railway

name	railway
title	Railway
category	Hosting & Deploy
description	Use to deploy services and databases from a GitHub repo, the CLI, a Docker image, or a template, with managed infrastructure.
tags	["hosting","deploy","paas","databases","docker","cli"]
official_docs	https://docs.railway.com
sources	["https://docs.railway.com/quick-start"]
last_verified	2026-08-10T00:00:00.000Z

Railway — Skillship

A PaaS that provisions infrastructure for you: deploy an app plus its Postgres/Redis from a repo, CLI, Docker image, or a marketplace template — all managed on a visual project canvas.

🧭 When to use this skill

• Use when: you want an always-on service (API/worker) and a database with minimal setup.
• Use when: you need a quick full-stack environment with managed Postgres/Redis.
• Don't use for: pure static sites (use Vercel/Netlify) or edge functions (use Cloudflare).

⚡ Quickstart

Option A — Deploy from GitHub

1. Dashboard → New Project → Deploy from GitHub repo.
2. Pick the repo → Deploy Now (or Add Variables first).
3. In service settings, Generate Domain to get a public URL.

Option B — Deploy from the CLI

npm i -g @railway/cli
railway login
railway init          # create a new project from the current directory
railway up            # build + deploy your local code
railway open          # open the project canvas

Add a database

• On the canvas: New → Database → Postgres/Redis/MySQL/Mongo. Railway injects connection variables (e.g. DATABASE_URL) that your service can reference.

🧩 Common recipes

Recipe: Reference service/database variables

# Railway exposes managed vars like DATABASE_URL, REDIS_URL to linked services automatically.
# Reference another service's var in the dashboard with ${{ Postgres.DATABASE_URL }}

Recipe: Deploy a prebuilt Docker image

• New Project → Empty project → Add a Service → Docker Image → enter e.g. org/image:tag (use the full registry URL for GHCR/GitLab/Quay). Private registries require a paid plan.

Recipe: Run a local command against Railway env

railway run npm run migrate    # runs locally with Railway's env vars injected
railway logs                   # tail build/deploy logs

🚀 Ship to production

• Secrets set as service Variables (Railway UI/CLI), not committed to the repo.
• Public domain generated (or custom domain added) for services that need ingress.
• App reads config from env vars (DATABASE_URL, PORT) — bind to process.env.PORT.
• Separate environments (e.g. production vs PR/testing) for isolation.
• Health/observability: check the built-in metrics/logs; scroll full build logs when diagnosing.
• Understand usage-based billing before scaling instances up.

🔐 Security & secrets

• Store credentials as service Variables; reference cross-service secrets via ${{ Service.VAR }} instead of copy-pasting.
• Don't expose database services publicly unless required; prefer Railway's private networking.

🐛 Common errors & fixes

Symptom	Likely cause	Fix
App deploys but URL 404/none	No public domain generated	Service settings → Generate Domain
App crashes binding to a port	Hardcoded port	Use process.env.PORT
Can't reach the database	Missing/!referenced var	Use the injected DATABASE_URL / ${{ Postgres.DATABASE_URL }}
Build failed, unclear why	Error mid-log	Scroll the entire build log; the real error is rarely last
Private image won't pull	Free plan / missing creds	Upgrade plan; add registry credentials

💰 Pricing gotchas

• Billing is usage-based (CPU/RAM/egress). Idle-but-running services still accrue cost — scale down unused ones.

📚 Sources

• https://docs.railway.com/quick-start