/yjs-scalable-ws-backend

Primary LanguageTypeScriptMIT LicenseMIT

Code from this repo live on erdtool.com!!!

NEW!!

Horizontally scaled websocket backend that can talk to any external api for persistence and provide read-only and read+write access levels:

Branch external_api_persistence

main-branch

About

Example of horizontally scalable websocket backend for y-js to be used with y-websocket provider with persitence to postgresql using knex. Uses redis-pubsub for horizontal scaling and uses redis-queues to provide eventual-consistency and better behaviour when users join a document being edited by other users.

Usage

  1. npm ci
  2. npm run build
  3. cp .env.example .env.local
  4. Fill in .env.local
  5. npm run tables -> creates the tables for doc persistence in db
  6. npm run dev -> starts the websocket-server
  7. On the client-side, initialize a yjs doc and use the y-websocket provider to connect to the websocket-server

How it works

This repo is a slightly reworked websocket server found in the y-websocket repo(link).

The websocket-server that comes with y-websocket essentially maintains a copy of the y-js document(s) in memory and syncs it between different clients connected to the same doc.

The websocket-server in this repo isolates the updates that clients send to it, persists these updates to the database, and publishes these updates (using redis-pubsub) in a channel for the document. Also, when a doc is created for the first time in the websocket-server, the server reads all the updates stored in the database, and applies those updates to the document, effectively initializing it.

This makes the websocket-server provided in this repo persistent and horizontally-scalable on paper.

The code in this repo hasn't been tested in a production system yet, and from the looks of it, will be a long time before I would be able to run it on a production system, but theoretically, the code in this repo should be sufficient and can be tweaked to suit any project.

The code from this repo is live on (https://erdtool.com)

Companion testing repo

repo

Deploy to Kubernetes

For local testing, we first have to spin up minikube and create some namespaces

minikube start
kubectl create ns backend
kubectl create ns postgresql
kubectl create ns redis

Then we have to install PostgreSQL and Redis

helm install psql bitnami/postgresql --version 11.9.11 --namespace postgresql --set global.postgresql.servicePort=5432 --set global.postgresql.postgresqlDatabase=yjs_db \
    --set global.postgresql.postgresqlUsername=test_user --set global.postgresql.postgresqlPassword=mypass

helm install redis bitnami/redis --version 17.3.7 --namespace redis --set auth.enabled=false

After building the Docker image

eval $(minikube docker-env)
docker build -f ./Dockerfile -t kapv89/yjs-backend-ws:test .

we can deploy the service with our helm charts

helm install yjs-backend-ws ./chart/yjs-backend-ws --namespace backend \
 --set image.tag=test \
 --set secret.db_host="psql-postgresql.postgresql.svc.cluster.local:5432" \
 --set secret.db_user="test_user" \
 --set secret.db_password="mypass" \
 --set secret.db_name="yjs_db" \
 --set secret.redis_host="redis-master.redis.svc.cluster.local" \
 --set secret.redis_port="6379" \
 --set secret.redis_prefix="backend.crdtwss."