Create an AWS Lambda powered API endpoint using a custom build Ubuntu base image for ECR container.
Install serverless framework via NPM or as a standalone binary.
$ sls --version
Framework Core: 3.31.0
Plugin: 6.2.3
SDK: 4.3.2
You should also have permissions to deploy to AWS Lambda, ECR, Cloudwatch, S3, and create IAM Lambda execution role.
Create an environment file .env
with AWS profile infomation.
The sample .env
file below also has environment vars necessary for currect serverless.yml.
AWS_PROFILE="my-profile-name"
AWS_REGION="us-east-1"
ENVIRON_VARIABLE_1="HelloWorld"
TAG_MANAGED_BY="[email protected]"
TAG_COST_CENTER="12345-05432"
Add AWS credentials to ~/.aws/credentials
To build container from dockerfile and deploy to ecr for lambda
sls deploy
To remove everything created by serverless
sls remove
Make sure to save the .serverless
directory as it contains JSON statefiles which are helpful for future updates.
The Dockerfile will copy over everything in the app
directory and install python requirements from requirements.txt
# Copy function code
COPY ./app/ ${LAMBDA_TASK_ROOT}/
# Install function dependencies
COPY requirements.txt /
RUN pip install \
-r "/requirements.txt" \
--target "${LAMBDA_TASK_ROOT}"
You can add additional dependencies and build up your code within the app
directory.
Lambda supports container images upto 10GB in size.
Best of luck!
https://aws.amazon.com/blogs/aws/new-for-aws-lambda-container-image-support/
https://docs.aws.amazon.com/lambda/latest/dg/images-create.html