big-c-note / genetic_algo_api

Genetic Algo Flask AP with "Frontend" API written in Golang. Deployed in Kubernetes via Helm Chart.

Genetic Algorithm API

Introduction and Project Description

The following will be a "tutorial" (read: set of instructions) for how I set up the project in this repository.

The project aims to deploy two containerized API applications into a kubernetes cluster. One API is written in Go, and it is the "consumer-facing" endpoint. The other is a "backend" flask API (python) for manipulating a genetic algorithm.

The consumer can post "things" which have associated "weight" and "value". Once complete, the consumer can request an optimization such that the value is maximized under a certain weight threshold. A use case would be if you are going on a trip, and you can only bring X lbs of stuff, but you want to bring the stuff that makes you happiest (ie; yields maximum "value"). This is the classic knapsack problem.

Resources Used and Special Thanks.

The following code bases were used to create the genetic alorithm and go api.

• https://github.com/kiecodes/genetic-algorithms
• https://github.com/viveksinghggits/library
• https://github.com/kubucation/go-rollercoaster-api

Technologies Used

For me, this project is largely practice in creating containerized applications and deploying them to a kubernetes cluster.

• go rest api
• python flask api
• genetic algorithm written in python
• docker for containerizing the above applications
• kubernetes for deploying the containerized applications (using minikube)
• helm for packaging the deployment to kubernetes

I am on a linux, ubuntu OS (Focal LTS).

Comsumer-facing API MVP Checklist

The following will be written in go

• Post request for single thing which contains "name", "value", and "weight".
• Get request for the list of things that have been posted.
• Get request that, when received, runs a separate get request to the "backend" API to run the genetic algorithm and returns the results.

Backend API MVP Checklist

• powers a get request to the consumer-facing api to receive data, then calls the genetic algorithm and sends a post request to the consumer-facing api with the results.

Kubernetes Deployment

• Create docker images.
• Create helm chart.

Deploy Locally

If you're just looking to run the project locally, without kubernetes, look at consumer_api/README.md and backend_api/README.md.

Deploy in Kubernetes

If you want to deploy in kubernetes.

Start by dockerizing the applications in consumer_api and backend_api directories.

There is documentation of this process in each directory's README.

Now start up minikube (installation instructions below).

minikube start

You can either apply the supplied yaml configs, or install the helm chart.

Apply the supplied YAML files.

The manifest files are in the manifests/consumer_api directory. After starting minikube, they are applied as such.

Important note: I did not leave up my docker images. If you want to run this project, you must dockerize the applications and push to your own repo. You'll need to change out the "image" references in manifests/consumer_api/consumerapi.yaml to your image repositories.

cd path/to/top/directory
kubectl apply -f manifests/consumer_api/ns.yaml
kubectl apply -f manifests/consumer_api/consumerapi.yaml
kubectl apply -f manifests/consumer_api/service.yaml

Finally you can forward your localhost:8080 to the consumer api service to use.

kubectl port-forward -n consumer-api svc/consumerapi-service 8080

(see play with the api below)

Run the Helm Chart Installation.

I also provided a helm chart. Install with

helm install genalgo helm

Finally you can forward your localhost:8080 to the consumer api service to use.

kubectl port-forward -n consumer-api svc/consumerapi-service 8080

(see play with the api below)

Finally, play with the api!

I left a set up curl commands in consumer_api/README.md that will show you how to interact with the api.

Feel free to mess around with the hardcoded parameters set in backend_api/backend_api/gen_algo.py.

Tear Down

Just in case you want to clean up after you install the project.

minikube stop
minikube delete
docker image ls 
# find an image you want to delete
docker image rm -f {image id}
# you might additionally want to delete the docker repositories you made

Instructions for installing the software needed

I installed the following for Ubuntu, but there may be instructions included for another OS in the below links.

Docker.

• Installation instructions
  • I used the install from repo instructions.
• Quickstart
  • Remember to use sudo in front of docker command and that when you login/push
  • the username must be lowercase. This may be linux only. Not sure.

Kubectl.

• Installation instructions
  • I used the curl instructions.

MiniKube.

• Installation instructions

Troublshooting,

--issue was I needed to do the post install for Docker: https://docs.docker.com/engine/install/linux-postinstall/

Install Helm

• Installation instructions
  • used the from script instructions.