Rate limiting WordPress Logins with Traefik

Marcel Aust

— 1 min read

I'm hosting a few Wordpress websites on my server for various people. Recently after setting up my Log monitoring with loki (https://blog.marcel-aust.de/monitoring-with-prometheus-loki-alloy-grafana-2/) I noticed a lot of login attempts by various bots there to the wp-login.php endpoint.

I'm running the Wordpress instances each in its own docker-compose stack. As already previously mentioned I'm using traefik as ingress proxy to proxy the requests and do the tls termination.

To fix the issue with the login attempts I employed a Traefik ratelimit (https://doc.traefik.io/traefik/middlewares/http/ratelimit/). This is what my new wordpress config inside the docker-compose stack looks like now:

wordpress:
    image: wordpress:latest
    restart: unless-stopped
    depends_on:
      - db
    environment:
      # ...
    labels:
      - "traefik.enable=true"
      - "traefik.http.routers.wordpress1.rule=Host(`www.example.com`)"
      - "traefik.http.routers.wordpress1.entrypoints=websecure"
      - "traefik.http.routers.wordpress1.tls.certresolver=le"
      - "traefik.http.routers.wordpress1.service=wordpress1"
      - "traefik.http.services.wordpress1.loadbalancer.server.port=80"
      # Login Ratelimit
      - "traefik.http.routers.wordpress1-rl.rule=(Host(`www.example.com`)) && PathPrefix(`/wp-login.php`)"
      - "traefik.http.routers.wordpress1-rl.entrypoints=websecure"
      - "traefik.http.routers.wordpress1-rl.tls.certresolver=le"
      - "traefik.http.routers.wordpress1-rl.middlewares=wp-login-rl@file"
      - "traefik.http.routers.wordpress1-rl.service=wordpress1"
      - "traefik.docker.network=traefik-net"
    volumes:
      - wordpress1:/var/www/html
    networks:
      - traefik-net
      - default

The corresponding wp-login-rl config looks like this:

http:
  middlewares:
    wp-login-rl:
      rateLimit:
        average: 10
        period: 60m

This will limit all request to the wordpress login page which reduced the login attempts drastically.

Read more