This project is a rough version, will keep enriching.

EKS as Backend

• Refer to README_eks.md
• Notice: CodeDeploy and deploy stage of CodePipeline don’t support EKS yet.
• After deploy, it's needed to push code to CodeCommit and manually adjust buildspec.yaml to have the EKS resources successfully created, to be fixed.
• When destroy with new VPC created earlier, kubectl delete svc/flask-svc deploy/flask or manually delete load balancer, then cdk destroy.

ECS-EC2 as Backend Setup:

• Specify the stack to create in the cdk/bin/cdk.ts file
• npm run build
• cdk deploy
• Manually push code to CodeCommit repo and pipeline will be triggered Destroy:
• cdk destroy

ECS-Fargate as Backend Setup:

• Specify the stack to create in the cdk/bin/cdk.ts file
• npm run build
• cdk deploy
• Manually push code to CodeCommit repo and pipeline will be triggered Destroy:
• cdk destroy

ECS-Fargate as Backend enriched version 1.0: with CloudFormation to Deploy in the Pipeline Setup:

• Specify the stack to create in the cdk/bin/cdk.ts file
• npm run build
• cdk deploy ClusterStack
• cdk deploy CodePipelineStack, AppServiceDeployStack will be deployed as nested stack
• Manually push code to CodeCommit repo and pipeline will be triggered Destroy:
• cdk destroy AppServiceDeployStack, if AppServiceDeployStack is not destroyed fiest, it will fail，then grant permissions to the stack role and delete the stack.
• cdk destroy CodePipelineStack
• cdk destroy ClusterStack

TODO:

• Split pipeline into dev, uat, prod
• Support Blue/Green deploy, for references mageDetail.json File for Amazon ECS Blue/Green Deployment Actions, Amazon ECS Blue/Green Deployment
• Grant limited but enough permissions to service Role
• Expose moren configurations of the entire infrastructure
• Support Github repo, use Secrets Manager to store Github Credentials
• Support existed ECR repo, EKS/ECS cluster (eks.Cluster.fromClusterAttributes()), Code repo
• Evaluate using cdk lib( @aws-cdk/aws-eks) to manage EKS resources instead of kubectl cli, but then it might be not as convenient as managing k8s yaml, also parameters passing
• Support monitoring, auto-scaling EKS/ECS cluster (metrics-server), logging
• For Eks, LB, ECR repo, S3 resources need to be manually deleted; For ECS, ECR repo; S3 resources need to be manually deleted.
• Prepare a flask app with DB connection more close to real case.
• Use AWS SSM parameter to store configurations of CDK stack, even DB configurations.
• Clean up the original code/Readme a little bit.
• Refactor the project into support choosing modules(Repo: CodeCommit and Github; CD backend: EKS, ECS-Fargate, ECS-EC2 etc.)
• Create a frontend UI console for easy use.