lambda-batch-trigger

An AWS lambda containing utility functions for AWS batch

Development

Contributions

Contributions are welcome.

Setup Development Environment

Install the following applications:

• AWS CLI
• AWS SAM CLI
• pre-commit
• pipenv

Install Requirements

Run pipenv install --dev to install both production and development requirements, and pipenv shell to activate the virtual environment. For more information see the pipenv docs.

After activating the virtual environment, run pre-commit install to install the pre-commit git hook.

Update Requirements

First, make any needed updates to the base requirements in Pipfile, then use pipenv to regenerate both Pipfile.lock and requirements.txt.

$ pipenv update --dev

We use pipenv to control versions in testing, but sam relies on requirements.txt directly for building the lambda artifact, so we dynamically generate requirements.txt from Pipfile.lock before building the artifact. The file must be created in the CodeUri directory specified in template.yaml.

$ pipenv requirements > requirements.txt

Additionally, pre-commit manages its own requirements.

$ pre-commit autoupdate

Create a local build

Use a Lambda-like docker container to build the Lambda artifact

$ sam build --use-container

Run unit tests

Tests are defined in the tests folder in this project, and dependencies are managed with pipenv. Install the development dependencies and run the tests using coverage.

$ pipenv run coverage run -m pytest tests/ -svv

Automated testing will upload coverage results to Coveralls.

Run integration tests

Running integration tests requires docker

$ sam local invoke HelloWorldFunction --event events/event.json

Deployment

Deploy Lambda to S3

Deployments are sent to the Sage cloudformation repository which requires permissions to upload to Sage bootstrap-awss3cloudformationbucket-19qromfd235z9 and essentials-awss3lambdaartifactsbucket-x29ftznj6pqw buckets.

sam package --template-file .aws-sam/build/template.yaml \
  --s3-bucket essentials-awss3lambdaartifactsbucket-x29ftznj6pqw \
  --output-template-file .aws-sam/build/lambda-batch-trigger.yaml

aws s3 cp .aws-sam/build/lambda-batch-trigger.yaml s3://bootstrap-awss3cloudformationbucket-19qromfd235z9/lambda-batch-trigger/master/

Publish Lambda

Private access

Publishing the lambda makes it available in your AWS account. It will be accessible in the serverless application repository.

sam publish --template .aws-sam/build/lambda-batch-trigger.yaml

Public access

Making the lambda publicly accessible makes it available in the global AWS serverless application repository

aws serverlessrepo put-application-policy \
  --application-id <lambda ARN> \
  --statements Principals=*,Actions=Deploy

Install Lambda into AWS

Parameters

The list of inputs for this lambda can be found in the Parameters section of the template.yaml file.

CFN Nested Stack

We recommend installing this lambda as a nested cloudformation stack.

Add this snippet to the parent cloudformation stack:

LambdaBatchTrigger:
  Type: 'AWS::CloudFormation::Stack'
  Properties:
    TemplateURL: 'https://bootstrap-awss3cloudformationbucket-19qromfd235z9.s3.amazonaws.com/lambda-batch-trigger/master/lambda-batch-trigger.yaml'
    Parameters:
      JobName: !Sub '${AWS::StackName}-my-job'
      JobQueue: !Ref JobQueue
      JobDefinition: !Ref JobDefition

Then deploy the parent stack.

Sceptre

Alternatively it can be deployed with sceptre by creating the following file config/prod/lambda-batch-trigger.yaml

template:
  type: http
  url: https://bootstrap-awss3cloudformationbucket-19qromfd235z9.s3.amazonaws.com/lambda-batch-trigger/master/lambda-batch-trigger.yaml
stack_name: "lambda-batch-trigger"
stack_tags:
  Department: "Platform"
  Project: "Infrastructure"
  OwnerEmail: "[email protected]"
parameters:
  JobName: my-batch-job
  JobQueue: !stack_output_external script-runner::JobQueueArn
  JobDefinition: !stack_output_external script-runner::JobDefinitionArn

Install the lambda using sceptre:

sceptre --var "profile=my-profile" --var "region=us-east-1" launch prod/lambda-batch-trigger.yaml

AWS Console

Steps to deploy from AWS console.

1. Login to AWS
2. Access the serverless application repository -> Available Applications
3. Select application to install
4. Enter Application settings
5. Click Deploy

Triggering

Once this lambda is deployed as a cloudformation stack, locate the SubmitJobApi output of the stack and make a request to it with the curl command. It will look something like this

curl https://jmu6d85zp3.execute-api.us-east-1.amazonaws.com/Prod/submitjob

That should trigger the batch job.

Debugging

This lambda writes debug messages out to a cloudwatch log group, it should be under /aws/lambda/... Locate it and view the messages.

Releasing

We have setup our CI to automate a releases. To kick off the process just create a tag (i.e 0.0.1) and push to the repo. The tag must be the same number as the current version in template.yaml. Our CI will do the work of deploying and publishing the lambda.