Traefik Forward Auth
Now that we have Traefik deployed, automatically exposing SSL access to our Docker Swarm services using LetsEncrypt wildcard certificates, let's pause to consider that we may not want some services exposed directly to the internet...
..Wait, why not? Well, Traefik doesn't provide any form of authentication, it simply secures the transmission of the service between Docker Swarm and the end user. If you were to deploy a service with no native security (Radarr or Sonarr come to mind), then anybody would be able to use it! Even services which may have a layer of authentication might not be safe to expose publically - often open source projects may be maintained by enthusiasts who happily add extra features, but just pay lip service to security, on the basis that "it's the user's problem to secure it in their own network".
Some of the platforms we use on our swarm may have strong, proven security to prevent abuse. Techniques such as rate-limiting (to defeat brute force attacks) or even support 2-factor authentication (tiny-tiny-rss or Wallabag support this).
Other platforms may provide no authentication (Traefik's own web UI for example), or minimal, un-proven UI authentication which may have been added as an afterthought.
Still other platforms may hold such sensitive data (i.e., NextCloud), that we'll feel more secure by putting an additional authentication layer in front of them.
This is the role of Traefik Forward Auth.
How does it work?
Normally, Traefik proxies web requests directly to individual web apps running in containers. The user talks directly to the webapp, and the webapp is responsible for ensuring appropriate authentication.
When employing Traefik Forward Auth as "middleware", the forward-auth process sits in the middle of this transaction - traefik receives the incoming request, "checks in" with the auth server to determine whether or not further authentication is required. If the user is authenticated, the auth server returns a 200 response code, and Traefik is authorized to forward the request to the backend. If not, traefik passes the auth server response back to the user - this process will usually direct the user to an authentication provider (Google, Keycloak, and Dex are common examples), so that they can perform a login.
The advantage under this design is additional security. If I'm deploying a web app which I expect only an authenticated user to require access to (unlike something intended to be accessed publically, like Linx), I'll pass the request through Traefik Forward Auth. The overhead is negligible, and the additional layer of security is well-worth it.
Under normal Oauth2 / OIDC auth, you have to tell your auth provider which URLs it may redirect an authenticated user back to, post-authentication. This is a security feature of the OIDC spec, preventing a malicious landing page from capturing your session and using it to impersonate you. When you're securing many URLs though, explicitly listing them can be a PITA.
@thomaseddon's traefik-forward-auth includes an ingenious mechanism to simulate an "auth host" in your OIDC authentication, so that you can protect an unlimited amount of DNS names (with a common domain suffix), without having to manually maintain a list.
How does it work?
Say for example, you're protecting radarr.example.com. When you first browse to
https://radarr.example.com, Traefik forwards your session to traefik-forward-auth, to be authenticated. Traefik-forward-auth redirects you to your OIDC provider's login, but instructs the OIDC provider to redirect a successfully authenticated session back to
https://auth.example.com/_oauth, rather than to
When you successfully authenticate against the OIDC provider, you are redirected to the "redirect_uri" of
https://auth.example.com. Again, your request hits Traefik, which forwards the session to traefik-forward-auth, which knows that you've just been authenticated (cookies have a role to play here). Traefik-forward-auth also knows the URL of your original request (thanks to the X-Forwarded-Host header). Traefik-forward-auth redirects you to your original destination, and everybody is happy.
This clever workaround only works under 2 conditions:
- Your "auth host" has the same domain name as the hosts you're protecting (i.e., auth.example.com protecting radarr.example.com)
- You explictly tell traefik-forward-auth to use a cookie authenticating your whole domain (i.e. example.com)
Traefik Forward Auth needs to authenticate an incoming user against a provider. A provider can be something as simple as a self-hosted dex instance with a single static username/password, or as complex as a Keycloak instance backed by OpenLDAP. Here are some options, in increasing order of complexity...
- Authenticate Traefik Forward Auth against a self-hosted Dex instance with static usernames and passwords
- Authenticate Traefik Forward Auth against a whitelist of Google accounts
- Authenticate Traefik Forward Auth against a self-hosted Keycloak instance with an optional OpenLDAP backend
Chef's notes 📓
Tip your waiter (sponsor) 👏
Did you receive excellent service? Want to compliment the chef? (..and support development of current and future recipes!) Sponsor me on Github / Ko-Fi / Patreon, or see the contribute page for more (free or paid) ways to say thank you! 👏
Employ your chef (engage) 🤝
Is this too much of a geeky PITA? Do you just want results, stat? I do this for a living - I'm a full-time Kubernetes contractor, providing consulting and engineering expertise to businesses needing short-term, short-notice support in the cloud-native space, including AWS/Azure/GKE, Kubernetes, CI/CD and automation.
Learn more about working with me here.
Flirt with waiter (subscribe) 💌
Want to know now when this recipe gets updated, or when future recipes are added? Subscribe to the RSS feed, or leave your email address below, and we'll keep you updated.