# Securing Kubernetes Dashboard

The following guide covers how to secure Kubernetes Dashboard using Pomerium. Kubernetes Dashboard is a powerful, web-based UI for managing Kubernetes clusters. Pomerium can act as a forward-auth provider and as an independent identity-aware access proxy improving and adding single-sign-on to Kubernetes Dashboard's default access control. This guide aims to demonstrate a concrete example of those two methods of access control.

fresh kubernetes dashboard install

This tutorial covers:


nginx-ingress version 0.26.2 contains a regression that breaks external auth and results in an infinite loop.

# Background

Though securing kubernetes dashboard as an example may seem contrived, the damages caused by an unsecured dashboard is a real threat vector. In late 2018, Telsa determined that the hackers who were running crypto-mining malware on their cloud accounts came in through an unsecured Kubernetes Dashboard instance.

tesla hacked from kubernetes dashboard

# Helm

First, we will install Helm. Helm is a package manager similar to apt-get or brew but for Kubernetes and it's what we'll use to install Pomerium, nginx-ingress, cert-manager, and the dashboard.

# Install

There are two parts to Helm: the client, and the server. This guide will cover the most common installation path. Please refer to the Helm install instructions for more details, and other options.

# Client

We'll install by installing the helm client on our client on our local machine.

OSX via homebrew.

brew install kubernetes-helm

Linux via snap.

sudo snap install helm --classic

A script for the trusting πŸ˜‰.

curl -L https://git.io/get_helm.sh | bash

Add the default repository

helm repo add pomerium https://helm.pomerium.io

# NGINX Ingress

NGINX ingress controller is a Kubernetes Ingress based on NGINX the is a very popular, full-feature reverse-proxy. We will use NGINX in two configurations: as a fronting proxy, and as proxy that delegates every request's access-control decision to Pomerium using forward-auth.

Also, please note that while this guide uses NGINX Ingress Controller, Pomerium can act as a forward auth-provider alongside other fronting ingresses like Traefik, Ambassador, and envoy in a similar fashion.

# Install

NGINX Ingress controller can be installed via Helm from the official charts repository. To install the chart with the release name helm-nginx-ingress:

helm install helm-nginx-ingress stable/nginx-ingress
NAME:   helm-nginx-ingress
NAMESPACE: default

Confirm the ingress has been installed, and that an external LoadBalancer IP has been set.

$kubectl get svc
NAME                                 TYPE           CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
helm-nginx-ingress-controller        LoadBalancer   localhost     80:31059/TCP,443:32402/TCP   15m
helm-nginx-ingress-default-backend   ClusterIP   <none>        80/TCP                       15m
kubernetes                           ClusterIP       <none>        443/TCP                      93m

We now have a kubernetes ingress resource that can be used to delegate access control decisions to or front-proxy for Pomerium.

# Certificates

Cert-manager is a Kubernetes plugin that helps automate issuance of TLS certificates. In our case, we will use cert-manager to retrieve certs to each of our configured routes.

# Install

Like in previous steps, we will use Helm to install Cert-manager.

# Install the CustomResourceDefinition resources separately
$ kubectl apply --validate=false -f https://github.com/jetstack/cert-manager/releases/download/v0.12.0/cert-manager.yaml
# Create the namespace for cert-manager
$ kubectl create namespace cert-manager
# Add the Jetstack Helm repository
helm repo add jetstack https://charts.jetstack.io
# Update your local Helm chart repository cache
helm repo update
# Install the cert-manager Helm chart
helm install \
  --namespace cert-manager \
  --version v0.12.0 \
  cert-manager \

And we'll confirm cert-manager is up and running.

$ kubectl get pods --namespace cert-manager
NAME                                       READY   STATUS    RESTARTS   AGE
cert-manager-756d9f56d6-brv6z              1/1     Running   0          23s
cert-manager-cainjector-74bb68d67c-7jdw6   1/1     Running   0          23s
cert-manager-webhook-645b8bdb7-8kgc9       1/1     Running   0          23s

# Configure

Now that cert-manager is installed, we need to make one more configuration to be able to retrieve certificates. We need to add a http-01 issuer for use with LetsEncrypt.

$ kubectl apply -f docs/recipes/yml/letsencrypt-prod.yaml
# letsencrypt-prod.yaml
# As LE has very strict usage limits, it is highly suggested you start with staging
# https://letsencrypt.org/docs/staging-environment/
apiVersion: cert-manager.io/v1alpha2
kind: Issuer
  name: letsencrypt-prod # letsencrypt-staging
  namespace: default
    server: https://acme-v02.api.letsencrypt.org/directory # https://acme-staging-v02.api.letsencrypt.org/directory
    email: your@email.example # your email :)
      name: letsencrypt-prod # letsencrypt-staging
      - selector: {}
            class: nginx

And confirm your issuer is set up correctly.

$ kubectl describe issuer
Name:         letsencrypt-prod
API Version:  cert-manager.io/v1alpha2
Kind:         Issuer
    Private Key Secret Ref:
      Name:  letsencrypt-prod
    Server:  https://acme-v02.api.letsencrypt.org/directory
      Http 01:
          Class:  nginx
    Last Registered Email:  ....
    Uri:                    https://acme-v02.api.letsencrypt.org/acme/acct/69070883
    Message:               The ACME account was registered with the ACME server
    Reason:                ACMEAccountRegistered
    Status:                True
    Type:                  Ready

If you see something like the above, cert-manager should be all set to help issue you new certificates when you create a new https protected ingress. Note, if you need wild-card certificates, you may also need a DNS-01 type issuer.

# Dashboard

Kubernetes Dashboard is a general purpose, web-based UI for Kubernetes clusters. It allows users to manage applications running in the cluster and troubleshoot them, as well as manage the cluster itself.

kubernetes dashboard login page

# Install

As with the previous steps, we can use Helm to install our instance of Kubernetes Dashboard.

helm install \
  helm-dashboard \
  stable/kubernetes-dashboard \
  --set ingress.enabled="false" \
  --set enableSkipLogin="true"

That's it. We've now configured kubernetes dashboard to use the default service account, if none-is provided. We've also explicitly told helm that we are going to deploy our own custom, nginx / Pomerium / cert-manager enabled ingress.

# Pomerium

Pomerium is an identity-aware access proxy that can used to serve as an identity-aware reverse proxy, or as a forward-auth provider.

# Configure

Before installing, we will configure Pomerium's configuration settings in config.yaml. Other than the typical configuration settings covered in the quick-start guides, we will add a few settings that will make working with Kubernetes Dashboard easier.

We can retrieve the token to add to our proxied policy's authorization header as follows.

$ kubectl describe secret helm-dashboard
Namespace:    default
Labels:       <none>
Annotations:  kubernetes.io/service-account.name: dashboard-kubernetes-dashboard
              kubernetes.io/service-account.uid: 18ab35ee-eca1-11e9-8c75-025000000001

Type:  kubernetes.io/service-account-token

ca.crt:     1025 bytes
namespace:  7 bytes
token:      eyJhbGciOiJSUzI1NiIsImtpZCI6IiJ9.......

The above token then needs to be assigned to our route configuration and policy.

# config.yaml
forward_auth_url: https://forwardauth.domain.example

  # this route is directly proxied by pomerium & injects the authorization header
  - from: https://dashboard-proxied.domain.example
    to: https://helm-dashboard-kubernetes-dashboard
      - user@domain.example
    tls_skip_verify: true # dashboard uses self-signed certificates in its default configuration
      Authorization: Bearer eyJhbGciOiJSUzI1NiIsImtpZCI6IiJ9.....

  # this route is indirectly checked for access using forward-auth
  - from: https://dashboard-forwardauth.domain.example
    to: https://helm-dashboard-kubernetes-dashboard
      - user@domain.example

We then add our configuration to Kubernetes as a ConfigMap.

# add our pomerium policy to kubernetes as a configmap
$ kubectl create configmap config --from-file="config.yaml"="config.yaml"

# Install

Finally, we get to install Pomerium! πŸŽ‰ Once again, we will use Helm to deploy Pomerium.

helm install \
	"helm-pomerium" \
	stable/pomerium \
	--set config.rootDomain="domain.example" \
	--set config.existingConfig="config" \
	--set authenticate.idp.provider="google" \
	--set authenticate.idp.clientID="YOUR_CLIENT_ID" \
	--set authenticate.idp.clientSecret="YOUR_SECRET"

# Putting it all together

Now we just need to tell external traffic how to route everything by deploying the following ingresses.

$kubectl apply -f docs/recipes/yml/pomerium.ingress.yaml
# pomerium.ingress.yaml
apiVersion: extensions/v1beta1
kind: Ingress
  name: pomerium-authenticate
    kubernetes.io/ingress.class: "nginx"
    cert-manager.io/issuer: "letsencrypt-prod" # see `le.issuer.yaml`
    nginx.ingress.kubernetes.io/backend-protocol: "HTTPS"
    - hosts:
        - authenticate.domain.example
      secretName: pomerium-authenticate-external-tls
    - hosts:
        - forwardauth.domain.example
      secretName: pomerium-forwardauth-external-tls

    - host: authenticate.domain.example
          - path: /
              serviceName: helm-pomerium-authenticate
              servicePort: https
    - host: forwardauth.domain.example
          - path: /
              serviceName: helm-pomerium-proxy
              servicePort: https
$kubectl apply -f docs/recipes/yml/dashboard-forwardauth.ingress.yaml
# dashboard-forwardauth.ingress.yaml
apiVersion: extensions/v1beta1
kind: Ingress
  name: dashboard-forwardauth
    kubernetes.io/ingress.class: "nginx"
    cert-manager.io/issuer: "letsencrypt-prod" # see `letsencrypt.issuer.yaml`
    nginx.ingress.kubernetes.io/backend-protocol: "HTTPS"
    nginx.ingress.kubernetes.io/auth-url: "https://forwardauth.domain.example/verify?uri=$scheme://$host$request_uri"
    nginx.ingress.kubernetes.io/auth-signin: "https://forwardauth.domain.example?uri=$scheme://$host$request_uri"

    - hosts:
        - dashboard-forwardauth.domain.example
      secretName: dashboard-forwardauth-tls
    - host: dashboard-forwardauth.domain.example
          - path: /
              serviceName: helm-dashboard-kubernetes-dashboard
              servicePort: https
$kubectl apply -f docs/recipes/yml/dashboard-proxied.ingress.yaml
# dashboard-proxied.ingress.yaml
apiVersion: extensions/v1beta1
kind: Ingress
  name: dashboard-proxied
    kubernetes.io/ingress.class: "nginx"
    cert-manager.io/issuer: "letsencrypt-prod" # see `letsencrypt.issuer.yaml`
    nginx.ingress.kubernetes.io/backend-protocol: "HTTPS"

    - hosts:
        - dashboard-proxied.domain.example
      secretName: dashboard-proxied-tls
    - host: dashboard-proxied.domain.example
          - path: /
              serviceName: helm-pomerium-proxy
              servicePort: https

And finally, check that the ingresses are up and running.

 $kubectl get ingress
NAME                    HOSTS                                                ADDRESS   PORTS     AGE
dashboard-forwardauth   dashboard-forwardauth.domain.example                               80, 443   42h
dashboard-proxied       dashboard-proxied.domain.example                               80, 443   42h
helm-pomerium           *.domain.example,authenticate.domain.example                   80, 443   42h

You'll notice this is the step where we put everything together. We've got nginx handling the initial requests, cert-manager handling our public certificates, and Pomerium handling access control.

# Conclusion

Though the net result will be similar between using forward-auth and direct proxying, there are a few differences:

  • By having Pomerium directly proxy the requests, you as an administrator have control control over the underlying request. In this example, we are able to inject an authenticating bearer token header to the downstream request which arguably makes for a better user experience.

  • Conversely, going the forward-auth route potentially means using the ingress / reverse proxy you are are already accustomed to or have already modified to support your particular deployment.

In the end, you should choose whichever option makes the most sense for your use-case and environment.

Whichever option you choose to go with, πŸŽ‰πŸΎπŸŽŠ congratulations πŸŽ‰πŸΎπŸŽŠ! You now have a single-sign-on enabled Kubernetes Dashboard protected by Pomerium and automatically renewing LetsEncrypt certificates.