Simple Voting

Pre-requirements

NodeJS v12+
Docker v19+

Setup

Create a .env file at root and copy the content from .env.example:
```
 cp .env.example .env
```
By default, port 27017 and 6379 will be used for our MongoDB and Redis instance, respectively. If these ports are already in use, please change the port number in your .env file and it should be align with docker-compose.yml file.
Launch MongoDB and Redis via docker compose:
```
docker-compose up
```
Install project dependencies:
```
 npm i
```
Launch server:
```
npm start
```
By default, port 8080 will be used by the NodeJS server. If this port is already in use, please pass a new port number through the PORT environment variable when launching, for example:
```
 PORT=8888 npm start
```

Unit Tests

There are few test cases that cover the core services. You may run through this command:

npm run test

API Doc

This server was generated by the swagger-codegen project. Please checkout the server API from the following page (You have to launch the server first):

http://localhost:8080/docs

Postman Collection

There is a postman-collection.json file (v2.1) at the project root which contains examples of how to interact with the HTTP API. You may import it into your postman application. Notice that you should change the IDs recorded in the collection to the actual values found in your runtime, or else almost of the requests will be failure.

The Campaign charts

You may found all the campaign charts by visiting http://localhost:8080/pages/charts, which also establish a websocket connection to the daemon of the program for receiving last campaign vote counts.

How the Voting works

The NodeJS program will launch a web server and a daemon. The web server provides HTTP interface to access campaigns and users data stored in MongoDB. User votes will be stored as bitmaps in Redis for fast write (AOF enabled).

Each time after writing user vote into the bitmap in Redis, it will asynchronously updates the vote counts of the campaign in MongoDB, which also update the index for fast read.

While the asynchronous action completed, a voteUpdate event will be emitted through Redis. The daemon will receives this event and then notify all its clients with the updated campaign data through websocket.

Here is how a voting request being processed internally:

To ensure that the vote counts in MongoDB are eventually consistent with the bitmaps in Redis, even encounter program crashes, the daemon also periodically updates the vote counts from Redis to MongoDB.

Since the program is designed to be horizontally scalable, there is a daemon lock implemented in Redis so that only one daemon can performs the above updates.

High Traffic and Scalable Architecture Design

While the NodeJS program is ease for scale, it would be better to have Cluster + Replica structure for both the Redis and MongoDB:

doggor/offsite-test-voting