This project helps you build a complete Amazon EKS cluster with nodegroup and CI/CD pipeline with CodeBuild 100% in AWS CDK.
This stack provisions the following resources with AWS CDK
- Amazon EKS cluster
- Amazon EKS nodegroup (with 2 m5.large workers)
- AWS CodeBuild Project for Amazon EKS CI/CD
- AWS CodeCommit as a sample source repo
- Amazon ECR repository
- AWS CDK >= 1.82.0 - check Getting Started with AWS CDK to setup your CDK environment. Run
cdk --versionto check the CLI version. - Docker - AWS CDK will build a docker image from local for codebuild environment. Make sure you have installed docker in your client.
Just deploy the stack with AWS CDK
$ git clone https://github.com/aws-samples/amazon-eks-cicd-codebuild.git
$ cd amazon-eks-cicd-codebuild
# install required packages
$ yarn install
# if you have not used cdk in this account previously you may be advised to create the necessary resources in the account
$ cdk bootstrap aws://ACCOUNTNUMBER/us-east-1
# check the diff before deployment
$ cdk diff
# deploy the complete stack
$ cdk deploy
# when finished with the demo delete the created resources
# note that the flask resources were created independently of the cdk and must be deleted first
$ kubectl delete svc/flask-svc deploy/flask-deployment
$ cdk destroyWhen you complete the cdk deploy, an empty CodeCommit repository will be created(check Resource List above to see all resource being created)
Outputs:
eks-cicd-codebuild-stack.ClusterConfigCommand43AAE40F = aws eks update-kubeconfig --name Cluster9EE0221C-96a69d282e344526afd711d2d3014150 --region ap-northeast-1 --role-arn arn:aws:iam::112233445566:role/eks-cicd-codebuild-stack-ClusterMastersRole9AA3562-1Q06U5NB9DP2Q
eks-cicd-codebuild-stack.ClusterGetTokenCommand06AE992E = aws eks get-token --cluster-name Cluster9EE0221C-96a69d282e344526afd711d2d3014150 --region ap-northeast-1 --role-arn arn:aws:iam::112233445566:role/eks-cicd-codebuild-stack-ClusterMastersRole9AA3562-1Q06U5NB9DP2Q
eks-cicd-codebuild-stack.CodeCommitCloneUrlHttp = https://git-codecommit.ap-northeast-1.amazonaws.com/v1/repos/eks-cicd-codebuild-stack-repo
eks-cicd-codebuild-stack.CodeCommitCloneUrlSsh = ssh://git-codecommit.ap-northeast-1.amazonaws.com/v1/repos/eks-cicd-codebuild-stack-repo
eks-cicd-codebuild-stack.CodeCommitRepoArn = arn:aws:codecommit:ap-northeast-1:112233445566:eks-cicd-codebuild-stack-repo
eks-cicd-codebuild-stack.CodeCommitRepoName = eks-cicd-codebuild-stack-repoVerify the Amazon EKS is running kubectl
# copy the 'aws eks update-kubeconfig' command string and run it in the terminal to generate/update the kubeconfig
$ aws eks update-kubeconfig --name Cluster9EE0221C-96a69d282e344526afd711d2d3014150 --region ap-northeast-1 --role-arn arn:aws:iam::112233445566:role/eks-cicd-codebuild-stack-ClusterMastersRole9AA3562-1Q06U5NB9DP2Q
# list the nodes with kubectl
$ kubectl get no
# deploy the initial flask sample service
$ kubectl apply -f flask-docker-app/k8s/flask.yaml
# list the service and deployment
$ kubectl get svc,deployCopy the ELB dns name from the EXTERNAL-IP column and open it in browser.
You will see the Flask-demo homepage.
# copy the ELB dns name from the EXTERNAL-IP column and open it in browser.
# You will see the Flask-demo homepage
# set codecommit as another upstream
$ git remote add codecommit ssh://git-codecommit.ap-northeast-1.amazonaws.com/v1/repos/eks-cicd-codebuild-stack-repo
# push all current repo to codecommit. This will trigger CodeBuild for CI/CD.
$ git push -u codecommit masterCheck the CodeBuild console to see the build status.
On build complete, reload the browser and see the Flask-demo homepage again. You will see the running platform string has changed from Amazon Web Services to Amazon EKS.
You may edit the Dockerfile in flask-docker-app directory and specify different PLATFORM value
ENV PLATFORM 'Amazon EKS'After you git add, git commit and git push to the CodeCommit source repository, CodeBuild will rebuild the docker image with new tag, push to Amazon ECR and immediately update the kubernetes deployment again. You may reload the browser to see the changes.
Q: when I cdk deploy, I got can't find CDK_DEFAULT_REGION or CDK_DEFAULT_ACCOUNT error.
A: You need configure your CDK environment, check this chapter in AWS CDK Developer Guide to configure your Environment correctly.
Q: when I cdk deploy, I got error while load metadata for public.ecr.aws/amazonlinux/amazonlinux:latest.
A: Try to run below command.
docker pull public.ecr.aws/amazonlinux/amazonlinux:latest
If you encounter the following error
Error response from daemon: pull access denied for public.ecr.aws/amazonlinux/amazonlinux, repository does not exist or may require 'docker login': denied: Your authorization token has expired. Reauthenticate and try again.
You may need docker login to authenticate against the ecr-public like
aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/amazonlinux
Q: How can I deploy to a default VPC or any existing one?
A:
# deploy to the default VPC
$ cdk deploy -c use_default_vpc=1
# Or to deploy to VPC ID vpc-xxxxxx
$ cdk deploy -c use_vpc_id=vpc-xxxxxx(You are encouraged to run cdk diff with the -c to view the difference first.)
However, if you create a new VPC, you might not be able to cdk destroy it after you kubectl apply some services in this VPC. Check #5 for more details and instructions.
Q: Got VPC is not symmetric error for default VPC
A: The ec2.Vpc.fromLookup() assumes the VPC specified is symmetric, check the doc here. If you specify the default VPC and get this error, you probably need to check if you have public and private subnets in every AZ and make sure they are symmetric. (Ref: aws/aws-cdk/issues/3407)
Q: Can I build this with my existing Amazon EKS cluster?
A: Yes. You can import the existing Amazon EKS cluster with eks.Cluster.fromClusterAttributes()
Q: The cdk destroy fails due to dependencies that cannot be deleted
A: The flask app dependencies must be deleted with kubectl before the cdk destroy
This library is licensed under the MIT-0 License. See the LICENSE file.