Traefik reverse proxy: one entrance for every service
When you need Traefik, what gets delivered, what is out of scope, and how to move over from an existing Nginx setup.
When you need it
Three services become twelve, ports stop being memorable, and nobody knows when the certificate expires. Adding one service means editing several files and reloading. Traefik pulls this into one entrance: route by hostname or path, renew certificates automatically.
What gets delivered
- A route table: hostname or path to backend service, and which ones need WebSocket.
- Static and dynamic configuration kept apart, so adding a service touches one place.
- Certificate setup: ACME HTTP-01 or DNS-01, the latter for home networks without port 80.
- Middleware as needed: HTTPS redirect, compression, rate limits, basic auth.
- A migration order: run alongside first, then switch hostname by hostname, with a way back.
- A debugging list: 502, 404, wrong certificate, dropped WebSocket; what to check first for each.
Not included
- The application's own login and permissions. Traefik handles the entrance, not who may use what.
- Domain registration, DNS hosting or Cloudflare accounts.
- Changing application code to fit a route.
- 24/7 on-call. After handover, you add services yourself.
Who it fits
You already run several services, or plan to put a NAS, model service and monitoring behind one entrance. Nginx or Caddy still works, but the config has started to sprawl.
How to start
Email what services you run, how domains and certificates are managed, and whether there is a public IP. Scope first, then a quote.
Questions
Can I migrate from Nginx?
Yes. Run both for a while, switch one hostname at a time, and keep the old config for rollback.
Will there be downtime?
Switching a hostname usually takes seconds. DNS changes are announced in advance.
Related
Not sure where to start?
Run the assessment first, or send your setup and goal by email. The reply says whether it fits and what happens next.