2.1.1.2. Version Control Integration with Pipelines

2.1.1.2. Version Control Integration with Pipelines

The connection between version control and your pipeline determines how fast your team gets feedback. A poorly configured trigger means developers wait minutes (or hours) for builds that should start in seconds.

Trigger mechanisms vary by source provider. CodeCommit repositories trigger pipelines via EventBridge rules that detect referenceCreated or referenceUpdated events. GitHub connections use webhooks via AWS CodeStar Connections (now CodeConnections). S3 sources require CloudTrail logging on the bucket — without it, polling is the only option (slow and costly).

Branch strategies for pipelines: teams practicing continuous deployment prefer trunk-based development — short-lived feature branches (hours to a day) merged into main frequently, which keeps merge conflicts small; long-lived GitFlow develop/release/hotfix branches suit scheduled releases. A common multi-environment pattern uses main for production, develop for staging, and feature branches for development. CodePipeline can be configured with multiple pipelines per repository — one triggered by main (deploys to prod) and another by develop (deploys to staging). Alternatively, use a single pipeline with a manual approval action before the production deploy stage.

Exam Trap: A V1 pipeline starts on every change to the single branch configured in its source action. CodePipeline V2 pipelines add trigger filters on connection-based sources (GitHub, GitLab, Bitbucket via CodeConnections): include/exclude by branch (glob patterns such as release/*), file path (e.g., only src/**), Git tag, or pull-request event. For CodeCommit or S3 sources, which use automated EventBridge change detection instead of triggers, branch-based routing still means EventBridge rules with content filtering on the branch name. EventBridge also retries delivery to its targets, whereas a failed webhook delivery is not retried automatically.

💡 Tip: Always disable PollForSourceChanges when using event-based triggers. Polling adds latency and costs API calls against your source provider.