lookout

A small uptime monitor. You add some URLs, it checks them on a schedule, and the dashboard shows what's up, what's down, and how fast each one responds.

Live demo

https://lookout-dusky.vercel.app

Log in with demo@lookout.dev / demo1234 to see 20 real URLs with a week of history. admin@lookout.dev / admin1234 also gets the metrics page. Or sign up with your own email and hit Try the sample.

The api is on Render's free tier, so if nobody has used it in a while the first load can be slow while it wakes up.

Stack

• Backend: Node 20, Express, Prisma + Postgres, BullMQ + Redis, pino for logs
• Frontend: React 18, Vite, Tailwind, recharts
• Hosting: Vercel for the frontend, Render for the api and Redis, Supabase for Postgres

Run it locally

You need Docker.

cp .env.example .env
# put a JWT_SECRET of 32+ chars in .env, e.g. from: openssl rand -base64 32
docker compose up --build -d

Open http://localhost:8080, sign up, and click Try the sample. You should see three URLs show up: two UP and one DOWN with a 404.

If you want the demo data too:

docker compose exec api node dist/scripts/seed.js   # add --force to redo it

That creates the same demo and admin logins as the live site. To make your own account an admin, put your email in ADMIN_EMAILS in .env before starting.

Running the app outside Docker

Start just Postgres and Redis in Docker, then run each part in its own terminal:

docker compose -f docker-compose.yml -f docker-compose.dev.yml up -d postgres redis

cd backend  && cp .env.example .env && npm install && npm run db:migrate && npm run dev
cd backend  && npm run worker
cd frontend && npm install && npm run dev    # http://localhost:5173

Tests: cd backend && npm test

How it works

• The api handles auth and CRUD, and puts check jobs on a Redis queue.
• A separate worker process picks those jobs up and actually hits the URLs, so a slow site never slows down the api.
• Each URL gets its own BullMQ scheduler that fires at its interval. When the worker starts it syncs the schedulers with Postgres, so if Redis gets wiped it just rebuilds them.
• A check is UP if the final response (after redirects) is a 2xx. Anything else is DOWN, saved with an error type like HTTP_ERROR, TIMEOUT, DNS_FAIL, TLS_ERROR or CONN_REFUSED.
• Raw checks are kept for 7 days. They also get rolled up into hourly stats, which are kept forever and power the 7d and 30d charts.
• Every request gets an x-request-id that's passed along to the worker, so you can follow one check through the logs of both processes.

Deploying

The api deploys to Render from render.yaml. Render has no free background workers, so there the worker runs inside the api process (EMBED_WORKER=true). start-hosted.sh runs the migrations and the seed before starting the api.

For DATABASE_URL, use Supabase's session pooler URL with ?connection_limit=5 on the end. The direct host is IPv6 only so Render can't reach it, and the transaction pooler breaks prisma migrate deploy.

The frontend is on Vercel with frontend as the root directory. frontend/vercel.json forwards /api/* to the Render api, so the browser only ever talks to the Vercel domain.

API

Everything is under /api and uses Authorization: Bearer <token>, except signup, login and /health.

Method	Path	Who	What
POST	/api/auth/signup	anyone	Create an account, returns user and token
POST	/api/auth/login	anyone	Log in, returns a token
GET	/api/auth/me	user	Current user
GET	/api/urls	user	Your URLs with their latest check (page, pageSize, status, all)
POST	/api/urls	user	Add a URL (url, optional intervalSeconds)
POST	/api/urls/bulk	user	Add up to 100 URLs at once
GET	/api/urls/:id	user	One URL
PATCH	/api/urls/:id	user	Change intervalSeconds or isActive
DELETE	/api/urls/:id	user	Delete a URL and its history
GET	/api/urls/:id/checks	user	Check history, newest first (limit, max 500)
GET	/api/urls/:id/stats	user	Uptime and response times (window = 24h, 7d, 30d)
POST	/api/urls/:id/check	user	Run a check right now
POST	/api/urls/check-all	user	Queue a check for every active URL (all for admins)
GET	/api/dashboard/summary	user	Up/down counts and 24h uptime
GET	/api/metrics	admin	Check, queue and process metrics
GET	/health	anyone	Says whether the db and Redis are reachable

More screenshots: bulk import, sample, admin metrics.

License

MIT