Deploy to Railway

Railway is a “deploy from git” PaaS with managed databases as add-ons. The deploy flow is: connect a GitHub repo, Railway detects Node, picks up npm start, builds, deploys. Mongo comes as a one-click add-on next to the service.

Quick reference


Cost (small app)	~$5/mo (app) + Mongo add-on
Managed Mongo?	Yes (Railway add-on) — or external Atlas
Cold start?	No
TLS	Automatic on `*.up.railway.app`; bring-your-own for custom domain
PR previews	Yes (per-environment)

Prerequisites

A Railway account.
The dAvePi project pushed to GitHub.

Deploy

Create a new project, point it at your GitHub repo, give Railway access. It auto-detects Node and proposes npm install && npm start — accept the defaults.
Add a MongoDB add-on. New service → Database → MongoDB. Railway provisions Mongo and exposes its connection string via the MONGO_URL reference variable.
Wire env vars on the app service. Go to the service’s Variables tab and add:
```
NODE_ENV=production
MONGO_URI=${{MongoDB.MONGO_URL}}
TOKEN_KEY=<long random string>
API_PORT=${{PORT}}
CORS_ORIGINS=https://your-frontend.example.com
```
${{MongoDB.MONGO_URL}} and ${{PORT}} are Railway variable references that resolve at runtime. Railway sets PORT itself; dAvePi’s index.js honours it.
Deploy. Railway builds on push to your selected branch. First deploy takes ~3-5 minutes. The dashboard shows the public URL once app.listen fires.

Verify.

curl https://your-app.up.railway.app/healthz

Custom domain + TLS

In the service → Settings → Networking → Custom Domain. Add the hostname, Railway provides the CNAME target. After the DNS record propagates and the TLS cert provisions (Railway uses Let’s Encrypt), the domain serves the app.

Backups

The Mongo add-on doesn’t ship with automated backups on the hobby tier. Two options:

Move Mongo to Atlas. Free tier includes continuous backup. Swap the MONGO_URI env var; remove the Railway Mongo add-on after migrating data with mongodump | mongorestore.
Scheduled mongodump job. Railway supports cron services — add a tiny container that runs mongodump and writes to S3 on a schedule.

See Backup & retention.

Scaling

Vertical: Railway exposes a per-service “Resources” slider for CPU and memory. Easy lever for moderate growth.
Horizontal: Replicas are available on the Pro plan; the service must be stateless (dAvePi is). Mongo doesn’t scale this way — promote to a larger DB tier or move to Atlas.
PR previews: enable in service settings. Each PR gets its own Railway environment with a temporary Mongo. Cleans up on PR close.

Observability

Logs: Railway captures stdout; UI shows the live log. For long-term, ship to an external collector via Railway’s Logflare integration.
Metrics: enable METRICS_ENABLED=true, but the /_metrics endpoint isn’t reachable from outside Railway by default. Pair with METRICS_TOKEN and a Grafana Cloud / Prometheus scrape job that hits the public URL.