Testing Connections and Viewing the Kubernetes Dashboard

ENTERPRISE

Learn to test your connection to private Kubernetes clusters and view the Kubernetes dashboard.

In this section, we will test our connections to our Kubernetes clusters one at a time and then lastly connect to the Kubernetes dashboard and view our Kubernetes clusters’ information. To test our connections we must ensure that the right ports are open, set our cluster’s context with the DC/OS Kubernetes service, get Kubernetes nodes’ information, and test with a simple NGINX deployment.

Check that Port :6443 and :6444 are open.

Before attempting to connect kubectl to the MKE clusters, you will want to check to ensure that ports :6443 and :6444 are accessible by your local machine to the DC/OS Cluster. Closed ports :6443 and :6444 will cause kubectl commands to just hang. Typically, these settings are handled by your systems administrator. Or, if using a cloud provider such as AWS, these would be rules configured in your EC2–>Security Groups tab.

Test the connection to kubernetes-cluster1 at port :6443

  1. First, configure your context to kubernetes-cluster1 at port :6443.

    If using Edge-LB Auto Pools, in your CLI, enter the following snippet:

    dcos kubernetes cluster kubeconfig \
    --insecure-skip-tls-verify \
    --context-name=kubernetes-cluster1 \
    --cluster-name=kubernetes-cluster1 \
    --apiserver-url=https://${EDGELB_PUBLIC_AGENT_IP}:6443/kubernetes-cluster1

    If not using Edge-LB Auto Pools, enter the following snippet:

    dcos kubernetes cluster kubeconfig \
    --insecure-skip-tls-verify \
    --context-name=kubernetes-cluster1 \
    --cluster-name=kubernetes-cluster1 \
    --apiserver-url=https://${EDGELB_PUBLIC_AGENT_IP}:6443

    which will configure DC/OS Kubernetes to connect to kubernetes-cluster1 from the Kubernetes CLI, kubectl. The EDGELB_PUBLIC_AGENT_IP should still be cached in your terminal environment from the previous section, if not, make sure to insert the proper value.

  2. Next, get your Kubernetes nodes’ info for kubernetes-cluster1.

    In your CLI, enter:

    kubectl get nodes

    to receive details regarding your the nodes of kubernetes-cluster1.

    If you do not get an almost immediate response, check your firewall settings and ensure that you can communicate with your public agent node.

  1. Then, create a NGINX deployment on kubernetes-cluster1.

    Enter the following in your CLI:

    kubectl apply -f https://k8s.io/examples/application/deployment.yaml

    to launch the deployment.

  2. View the NGINX deployment just launched.

    In your CLI, run:

    kubectl get deployments

    The resulting output should look similar to the following:

    $ kubectl get deployments
NAME               DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
nginx-deployment   2         2         2            2           23s

  3. Lastly, delete the NGINX deployment to finish up this connection test for kubernetes-cluster1.

    Run the following command in your CLI:

    kubectl delete deployment nginx-deployment

    and the deployment will be deleted.

Test the connection to kubernetes-cluster2

Now you will do the same to test kubernetes

  1. Connect the second Kubernetes cluster to the kubeconfig in order to send kubectl commands to it.

    If using Edge-LB Auto Pools, in your CLI, paste in the following, again, your EDGELB_PUBLIC_AGENT_IP should still be cached in your terminal environment from the previous section:

    dcos kubernetes cluster kubeconfig \
    --insecure-skip-tls-verify \
    --context-name=kubernetes-cluster2 \
    --cluster-name=kubernetes-cluster2 \
    --apiserver-url=https://${EDGELB_PUBLIC_AGENT_IP}:6443/kubernetes-cluster2

    If not using Edge-LB Auto Pools, enter the following snippet:

    dcos kubernetes cluster kubeconfig \
    --insecure-skip-tls-verify \
    --context-name=kubernetes-cluster2 \
    --cluster-name=kubernetes-cluster2 \
    --apiserver-url=https://${EDGELB_PUBLIC_AGENT_IP}:6444

    Your kubectl commands should now connect to kubernetes-cluster2.

    Once the 2 contexts are configured, they will be saved in ~/.kube/config. To switch, you can either set the context generally:

    kubectl config use-context kubernetes-cluster2

    Or use the flag --context kubernetes-cluster1 in commands like so:

    kubectl get nodes --context kubernetes-cluster1

  2. As before for kubernetes-cluster1, get the nodes’ information for kubernetes-cluster2.

    In your CLI, enter:

    kubectl get nodes

    and note that the output should show that you are now using kubernetes-cluster2, similar to the following:

    $ kubectl get nodes
NAME                                                      STATUS   ROLES    AGE    VERSION
kube-control-plane-0-instance.kubernetes-cluster2.mesos   Ready    master   145m   v1.16.9
kube-node-0-kubelet.kubernetes-cluster2.mesos             Ready    <none>   142m   v1.16.9

  3. Next, create a NGINX deployment on kubernetes-cluster1:

    Enter the following in your CLI:

    kubectl apply -f https://k8s.io/examples/application/deployment.yaml

    and the deployment will begin.

  4. View that NGINX deployment in action:

    kubectl get deployments

    The resulting output should look similar to below:

    $ kubectl get deployments
NAME               DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
nginx-deployment   2         2         2            2           23s
Delete NGINX deployment:

  5. Finally, delete your test deployment.

    Simply enter:

    kubectl delete deployment nginx-deployment

    and the deployment will be shut down.

Access the Kubernetes Dashboard from your browser

In one simple but satisfying validation of our connection into our Kubernetes cluster manager from outside the cluster, we will access the Kubernetes Dashboard and view our cluster currently in context.

  1. Access the Kubernetes dashboard.

    In your CLI, enter:

    kubectl proxy

    which should start the proxy server.

  2. Then, navigate to: http://127.0.0.1:8001/api/v1/namespaces/kube-system/services/https:kubernetes-dashboard:/proxy/ and you should see your Kubernetes dashboard and information.

When the Kubernetes login screen is shown, you should choose the Kubeconfig option, click the Choose kubeconfig file text box and pick the location of your kubeconfig file (typically, $HOME/.kube/config).

Note: When accessed, and depending on whether you are running DC/OS or DC/OS EE (as well as on your browser’s configuration) you may be presented a warning indicating that the TLS certificate being used by the Kubernetes Dashboard is not trusted. It is generally safe to permanently trust this TLS certificate by adding an exception in your browser, or to skip past it. To learn more about TLS certificates, visit the Kubernetes Dashboard section.

Mission Complete!

Well done! You have successfully completed the Getting Started Guide for Kubernetes. You have set up your DC/OS Enterprise cluster to be able to run Kubernetes as a service on DC/OS.

For more information regarding operating Kubernetes as a service on DC/OS, check out the Operations section.