Wondering what’s next for npm?Check out our public roadmap! »

TypeScript icon, indicating that this package has built-in type declarations

1.2.0 • Public • Published

CDK Construct for deploying a webapp release

This project contains a CDK Construct for an AWS Lambda Function to handle deployment of a bundled static web application to a S3 bucket while preserving files from previous deployments within a time threshold.

What it does:

  1. Fetch deployment log from S3
  2. Fetch bundled artifact and extract locally with optional filtering
  3. Upload all non-html files to S3
  4. Upload html files to S3
  5. Add uploaded items to deployment log
  6. Delete old items from S3
  7. Prune old deployments from deployment log
  8. Store deployment log to S3 for next run
  9. Optionally invalidate CloudFront distribution

Preserving old files

A single-page application using code splitting will cause the client to defer loading lots of files. To avoid a deployment disrupting the user, we cannot delete the previous files (e.g. using aws s3 sync --delete), as that would cause the client to get 404 errors when navigating through the app. CloudFront will in many cases hide this issue for most of the users due to its edge caches, but it will still be an issue for users that haven't updated the application since the day before.

One way of handling this would be to never delete any files from the S3 bucket. That means the bucket will fill up with a lot of old files. This project approaches it differently by deleting old files.

A user that still uses an application deployed five days ago should not be disrupted. To keep this promise we keep the newest deployment that happened more than five days ago, and delete files from older ones that no longer have any reference to them.

Triggering a deployment

aws lambda invoke \
  --function-name my-deploy-lambda \
  --payload '{
    "artifactS3Url": "s3://my-bucket/my-release.tgz"
  }' \


Testing locally:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Adjust to your project. See config.py for full list.
export TARGET_BUCKET_URL=s3://my-website/web
export EXPIRE_SECONDS=86400
export DEPLOY_LOG_BUCKET_URL=s3://my-website/deployments.log

# Adjust artifact path.
python -m webapp_deploy.main s3://my-bucket/my-release.tgz


  • To avoid a race condition in using and updating the deployment log, no concurrent execution of the lambda should be allowed.

  • As we use S3 to hold the deployment log, there is a potential for a successive deployment to read an older deployment log. This is due to the fact that S3 is eventually consistent. If this happens the files that were deployed by the previous run (being overwritten) will be kept forever, unless they are also references by another deployment.


npm i @capraconsulting/webapp-deploy-lambda

DownloadsWeekly Downloads






Unpacked Size

50.8 kB

Total Files


Last publish


  • avatar
  • avatar
  • avatar
  • avatar
  • avatar
  • avatar
  • avatar
  • avatar
  • avatar
  • avatar
  • avatar