kharsh77 / personal_finance

Personal Finance App

Introduction

• A Python based personl finance app which helps to track income/expense transactions.
• The app supports multi-tenancy. A client can have isolation on how compute and storage happens on the app.

Walkthrough Video: https://www.loom.com/share/fd744f0fa86b4c189a0fcec7885bfb2a?sid=c4d374e7-547e-48fe-8d5a-cd2a2383210a

Design Considerations

• The app uses flask framework to expose few apis and postgres for data storage.
• Kubernates is used to keep seggregation of compute and storage based on namespace.
• Different clients will have a separate base URL which is generated per namespace wise. This unique URL for each client would be used by their users.
• Compute and storage isolation would be maintained for each these URLs.
• The APIs are authenticated by JWT token.

Technologies Used:

• Flask, Postgres, docker, Kubernates and github actions(CI/CD). AWS ECR for deploying image to cloud. AWS EKS for using cloud cluster (Few configurations established).
• Kubernates deployment has been tested locally using minukube.

APIs Exposed

• /register: This api registers a user into the app based on a username and password.
• /login: After registration, user can login using this API and get a access token for using other transaction APIs.
• /transaction: A user can add their transaction amount with details. A income or expense type of input is expected.
• /analytics: Based on the transactions added, a user can get current month or last month aggregated data.
• /heath: To check health status of service.

• we need to add access token generated from response of /login api to header "x-access-token" of /transaction & /analytics api.
• Postman Collection: https://www.postman.com/kharsh77/workspace/projects/request/1228171-53cf1988-0b32-4cd7-a520-344adbba4ef3

Configure app based on cluster

• Remote:
• • Below are step to configure EKS.
• • create cloudformation stack on aws
• • update cluster.ymal file & update subnets from above setup of cloudformation stack
• • eksctl create cluster -f cluster.yaml --kubeconfig=~/.kube/config
• Local (Minikube):
• • We need to start minikube and allow local docker repositories to be accessed by minikube.
• • minikube start
• • eval $(minikube docker-env)
• • minikube addons enable registry

Update application docker image

• Remote: (ECR Registry)
• • We can use github action rule: publish_image_to_ecr
• Locally
• • docker-compose build --no-cache

Add a new Tenant

• We have deployment setup for each tenant in kubernates/<tenant-name> folder.

• For adding a new tenant we need create a namespace for the tenant. The sample config file can be found in kubernetes/<tenant-name>/dev-namespace.yaml file.

• To apply a namespace we need to run kubectl apply -f kubernetes/<tenant-name>/dev-namespace.yaml -n <tenant-name>

• After creating a namespace, we can add apply all other configurations for that client by kubectl apply -f kubernetes/<tenant-name>/. -n <tenant-name>

• All other configurations has to setup only once while tenant onboarding in kubernates/ folder

• Remote: (EKS) [Pending]

• • We can use github action rule: deploy_to_eks
• • Due to lack to production level EKS cluster this functionality is not fully functional.

• Locally: (Minikube)

• • Namespace for the tenant has to be active for deploying app for a tenant.
• • kubectl apply -f kubernetes/<tenant-name>/. -n <tenant-name>

• We will wait till all containers are RUNNING by using this command: kubectl get po,svc -n <tenant-name>

• • minikube service app --url -n <tenant-name> This command will output a specific URL which can be used via postman
• • Each tenant will have a separate url which can be accessed by their users.