This PR aims to remove kes and use serverless framework for AWS stack management.
Important
We choose to split chip-n-scale pipeline in two separate AWS Stack to ease iteration and deployement:
Inference Backend (ECS + RDS + VPC)
Lambda prediction
The idea is to create a unique Inference backend that could be use by multiple Prediction Lambda functions.
To Do
[ ] Fix Inference deployement
The inference stack deployement is failing with
the target group with targetGroupArn arn:aws:elasticloadbalancing:us-east-1:....:targetgroup/infer-Targe-1DDILH8D0FGF5/95c344ab813dab6d does not have an associated load balancer. (Service: AmazonECS; Status Code: 400; Error Code: InvalidParameterException; Request ID: 4d5be02d-44bf-4f5a-880e-bf5121506d57)
[ ] Make RDS Table Name configurable in the lambda
[ ] Test Lambda deployement
[ ] Test connectivity between the lambda and the Inference backend
This PR aims to remove
kes
and useserverless
framework for AWS stack management.Important
We choose to split chip-n-scale pipeline in two separate AWS Stack to ease iteration and deployement:
The idea is to create a unique Inference backend that could be use by multiple Prediction Lambda functions.
To Do
cc @drewbo