Amazon now provides images for Lambda runtimes on Docker Hub. See amazon/aws-sam-cli-emulation-image-python. It is important to use this specific set of images to avoid missing binaries, which can occur with similar Python 3.x containers from amazon/.

An easier way to build AWS Lambda deployment packages — with Docker instead of EC2

• Logging in to EC2 and creating a deployment package by hand is clumsy
• Instead, script package creation around the amazonlinux image (blessed as an official repository and linked from this AWS user guide)

• build - introduced to build wheels per #2. In theory this should not affect how the environment functions, but in practice use image id 823205fae3ed if you want to be as close to the official lambda environment as possible. On 9/1/21 the build tag has been updated to support the Python3.8 runtime (only).

docker pull quiltdata/lambda

docker run --rm -v $(pwd)/create_table:/io -t \
	-e GIT_REPO quiltdata/lambda \
	bash /io/package.sh

• Mount /io as a docker volume
  • /io should contain package.sh and your lambda code \
  • /io is where the deployment package, lambda.zip, is written \
• Pass environment variables with -e
• --rm so that, for example, secure envs aren't written to disk

Modify package.sh to suit your own purposes.

docker build -t quiltdata/lambda .

Use a personal access token:

git clone https://${TOKEN}@github.com/USER/REPO