Lab 03 - Deploy Apps on EKS

In this project , you would write definitions for deploying the instavote application stack with all components/tiers which include,

  • vote
  • redis
  • worker
  • db
  • result

Project Specs

  • Clone the code with git clone https://github.com/initcron/k8s-code.git and switch to k8s-code/projects/instavote/dev path.
  • Apply the deployment and service code for the applications marked as ready
  • Complete the code for deployments and services marked as TODO
  • Apply the definitions that you have completed
  • Validate the application workflow is operational by loading vote and result applications from the browser

⠀ Following table depicts the state of readiness of the above services.

App Deployment Service
vote TODO ready
redis ready ready
worker TODO n/a
db ready ready
result TODO TODO

Phase I - Apply existing code

Create a namespace and switch to it

kubectl create namespace instavote
kubectl config set-context --current --namespace=instavote
kubectl config get-contexts

Apply the existing manifests

cd k8s-code/projects/instavote/dev/

kubectl apply  -f vote-svc.yaml
kubectl apply -f redis-deploy.yaml -f redis-svc.yaml
kubectl apply -f db-deploy.yaml -f db-svc.yaml

validate

kubectl get all

Where you should see,

  • deplyoment and services for redis and db created
  • nodeport service for vote app

⠀ If you see the above objects, proceed with the next task.

Phase II - Migrate to a New Node Group

You may notice that your pods are not running or stuck in containerCreating stage. This could be due to the insufficient capacity due to the t2.micro instance types. You could mitigate this by migrating to a new Node Group.

This could be done by adding a new node group e.g.


....


managedNodeGroups:
  - name: ng-1-workers
    labels: { role: workers }
    instanceType: t3.micro
    desiredCapacity: 1
    minSize: 1
    maxSize: 5
    maxPodsPerNode: 17
    ssh:
      allow: true
      publicKeyName: eks-spore
    tags:
      k8s.io/cluster-autoscaler/enabled: "true"
      k8s.io/cluster-autoscaler/eks-cluster-01: "owned"
    updateConfig:
      maxUnavailable: 1

  - name: ng-2-workers
    labels: { role: workers }
    instanceType: t3.small
    desiredCapacity: 2
    minSize: 1
    maxSize: 4
    maxPodsPerNode: 17
    ssh:
      allow: true
      publicKeyName: eks-spore
    tags:
      k8s.io/cluster-autoscaler/enabled: "true"
      k8s.io/cluster-autoscaler/eks-cluster-01: "owned"
    updateConfig:
      maxUnavailable: 1

Notice the ng-2-workers node group definition above. It will create a new set of nodes using t3.medium instance type.

To launch the new node group: 

eksctl create nodegroup -f cluster.yaml --include=ng-2-workers

and to delete the previous one:

eksctl delete nodegroup -f cluster.yaml --include=ng-1-workers --approve

This should help you migrate the workloads to the new node group and help it run with sufficient resources.

You could also scale it up with

eksctl scale nodegroup --cluster eks-cluster-01 ng-2-workers --nodes 2

Phase III - Create Deployments and Services for Remaining Services

You may find the files available in the same directory as above i.e. k8s-code/projects/instavote/dev/ with either partial or blank code. Your job is to complete the deployment and service yaml specs and apply those. While writing the specs, you could refer to the following specification.

  • vote
    • image: schoolofdevops/vote:v1
    • application port: 80
    • replicas: 2
    • service type: NodePort
    • nodePort : 30000
  • worker
    • image: schoolofdevops/worker:latest
  • results
    • image: schoolofdevops/vote-result
    • application port: 80
    • replicas: 1
    • service type: NodePort
    • nodePort : 30100

⠀

To Validate:

kubectl get all

The above command should show, five deployments and four services created services for vote and result app should have been exposed with NodePort

Find out the NodePort for vote and service apps and load those from your browser.

This is how the vote application should look like.

You could also load the Result page on the 30100 port of your Node IP/Address as,

Above is how the result app should look like.

Final validation is, if you submit the vote, you should see the results changing accordingly. You would see this behavior only if all deployments and services are created and configured properly.