deploy-aws-s3-cloudfront-with-cache-contol

Syncs a local directory to an AWS S3 bucket, optionally invalidating affected CloudFront paths.

Installation

npm install --save deploy-aws-s3-cloudfront-with-cache-contol

Authentication

This packages uses the AWS SDK for Node.js and defers authentication to the SDK.

If you are relying on credentials stored in ~/.aws/credentials you can use the --profile option to specify a named profile, if required.

Usage

deploy-aws-s3-cloudfront-with-cache-contol --bucket <bucket> [options]

Options

--bucket <name>

The name of the S3 bucket to sync to.

--source <path>

Path to local directory to sync from.

Default value .

--distribution <ID>

The CloudFront distribution ID to invalidate after successful deployment.

If omitted, no invalidation will be performed.

--exclude <pattern> [--exclude <pattern>...]

Exclude local paths from being synced to the bucket. Refer to the fast-glob documentation for supported patterns.

Multiple paths can be specified by passing multiple --exclude options.

--delete

If used, objects that do not exist locally will be deleted from the bucket.

--invalidation-path <path>

When used with the --distribution option, this can be used to set the invalidation path. If omitted, only the added, modified and deleted objects (if --delete option is used) are invalidated.

This option is typically used to reduce invalidation costs by using a wildcard pattern (e.g. --invalidation-path "/*").

--profile <name>

If depending on a named profile in ~/.aws/credentials for authentication, use this option to provide the profile name.

--non-interactive

Never prompt for confirmation. This is particularly useful for automated deployment pipelines.

--cache-control

Set cache control headers

Installation as a run-script alias (optional)

Add a deploy script alias to your package.json file:

{
  ...
  "scripts": {
    ...
    "deploy": "deploy-aws-s3-cloudfront-with-cache-contol --bucket my-bucket"
  }
}

Run npm run build to build then npm run deploy to deploy.

If you need to pass user-level options that you don't want committed into package.json, the you can provide these options at call-time, e.g. npm run deploy -- --profile <profile>.

Configuration for create-react-app projects

Set the --source option to /.build/:

{
  ...
  "scripts": {
    ...
    "deploy": "deploy-aws-s3-cloudfront-with-cache-contol --bucket my-bucket --source=./build/"
  }
}

Then simply run npm run build then npm run deploy to deploy the latest build output.

Alternatively, you can force a build before every deployment:

{
  ...
  "scripts": {
    ...
    "deploy": "npm run build && deploy-aws-s3-cloudfront-with-cache-contol --bucket my-bucket --source=./build/"
  }
}

Alternatives (and why this package exists!)

• AWS S3 Sync (bundled with AWS CLI)

  The aws s3 sync command uses the modification time to identify modified assets. This doesn't work well when building a project often involves regenerating files with fresh timestamps but identical content.

  This package will instead perform a checksum comparison to minimise the deployment payload. The MD5 checksum will be computed against local files then compared against the ETag of the corresponding remote objects.

• react-deploy-s3

  For React apps, the react-deploy-s3 provides similar behaviour to this package. However, react-deploy-s3 expects your AWS credentials to be passed in as command arguments and requires additional configuration to get set up. In contrast, this package defers authentication to the AWS SDK and therefore supports multiple authentication strategies (e.g. IAM roles, environment variables and profiles).

  Additionally, react-deploy-s3 will purge everything from your S3 bucket before re-uploading the entire build directory. Here, however, deployments are incremental resulting in a smaller payload and minimal interruption. Likewise, this package will perform a more efficient CloudFront purge by executing an invalidation on the affected paths only, as opposed to a site-wide refresh as performed by react-deploy-s3.