LambdaMaker.jl

This package is intended to generate all the boilerplate code required to deploy an AWS Lambda function via AWS SAM.

You will need to download and install the following:

• AWS SAM
• Docker

You will need to generate a new project by running,

using LambdaMaker

create_lambda_package("project_name")

If a directory already exists with the project_name, an error is thrown and the directory is left untouched. You can pass force=true as a kwarg to overwrite it.

This will create a new directory in your working directory with the following files:

• bootstrap: This is the file which Lambda will use when it first runs
• Dockerfile: Defines the Docker image that will run in the Lambda environment
• handle_requests.jl: Reads the Lambda invocations, calls your code, and writes the responses
• Project.toml: Basic TOML for defining your package to be deployed on Lambda
• template.yml: The CloudFormation stack which will be created when deploying via AWS SAM
• src/{project_name}.jl: A module with the boilerplate handle_event(event_data, headers) function

You will need to modify the src/{project_name}.jl file to include whatever code you wish to run in the Lambda function. The entrypoint for Lambda will be the handle_event(event_data, headers) function, you will need to modify this to call the functions you wish to invoke.

Note: If you want to specify an ENTRYPOINT script in the Dockerfile (e.g. to set some environment variables), you must explicitly run the bootstrap file in that script.

Dockerfile:

WORKDIR /var/task
ENTRYPOINT [ "script.sh" ]
CMD [ "Module.function" ]

script.sh:

#!/bin/bash
export FOO=1
./bootstrap

AWS SAM documentation can be found here. The short-handed version is,

Build your Docker image as:

sam build

Test locally by running:

sam local invoke JuliaFunction

If your function requires some form of input see the appendix for more information.

When you are satisfied and want to deploy your function,

sam deploy --guided

If you want to update your function, you can use SAM sync.

• tk3369/aws-lambda-container-julia

Passing inputs is usually done in the JSON format. You will need to add a package to parse this information, such as JSON3.jl.

After generating the boilerplate code and adding in your JSON parsing functionality you will need to modify the src/project_name.jl. Modify the handle_event() function as,

module MyPackage

using JSON3

function handle_event(event_data, headers)
    input = JSON3.read(event_data)[:input_variable]
    return your_function(input)
end

end

You will then need to define a JSON file to pass into your function as,

{
    "input_variable": 10
}

To pass this event JSON file when localy testing as,

sam local invoke -e event.json JuliaFunction