n8n Download & Self-Host Guide (Docker, npm, VPS)

n8n download is a misleading query — there is no installer. What people actually need is a running n8n instance. This guide covers all three legitimate paths (Docker, npm, source) and gives you the production hardening checklist nobody else publishes.

Path 1: Docker (recommended)

Docker is the path n8n GmbH officially recommends. It works on any VPS — DigitalOcean, Hetzner, AWS Lightsail, Linode, Contabo. Cost: USD 5–20 per month for most workloads.

Full step-by-step in our n8n Docker guide. Minimal compose:

services:
  n8n:
    image: docker.n8n.io/n8nio/n8n
    restart: always
    ports:
      - "5678:5678"
    environment:
      - N8N_HOST=n8n.yourcompany.com
      - N8N_PROTOCOL=https
      - WEBHOOK_URL=https://n8n.yourcompany.com/
      - GENERIC_TIMEZONE=Asia/Kolkata
      - TZ=Asia/Kolkata
    volumes:
      - n8n_data:/home/node/.n8n

volumes:
  n8n_data:

Path 2: npm install

Quick for local development. Requires Node.js 22.12 or higher and a persistent home directory for the SQLite database.

npm install n8n -g
n8n start

Not recommended for production: you lose the easy upgrade story Docker gives you, and you must manage process supervision (PM2, systemd) yourself.

Path 3: Build from source

Clone the n8n GitHub repo, run pnpm install and pnpm build. Only worth doing if you are contributing back upstream or running a fork. For everyone else, the Docker image is identical to what source builds produce.

Minimum server requirements

Workload	RAM	vCPU	Disk
Personal / hobby	1–2 GB	1	10 GB SSD
Light agency (10 workflows)	2 GB	1–2	20 GB SSD
Heavy agency / AI agents	4–8 GB	2–4	40 GB+ SSD
High concurrency, multiple runners	16 GB+	4+	100 GB+ SSD

HTTPS reverse proxy

In production you must put TLS in front of n8n. Pick one:

• Caddy — easiest. Automatic Let's Encrypt certificates, one-line config.
• Traefik — best for Docker stacks with multiple services.
• Nginx — most familiar to ops engineers.
• Cloudflare Tunnel — no open ports needed; great for home-server setups.

Whichever you choose, ensure X-Forwarded-Proto is set to https and your WEBHOOK_URL env matches your public URL exactly. Misconfigured headers are the #1 cause of the redirect-loop login bug.

Backups

n8n stores everything in ~/.n8n — workflows, credentials (encrypted), and the SQLite database. Back up:

1. The volume mount (e.g. n8n_data in compose).
2. The N8N_ENCRYPTION_KEY environment variable — without it, restored credentials are unusable.
3. Your reverse-proxy config and TLS certificates (or rely on Let's Encrypt's auto-renew).

A nightly cron that rsyncs the volume to object storage (S3, Backblaze B2, Hetzner Storage Box) handles 95% of disaster recovery scenarios.

Upgrades

With Docker the upgrade flow is:

1. docker compose pull
2. docker compose up -d
3. Open the editor — n8n will run any database migrations automatically.

Always read release notes for major version bumps. Test on a staging instance with a snapshot of your prod volume before upgrading critical workflows.

Security hardening

• Set a strong N8N_BASIC_AUTH_USER / N8N_BASIC_AUTH_PASSWORD in addition to the owner account.
• Put Cloudflare in front and restrict the editor to your office IPs.
• Disable the public API in production unless you genuinely need it (N8N_PUBLIC_API_DISABLED=true).
• Rotate N8N_ENCRYPTION_KEY on a known cadence — but keep the old one until you have re-encrypted all credentials.
• Pin your Docker image tag instead of using latest.

Cost estimate

A real-world agency running ~30 active workflows with light AI usage on a 2-vCPU 4 GB VPS:

• VPS (Hetzner CX22 or DO Premium Intel): $5–8/month
• Object storage backups: $1/month
• Cloudflare free tier in front: $0
• Domain: $10–15/year

Versus the equivalent SaaS-only setup on Zapier or Make: typically 5–10× more once you factor in per-task pricing on AI-heavy flows.

After install: import templates

• Lead Gen Systems
• Chat Bot
• Voice Agent
• All free workflows

Related: Docker setup · n8n Cloud comparison · Pricing math