KOps-Cluster-Helm-Deployment

Problem Statement

• Create a K8S cluster on AWS account using kops (1 master, 1 worker node) via terraform.
• Create a CI-CD pipeline that would build/perform code.
• Use any code scanner to identify vulnerabilities/code quality (Example-SonarQube or SonarCloud)
• Setup Quality Gates in SonarQube to 70%.
• If code does not meet code quality job should fail and email should be sent to relevant stakeholders.
• If code passes 70% quality, then it should proceed with next steps defined in pipeline.
• Store all the artifacts with proper versioning in any DMS like artifactory/nexus.
• Create a docker file to containerize application and push to ECR with proper tagging.
• Pull the latest image from ECR and deploy using Helm charts on a K8S cluster [setup to be done via Kops]
• Deploy few more applications on cluster (nginx, tomcat) so that multiple applications are hosted on same K8S Cluster.
• Use Prometheus to capture metrics of these applications and visualize the same on Grafana Dashboard.

Jenkins Plugins Used

• SonarQube Scanner
• Nexus Artifactory Uploader
• Pipeline Utility Steps
• AWS Credentials
• Docker
• Amazon ECR

Installations

• AWS CLI => run "aws configure", then set access-key, secret-key, region.
• KOps => refer to their latest stable release from official GitHub repo
• Terraform
• kubectl
• helm

---

Extra Info:

1. Check out more info regarding kops commands.

   • kops create cluster --help | less

2. Get instance groups.

   • kops get ig --name vikrant.k8s.local

3. SSH inside the control or worker node

   • ssh to the master: ssh -i ~/.ssh/id_rsa ubuntu@

---

Next Steps

1. SetUp Kubernetes Cluster using KOps on AWS

1. Create an S3 bucket, and its enable versioning.

   • aws s3api create-bucket --bucket vikrant-kops-store --region us-east-1
   • aws s3api put-bucket-versioning --bucket vikrant-kops-store --versioning-configuration Status=Enabled

2. Export the S3 bucket, so that we don't have to specify the bucket with --state flag with every KOps command.

   • export KOPS_STATE_STORE="s3://vikrant-kops-store"

3. Create Kubernetes Cluster.

   • kops create cluster --name vikrant.k8s.local --zones us-east-1a
     --control-plane-size t2.medium --node-size t2.medium --node-count 1
     --out . --target terraform
   • terraform init
   • terraform plan
   • terraform apply
   • kops export kubecfg --name vikrant.k8s.local --state s3://vikrant-kops-store --admin
   • kops get clusters (It will list all the clusters)

   OR

   • kops create cluster --name vikrant.k8s.local --zones us-east-1a
     --control-plane-size t2.medium --node-size t2.medium --node-count 1
   • kops edit cluster --name vikrant.k8s.local (if, we want to make any changes to the cluster configuration)
   • kops edit ig nodes-us-east-1a --name vikrant.k8s.local (if, we want to make any changes to the worker node configuration)
   • kops edit ig control-plane-us-east-1a --name vikrant.k8s.local (if, we want to make any changes to the master node configuration)
   • kops update cluster --name vikrant.k8s.local --yes
   • kops rolling-update cluster --name vikrant.k8s.local --yes --cloudonly (run this command, if made any changes to the cluster/node configuration)

4. Verify the cluster creation setup (cluster creation may take up to 10-15 min).

   • kops validate cluster --name vikrant.k8s.local
   • kubectl get nodes (If, correctly setup, all nodes master and worker will show up)
   • kubectl get pods -n kube-system

5. Delete the cluster.

   • kops delete cluster --name vikrant.k8s.local --yes

2. Deploy the application on the cluster using Helm Chart

1. Authenticate Docker to ECR.

   • aws ecr get-login-password --region eu-north-1 | docker login --username AWS --password-stdin 876724398547.dkr.ecr.eu-north-1.amazonaws.com/vikrantkatoch/calculator-app

2. Pull the latest docker image from ECR.

   • docker pull 876724398547.dkr.ecr.eu-north-1.amazonaws.com/vikrantkatoch/calculator-app:${BUILD_NUMBER}

3. Create Helm Chart

   • helm create calculator-app

4. Modify the image section inside the values.yaml file.

5. Deploy Your Application to the Kubernetes Cluster

   • kubectl create namespace test-1
   • helm install calc-app ./calculator-app --namespace test-1

6. Verify the Deployment

   • helm list --namespace test-1 (List all the helm deployments)
   • kubectl get pods --namespace test-1 (Check your Kubernetes pods)
   • kubectl logs calc-app-calculator-app-7476d77676-8kjsc --namespace test-1 (Check logs or troubleshoot a pod)
   • kubectl describe pod calc-app-calculator-app-7476d77676-8kjsc --namespace test-1
   • kubectl get services --namespace test-1
   • kubectl get all -n test-1

7. Update the Deployment (If needed)

   • helm upgrade calc-app ./calculator-app --namespace test-1
   • helm upgrade --install calc-app ./calculator-app --namespace test-1
     --set image.repository=876724398547.dkr.ecr.eu-north-1.amazonaws.com/vikrantkatoch/calculator-app,image.tag=46

8. Clean Up (If needed)

   • helm delete calc-app --namespace test-1
   • kops delete cluster --name vikrant.k8s.local --yes

3. SetUp Prometheus-Grafana Monitoring using Helm

• helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
• helm repo update
• helm upgrade --install prometheus prometheus-community/kube-prometheus-stack --namespace test-1
• kubectl patch svc prometheus-grafana -p '{"spec": {"type": "LoadBalancer"}}' -n test-1 (Change the prometheus-grafana service from "ClusterIP" type to "LoadBalancer" type)