Using the Serverless CLI tool, you can package your project without deploying it to AWS. This is best used with CI / CD workflows to ensure consistent deployable artifacts.
Running the following command will build and save all of the deployment artifacts in the service's .serverless directory:
However, you can also use the --package option to add a destination path and Serverless will store your deployment artifacts there (../my-artifacts in the following case):
serverless package --package my-artifacts
Sometimes you might like to have more control over your function artifacts and how they are packaged.
You can use the
patterns configuration for more control over the packaging process.
Patterns allows you to define globs that will be excluded / included from the resulting artifact. If you wish to exclude files you can use a glob pattern prefixed with
! such as
Serverless will run the glob patterns in order so you can always re-include previously excluded files and directories.
By default, serverless will exclude the following patterns:
and the serverless configuration file being used (i.e.
serverless.yml). In addition, if
useDotenv is set, all files satisfying pattern
.env* will be excluded as well.
Exclude all node_modules but then re-include a specific modules (in this case node-fetch) using
package: patterns: - '!node_modules/**' - 'node_modules/node-fetch/**'
Exclude all files but
package: patterns: - '!src/**' - src/function/handler.js
Note: Don't forget to use the correct glob syntax if you want to exclude directories
package: patterns: - '!tmp/**' - '!.git/**'
For complete control over the packaging process you can specify your own artifact zip file.
Serverless won't zip your service if this is configured and therefore
patterns will be ignored. Either you use artifact or patterns.
The artifact option is especially useful in case your development environment allows you to generate a deployable artifact like Maven does for Java.
service: my-service package: artifact: path/to/my-artifact.zip
You can also use this to package functions individually:
service: my-service package: individually: true functions: hello: handler: com.serverless.Handler package: artifact: hello.jar events: - http: path: hello method: get
Artifacts can also be fetched from a remote S3 bucket. In this case you just need to provide the S3 object URL as the artifact value. This applies to both, service-wide and function-level artifact setups.
service: my-service package: artifact: https://s3.amazonaws.com/some-bucket/service-artifact.zip
service: my-service package: individually: true functions: hello: handler: com.serverless.Handler package: artifact: https://s3.amazonaws.com/some-bucket/function-artifact.zip
If you want even more controls over your functions for deployment you can configure them to be packaged independently. This allows you more control for optimizing your deployment. To enable individual packaging set
individually to true in the service or function wide packaging settings.
Then for every function you can use the same
artifact config options as you can service wide. The
patterns option will be merged with the service wide options to create one
patterns config per function during packaging.
service: my-service package: individually: true patterns: - '!excluded-by-default.json' functions: hello: handler: handler.hello package: # We're including this file so it will be in the final package of this function only patterns: - excluded-by-default.json world: handler: handler.hello package: patterns: - '!some-file.js'
You can also select which functions to be packaged separately, and have the rest use the service package by setting the
individually flag at the function level:
service: my-service functions: hello: handler: handler.hello world: handler: handler.hello package: individually: true
Serverless will auto-detect and exclude development dependencies based on the runtime your service is using.
This ensures that only the production relevant packages and modules are included in your zip file. Doing this drastically reduces the overall size of the deployment package which will be uploaded to the cloud provider.
You can opt-out of automatic dev dependency exclusion by setting the
excludeDevDependencies package config to
package: excludeDevDependencies: false