The storage construct deploys S3 buckets to store files.
serverless plugin install -n serverless-liftservice: my-app
provider:
name: aws
constructs:
avatars:
type: storage
plugins:
- serverless-liftOn serverless deploy, a preconfigured S3 bucket will be created.
The storage construct creates and configures the S3 bucket for production:
- Files stored in the bucket are automatically encrypted (S3 takes care of encrypting and decrypting data on the fly, without change to our applications).
- File versioning is enabled to prevent any accidental data loss. Old versions are automatically purged after 30 days to avoid extra costs.
- Storage costs are optimized automatically via intelligent tiering.
To learn more about the architecture of this construct, read this article.
All storage constructs expose the following variables:
bucketName: the name of the deployed S3 bucketbucketArn: the ARN of the deployed S3 bucket
When publicPath is set, an additional variable is exposed:
publicUrl: the base URL of the bucket (e.g.https://<bucket>.s3.<region>.amazonaws.com), used to build URLs to public files
This can be used to reference the bucket from Lambda functions, for example:
constructs:
avatars:
type: storage
functions:
myFunction:
handler: src/index.handler
environment:
BUCKET_NAME: ${construct:avatars.bucketName}How it works: the ${construct:avatars.bucketName} variable will automatically be replaced with a CloudFormation reference to the S3 bucket.
By default, all the Lambda functions deployed in the same serverless.yml file will be allowed to read/write into the bucket and use common S3 object features such as ACLs, tags, multipart uploads, object attributes, and restores.
In the example below, there are no IAM permissions to set up: myFunction will be allowed to read and write into the avatars bucket.
constructs:
avatars:
type: storage
functions:
myFunction:
handler: src/index.handler
environment:
BUCKET_NAME: ${construct:avatars.bucketName}Automatic permissions can be disabled: read more about IAM permissions.
By default, files are encrypted using the default S3 encryption mechanism (free).
Alternatively, for example to comply with certain policies, it is possible to use KMS:
constructs:
avatars:
# ...
encryption: kmsYou can configure custom S3 lifecycle rules to automatically delete or transition objects:
constructs:
avatars:
type: storage
lifecycleRules:
- prefix: tmp/
expirationInDays: 1
- prefix: cache/
expirationInDays: 7The configuration maps directly to CloudFormation LifecycleConfiguration Rules with the following conveniences:
Status: Enabledis added by default (can be overridden)- Property names can be written in camelCase (e.g.
expirationInDays) or PascalCase (e.g.ExpirationInDays)
These rules are added to the default lifecycle rules that Lift adds (intelligent tiering and old version cleanup).
Since April 2023, S3 buckets have ACLs disabled by default. However, many tools and libraries (including PHP's Flysystem, used by Laravel) send ACL headers on S3 operations. Without enabling ACLs, these operations will fail.
To let the bucket accept ACL headers while keeping the bucket owner in full control:
constructs:
storage:
type: storage
allowAcl: trueThis sets the S3 bucket's Object Ownership to BucketOwnerPreferred.
To allow browser-based uploads (e.g. via presigned URLs), you can configure CORS on the bucket.
Simple form — allow a single origin with default methods (GET, PUT, DELETE) and all headers:
constructs:
storage:
type: storage
cors: "${construct:website.url}"Use cors: "*" to allow all origins.
Full form — define complete CORS rules (property names can be camelCase or PascalCase):
constructs:
storage:
type: storage
cors:
- allowedOrigins:
- "${construct:website.url}"
allowedMethods:
- PUT
allowedHeaders:
- "*"The full form maps directly to CloudFormation CorsRules.
By default, the bucket is completely private. To expose a subset of the bucket publicly (for example user-uploaded avatars served on a website), use publicPath:
constructs:
avatars:
type: storage
publicPath: publicEvery object stored under the public/ prefix immediately becomes publicly readable over HTTPS (via a bucket policy), while the rest of the bucket stays private. Files are served directly from S3 at https://<bucket>.s3.<region>.amazonaws.com/public/....
Unlike the assets of the server-side-website construct, a storage bucket is never emptied or cleared on deploy, which makes publicPath the right place for files generated or uploaded by the application at runtime.
To make the entire bucket public instead of a single prefix, set publicPath to / or *:
constructs:
avatars:
type: storage
publicPath: "*" # every object in the bucket is publicly readable/ or *, all files in the bucket are publicly readable — only do this for buckets that store exclusively public files. Prefer a sub-path (e.g. public) when the bucket also holds private files.
You can specify an extensions property on the storage construct to extend the underlying CloudFormation resources. In the exemple below, the S3 Bucket CloudFormation resource generated by the avatars storage construct will be extended with the new AccessControl: PublicRead CloudFormation property.
constructs:
avatars:
type: storage
extensions:
bucket:
Properties:
AccessControl: PublicRead| Extension key | CloudFormation resource | CloudFormation documentation |
|---|---|---|
| bucket | AWS::S3::Bucket | Link |
Feel like a common extension pattern should be implemented as part of the construct configuration? Open a GitHub issue.