Connecting and deploying to an Amazon EKS cluster

In this tutorial, we will show how to integrate an Amazon EKS cluster with GitLab and begin deploying applications.

Introduction

For an end-to-end walkthrough we will:

Start with a new project based on the sample Ruby on Rails template.
Integrate an EKS cluster.
Utilize Auto DevOps to build, test, and deploy our application.

You will need:

An account on GitLab, like GitLab.com.
An Amazon EKS cluster (with worker nodes properly configured).
kubectl installed and configured for access to the EKS cluster.

If you don't have an Amazon EKS cluster, one can be created by following the EKS getting started guide.

Creating a new project

On GitLab, create a new project by clicking on the + icon in the top navigation bar and selecting New project.

On the new project screen, click on the Create from template tab, and select "Use template" for the Ruby on Rails sample project.

Give the project a name, and then select Create project.

Connecting the EKS cluster

From the left side bar, hover over Operations > Kubernetes > Add Kubernetes cluster, then click Add an existing Kubernetes cluster.

A few details from the EKS cluster will be required to connect it to GitLab:

Retrieve the certificate: A valid Kubernetes certificate is needed to authenticate to the EKS cluster. We will use the certificate created by default. Open a shell and use kubectl to retrieve it:
- List the secrets with kubectl get secrets, and one should named similar to default-token-xxxxx. Copy that token name for use below.
- Get the certificate with:
```
kubectl get secret <secret name> -o jsonpath="{['data']['ca\.crt']}" | base64 -D
```

Create admin token: A cluster-admin token is required to install and manage Helm Tiller. GitLab establishes mutual SSL auth with Helm Tiller and creates limited service accounts for each application. To create the token we will create an admin service account as follows:

2.1. Create a file called eks-admin-service-account.yaml with contents:

apiVersion: v1
kind: ServiceAccount
metadata:
  name: eks-admin
  namespace: kube-system

2.2. Apply the service account to your cluster:

kubectl apply -f eks-admin-service-account.yaml

Output:

serviceaccount "eks-admin" created

2.3. Create a file called eks-admin-cluster-role-binding.yaml with contents:

apiVersion: rbac.authorization.k8s.io/v1beta1
kind: ClusterRoleBinding
metadata:
  name: eks-admin
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: cluster-admin
subjects:
- kind: ServiceAccount
  name: eks-admin
  namespace: kube-system

2.4. Apply the cluster role binding to your cluster:

kubectl apply -f eks-admin-cluster-role-binding.yaml

Output:

clusterrolebinding "eks-admin" created

2.5. Retrieve the token for the eks-admin service account:

kubectl -n kube-system describe secret $(kubectl -n kube-system get secret | grep eks-admin | awk '{print $1}')

Copy the <authentication_token> value from the output:

Name:         eks-admin-token-b5zv4
Namespace:    kube-system
Labels:       <none>
Annotations:  kubernetes.io/service-account.name=eks-admin
              kubernetes.io/service-account.uid=bcfe66ac-39be-11e8-97e8-026dce96b6e8

Type:  kubernetes.io/service-account-token

Data
====
ca.crt:     1025 bytes
namespace:  11 bytes
token:      <authentication_token>

The API server endpoint is also required, so GitLab can connect to the cluster. This is displayed on the AWS EKS console, when viewing the EKS cluster details.

You now have all the information needed to connect the EKS cluster:

Kubernetes cluster name: Provide a name for the cluster to identify it within GitLab.
Environment scope: Leave this as * for now, since we are only connecting a single cluster.
API URL: Paste in the API server endpoint retrieved above.
CA Certificate: Paste the certificate data from the earlier step, as-is.
Paste the token value.
Project namespace: This can be left blank to accept the default namespace, based on the project name.

Click on Add Kubernetes cluster, the cluster is now connected to GitLab. At this point, Kubernetes deployment variables will automatically be available during CI/CD jobs, making it easy to interact with the cluster.

If you would like to utilize your own CI/CD scripts to deploy to the cluster, you can stop here.

Disable Role-Based Access Control (RBAC) (optional)

When connecting a cluster via GitLab integration, you may specify whether the cluster is RBAC-enabled or not. This will affect how GitLab interacts with the cluster for certain operations. If you did not check the "RBAC-enabled cluster" checkbox at creation time, GitLab will assume RBAC is disabled for your cluster when interacting with it. If so, you must disable RBAC on your cluster for the integration to work properly.

Presently, Auto DevOps and one-click app installs do not support Kubernetes role-based access control. Support is being worked on, but in the interim RBAC must be disabled to utilize for these features.

NOTE: Note: Disabling RBAC means that any application running in the cluster, or user who can authenticate to the cluster, has full API access. This is a security concern, and may not be desirable.

To effectively disable RBAC, global permissions can be applied granting full access:

kubectl create clusterrolebinding permissive-binding \
  --clusterrole=cluster-admin \
  --user=admin \
  --user=kubelet \
  --group=system:serviceaccounts

Deploy services to the cluster

GitLab supports one-click deployment of helpful services to the cluster, many of which support Auto DevOps. Back on the Kubernetes cluster screen in GitLab, a list of applications is now available to deploy.

First, install Helm Tiller, a package manager for Kubernetes. This enables deployment of the other applications.

Deploying NGINX Ingress (optional)

Next, if you would like the deployed app to be reachable on the internet, deploy the Ingress. Note that this will also cause an Elastic Load Balancer to be created, which will incur additional AWS costs.

Once installed, you may see a ? for "Ingress IP Address". This is because the created ELB is available at a DNS name, not an IP address. To get the DNS name, run:

kubectl get service ingress-nginx-ingress-controller -n gitlab-managed-apps -o jsonpath="{.status.loadBalancer.ingress[0].hostname}"

Note that you may see a trailing % on some Kubernetes versions, do not include it.

The Ingress is now available at this address and will route incoming requests to the proper service based on the DNS name in the request. To support this, a wildcard DNS CNAME record should be created for the desired domain name. For example, *.myekscluster.com would point to the Ingress hostname obtained earlier.

Deploying the GitLab Runner (optional)

If the project is on GitLab.com, free shared Runners are available and you do not have to deploy one. If a project specific Runner is desired, or there are no shared Runners, it is easy to deploy one.

Simply click on the Install button for the GitLab Runner. It is important to note that the Runner deployed is set as privileged, which means it essentially has root access to the underlying machine. This is required to build docker images, and so is on by default.

Deploying Prometheus (optional)

GitLab is able to monitor applications automatically, utilizing Prometheus. Kubernetes container CPU and memory metrics are automatically collected, and response metrics are retrieved from NGINX Ingress as well.

To enable monitoring, simply install Prometheus into the cluster with the Install button.

Create a default Storage Class

Amazon EKS doesn't have a default Storage Class out of the box, which means requests for persistent volumes will not be automatically fulfilled. As part of Auto DevOps, the deployed Postgres instance requests persistent storage, and without a default storage class it will fail to start.

If a default Storage Class doesn't already exist and is desired, follow Amazon's guide on storage classes to create one.

Alternatively, disable Postgres by setting the project variable POSTGRES_ENABLED to false.

Deploy the app to EKS

With RBAC disabled and services deployed, Auto DevOps can now be leveraged to build, test, and deploy the app.

Enable Auto DevOps if not already enabled. If a wildcard DNS entry was created resolving to the Load Balancer, enter it in the domain field under the Auto DevOps settings. Otherwise, the deployed app will not be externally available outside of the cluster.

A new pipeline will automatically be created, which will begin to build, test, and deploy the app.

After the pipeline has finished, your app will be running in EKS and available to users. Click on CI/CD > Environments.

You will see a list of the environments and their deploy status, as well as options to browse to the app, view monitoring metrics, and even access a shell on the running pod.

Learn more

To learn more on automatically deploying your applications, read about Auto DevOps.