publine

Publine is a proof-of-concept for -

1. DynamoDB events getting sent over the DynamoDB stream on TTL expiry
2. an AWS Lambda ingesting events from a DynamoDB stream
3. a second AWS Lambda that ensures that items are deleted on expiry and placed on Dynamodb stream

There are two major steps to deploy - building the lambda and deploying the lambda alongwith DynamoDB and related infrastructure

Build lambda (runner)

This is the lambda (named runner) that ingests events from DynamoDB streams. There is a Makefile in the folder of runner that has the instruction to build the lambda

1. make runner

Deploy infrastructure

Prerequisites:

• Setup your AWS access key and secret in ~/.aws/credentials
• In your S3 account, create a bucket called publinestate. This name is currently hardcoded in infrastructure/main.tf

Run the following -

1. export AWS_PROFILE=name_of_your_profile_in_aws_credentials
2. terraform init
3. terraform apply

Note: A deploy of a lambda requires both the major steps listed above - building the lambda and then redoing terraform apply

Branch descriptions

These branches are in increasing order of development. After the first branch, the others are pretty incremental in change

• 1.0: Basic infrastructure setup and lambda deploy
• 1.1: Update lambda function structure
• 1.2: Attribute mapping from DynamoDB streamed event into Lambda
• 1.3: Filter criteria on events before they are sent to the Lambda
• 2.0: Lambda triggered as a cronjob by Cloudwatch Alarm
• 2.1: Lambda named cleaner ensures that items are deleted on expiry and placed on Dynamodb stream