cdk-lambda-bash

Deploy Bash Lambda Functions with AWS CDK

Why

AWS Lambda has the docker container image support since AWS re:Invent 2020 which allows you to run your Lambda code in a custom container image. Inspired by nikovirtala/cdk-eks-experiment, cdk-lambda-bash allows you to specify a local shell script and bundle it up as a custom resource in your cdk stack. On cdk deployment, your shell script will be executed in a Lambda container environment.

BashExecFunction

At this moment, we are offering BashExecFunction construct class which is a high-level abstraction of lambda.Function. By defining the script property which poins to your local shell script, on cdk deploy, this script will be bundled into a custom docker image and published as a lambda.DockerImageFunction.

If you fn.run(), a custom resource will be created and the lambda.DockerImageFunction will be executed on deployment.

Install

Use the npm dist tag to opt in CDKv1 or CDKv2:

// for CDKv2
npm install cdk-lambda-bash
or
npm install cdk-lambda-bash@latest

// for CDKv1
npm install cdk-lambda-bash@cdkv1 

Sample

const app = new cdk.App();

const stack = new cdk.Stack(app, 'my-stack');

// bundle your Lambda function to execute the local demo.sh in container
const fn = new BashExecFunction(stack, 'Demo', {
  script: path.join(__dirname, '../demo.sh'),
})

// run it as custom resource on deployment
fn.run();

Re-execution on assets update

By default, if you update your shell script or Dockerfile and re-deploy your CDK application, the BashExecFunction will not be re-executed. Use runOnUpdate to enable the re-execution on update.

fn.run({ runOnUpdate: true });

Custom Dockerfile

In some cases, you may customize your own Dockerfile, for instances:

1. You need extra tools or utilities such as kubectl or helm
2. You need build from your own base image

In these cases, create a custom Dockerfile as below and add extra utilities i.e. kubectl:

FROM public.ecr.aws/lambda/provided:al2

RUN yum install -y unzip jq

# install aws-cli v2
RUN curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip" && \
  unzip awscliv2.zip && \
  ./aws/install

# install kubectl
RUN curl -o kubectl https://amazon-eks.s3.us-west-2.amazonaws.com/1.19.6/2021-01-05/bin/linux/amd64/kubectl && \
  chmod +x kubectl && \
  mv kubectl /usr/local/bin/kubectl

COPY bootstrap /var/runtime/bootstrap
COPY function.sh /var/task/function.sh
COPY main.sh /var/task/main.sh
RUN chmod +x /var/runtime/bootstrap /var/task/function.sh /var/task/main.sh

WORKDIR /var/task
CMD [ "function.sh.handler" ]

Specify your own Dockerfile with the dockerfile property.

new BashExecFunction(stack, 'Demo', {
  script: path.join(__dirname, '../demo.sh'),
  dockerfile: path.join(__dirname, '../Dockerfile'),
});

Conditional Execution

In the user script(e.g. demo.sh), you are allowed to determine the event type and act accordingly.

For example

const installArgoCD = new BashExecFunction(...)

installArgoCD.run({runOnUpdate: true});

When you run this sample, demo.sh will receive onCreate event and you can run your custom logic to "install ArgoCD" like kubectl apply -f URL. However, if you comment it off and deploy again:

const installArgoCD = new BashExecFunction(...)

//installArgoCD.run({runOnUpdate: true});

Under the hood, demo.sh will receive onDelete event and you can run your custom logic to "uninstall ArgoCD" like kubectl delete -f URL.

Check the full sample code below:

#!/bin/bash

# implement your business logic below
function onCreate() {
  echo "running kubectl apply -f ..."
}

function onUpdate() { 
  echo "do nothing on update"
}

function onDelete() { 
  echo "running kubectl delete -f ..."
}

function getRequestType() {
  echo $1 | jq -r .RequestType
}

function conditionalExec() {
  requestType=$(getRequestType $EVENT_DATA)

  # determine the original request type
  case $requestType in
    'Create') onCreate $1 ;;
    'Update') onUpdate $1 ;;
    'Delete') onDelete $1 ;;
  esac
}

echo "Hello cdk lambda bash!!"

conditionalExec

exit 0

In Action

See this tweet