System Playground

You can try this up and running production system! (No longer available)

• First create a user_id by making a POST request to:
  • https://YOUR-URL.amazonaws.com/user
• Then create a ticket by making another POST request to:
  • Add this body- {"user_id": "YOUR-USER-ID"}
  • https://YOUR-URL.amazonaws.com/ticket
• Now you can check on your ticket status by making this GET request:
  • https://YOUR-URL.amazonaws.com/ticket?ticket_id=<YOUR_TICKET_ID>

Setup Instructions

Prerequisites:

• AWS account
• Disclaimer: the cloud_formation.yaml is not production ready, but gives a sense of the design
  • You can paste and view it in the "Application Composer" on AWS console

Step 1: Create a new Aurora PostgreSQL cluster in the AWS Console

• Enable DATA API for your database!
• Use the "Query Editor" on AWS console to connect to the DB and create these tables:

CREATE TABLE IF NOT EXISTS users (
user_id VARCHAR(255) PRIMARY KEY,
time_created VARCHAR(255) NOT NULL);

CREATE TABLE IF NOT EXISTS tickets (
ticket_id VARCHAR(255) PRIMARY KEY, 
user_id VARCHAR(255) REFERENCES users(user_id), 
time_created VARCHAR(255) NOT NULL, 
time_finished VARCHAR(255), 
response TEXT);

Step 2: Create two new Standard Queues in the AWS SQS Console

• one for Responses and one for Errors

Step 3: Create Lambda Functions for each component in the lambdas directory.

• Runtime: Python 3.12
• Attach correct IAM roles that allows access to RDS and SQS for each lambda
• Set the required environment variables for each lambda
• Give the ticket_creator lambda permissions to invoke the ml_process, and the vice-versa
• Configure the ticket_status_updater lambda to be triggered by the "Response" SQS when receiving a message
• Timeout: Increase to 10 seconds for all lambdas (for "ml_process" make it 7 minutes!)

Step 4: Create an API Gateway for an HTTP API

• Create a POST endpoint and link it to the ticket_creator
• Create a GET endpoint and link it to the ticket_getter
• Create a POST endpoint and link it to the user_creator

Step 5: Test the setup

• Make a POST request to the API Gateway endpoint to create "user_id" and save it
• Make another POST request to /ticket endpoint to create a ticket_id
• Try to GET from /ticket endpoint with the ticket_id and see what you get
• Check if the request is stored in the RDS database

Enjoy!