dashboard. what I used is version 7.0.x, maybe current latest. I fetch ver 5.11.0 (another trainee used on this forum)

$helm pull kubernetes-dashboard/kubernetes-dashboard --version 5.11.0 --untar

and re-installed , then it works.

In order to keep students from confusing , I really appreciate to explain where to be referred as previous labs, and version information in advance. (Really took time to solve. of course this is good experience though.)

Thanks

chrispokorni

Hi @m.taniguchi,

In case it did not come up in your search, there is a workaround I provided on a previous discussion thread

https://forum.linuxfoundation.org/discussion/comment/38139/#Comment_38139

Regards,
-Chris

m.taniguchi

Thanks, in case using dashboard chart v5.11, dashboard doesn't show the charts from metric-server(v0.6.7) 's values. this seems to be similar to the previous post (of course metric server works fine with kubectl top xxx ).

https://forum.linuxfoundation.org/discussion/859748/lab-13-3-charts-on-dashboard

I will also try latest dashboard chart and edit after deployment of it base on the link you gave me.

In case it did not come up in your search, there is a workaround I provided on a previous discussion thread
https://forum.linuxfoundation.org/discussion/comment/38139/#Comment_38139

chrispokorni

Hi @m.taniguchi,

I attempted to reproduce the dashboard issues reported above, but was not able to do so with dashboard Helm chart versions 7.0.x. None of the installation and configuration recommendations from the chart's home page seemed to help.

However, installing an earlier version of the chart allowed me to see the expected results in the dashboard web UI, while following closely the instructions presented in the lab guide.

I used the following commands to add the repo, fetch and untar the dashboard chart version 6.0.8:

helm repo add kubernetes-dashboard https://kubernetes.github.io/dashboard/

helm pull kubernetes-dashboard/kubernetes-dashboard --version=6.0.8 --untar

Then I made the following changes in values.yaml:

cd $HOME/kubernetes-dashboard

vim values.yaml

service:
  type: NodePort       #<<-- line 145

metricsScraper:
  enabled: true         #<<-- line 260

Installed a release and assigned the desired permissions:

helm install dashboard .

kubectl create clusterrolebinding dashaccess \
  --clusterrole=cluster-admin \
  --serviceaccount=default:dashboard-kubernetes-dashboard

Regards,
-Chris

m.taniguchi

Hi, @chrispokorni , Thanks for your support!

I used the following commands to add the repo, fetch and untar the dashboard chart version 6.0.8:

Following your instruction, metrics data have been displayed as expected!

Anyway, I think it was a version sensitive issue. in my case worked with;

• metrics-server: 0.6.7
• dashboard helm chart : 6.0.8

Thanks again your support.

rebekah

I used the kubectl create command in the docs: https://kubernetes.io/docs/tasks/access-application-cluster/web-ui-dashboard/. Then I updated the service that was created to be of type NodePort. Then I noted the port by kubectl -n kubernetes-dashboard get services. I altered the service account to reference the correct service account name: kubernetes-dashboard and namespace: kubernetes-dashboard. Then I created the ClusterRoleBindinding as in the course documentation but switched out the namespace: kubectl create clusterrolebinding dashaccess --clusterrole=cluster-admin --serviceaccount=kubernetes-dashboard:kubernetes-dashboard. And then you can create the token for the service account: kubectl create token kubernetes-dashboard. Grab the external ip - I get it from the google console as I am running my cluster in GCE. Then go to <your_control_planes_external_ip>:<your_node_port> in your browser. You may have to proceed to unsafe. Then put in your token. And violà it should work.