Skip to main content

AWS CodePipeline

Core concept: CodePipeline orchestrates your entire release process — source → build → test → deploy — end-to-end.

Pipeline Anatomy​

Stage 1: Source          Stage 2: Build         Stage 3: Deploy
──────────────────       ──────────────────      ──────────────────
Action: CodeCommit    →  Action: CodeBuild    →  Action: CodeDeploy
(output: source.zip)     (output: build.zip)     (to staging)
                                                       ↓
                         Stage 4: Approve       Stage 5: Deploy
                         ──────────────────      ──────────────────
                         Action: Manual      →  Action: CloudFormation
                         Approval                (to production)

Action Types​

CategoryProviders
SourceCodeCommit, S3, GitHub, ECR
BuildCodeBuild, Jenkins
TestCodeBuild, Device Farm, third-party
DeployCodeDeploy, Beanstalk, ECS, CloudFormation, S3
ApprovalManual approval
InvokeLambda, Step Functions

Artifact Store​

Every stage passes artifacts via S3:

CodeCommit source → S3 artifact (source.zip)
                                ↓
                     CodeBuild reads → builds → outputs build.zip
                                                        ↓
                                            CodeDeploy reads build.zip

S3 artifacts are encrypted with KMS for cross-account pipelines.

Manual Approval​

# CloudFormation
ApprovalStage:
  Type: StageDeclaration
  Name: ManualApproval
  Actions:
    - Name: ApproveProduction
      ActionTypeId:
        Category: Approval
        Owner: AWS
        Provider: Manual
        Version: "1"
      Configuration:
        NotificationArn: !Ref ApprovalSNSTopic
        CustomData: "Review test results at: https://dashboard.example.com"
        ExternalEntityLink: "https://jira.example.com/RELEASE-123"

EventBridge Integration​

React to pipeline state changes:

// EventBridge rule: alert on pipeline failure
{
  "source": ["aws.codepipeline"],
  "detail-type": ["CodePipeline Pipeline Execution State Change"],
  "detail": {
    "state": ["FAILED"],
    "pipeline": ["my-prod-pipeline"]
  }
}

🧪 Practice Questions​

Q1. A pipeline needs to deploy to staging automatically, then wait for a human to approve before deploying to production. Which CodePipeline action provides this?

A) A CodeBuild step with approval logic
B) Lambda function that checks a DynamoDB approval flag
C) Manual Approval action between staging and prod stages
D) EventBridge rule that pauses the pipeline

✅ Answer & Explanation

C — CodePipeline's built-in Manual Approval action pauses execution and sends an SNS notification. The pipeline resumes only when an authorized user approves via console, CLI, or API.

Q2. Between stages in a CodePipeline, how are artifacts (build outputs) passed?

A) Directly via Lambda invocation
B) Via an SQS message
C) Via S3 — each stage reads/writes to the pipeline's artifact store bucket
D) In-memory within the pipeline

✅ Answer & Explanation

C — CodePipeline uses an S3 bucket as the artifact store. Each stage's output is uploaded to S3, and the next stage downloads it. For cross-account pipelines, the S3 bucket and KMS key must allow cross-account access.

🔗 Resources​