junqueira / sam-pyspark

Serverless PySpark

SAM PySpark

Serverless PySpark

Requirements

• SAM CLI - Install the SAM CLI
• Python installed
• Docker - Install Docker community edition

Environment

Create you Python environment for development

virtualenv -p python2.7 .env
source .env/bin/activate
pip install -U -r requirements.txt
aws s3 cp s3://didone-spark/drivers.zip libs/
aws s3 cp s3://didone-spark/java.zip libs/
aws s3 cp s3://didone-spark/python.zip libs/

Build

rm -Rf .aws-sam
# Create bucket to deploy
DEPLOY_BUCKET=didone-spark-sam
aws s3 mb "s3://${DEPLOY_BUCKET}"
# Build lambdas
sam build
# Change the runtime python2.7 to 'provided'
sed 's+ python2.7+ provided+g' template.yaml > template.tmp
# Package and generate the template for cloudformation
sam package --template template.tmp --output-template-file packaged.yaml --s3-bucket $DEPLOY_BUCKET

Deploy

sam deploy --guided

or you can use the AWS Cloudformation to manage your deployments as Application Stacks using the packaged.yaml file

# Create stack
aws cloudformation create-stack --stack-name pyspark-app --template-body file://packaged.yaml --capabilities "CAPABILITY_IAM" "CAPABILITY_NAMED_IAM" "CAPABILITY_AUTO_EXPAND"
# View stack
aws cloudformation describe-stacks --stack-name pyspark-app --output table

# Remove stack
aws cloudformation delete-stack --stack-name pyspark-app