Will definitely let you know if I figure it out, for sure.

Before putting Pi-hole behind Traefik, it worked perfectly via :/admin. And the logs for Pi-hole now in Traefik show that it is up and working, and I get the login page. But just can’t get beyond it.

The guides I’ve seen show how to structure the Traefik labels with and without the addprefix middleware, and both apparently work. So I’m wondering if by following several guides and taking bits from each, I’ve ended up overlooking something.

I’ll try and expose 80 and see if it makes a difference, but like I say everything is up and running in the backend, I just can’t get past the login screen on the frontend.

Thanks. Yeah it’s really frustrating. There are guides out there that show that it’s possible, but I think I’m missing something somewhere. I’m combining guides together so more than likely I’ve missed a step somewhere.

Just a quick update on where I’m up to…

I’ve managed to get all my containers working behind the Traefik reverse proxy with SSL. I’ve also deployed a Cloudflare DDNS container in Docker and have linked the external IP address of my Synology NAS to Cloudflare. I haven’t port forwarded 80 and 443, though, so it’s not accessible over the internet. So I’ve added local DNS into Pi-hole so I can access all the containers using subdomains.

I’ve also deployed an Authelia container and have started running through my containers adding 2FA in front of them all.

I should probably point out at this juncture, that if I encounter any errors, the HTTP 404 page that I get is a Cloudflare one - I assume that’s expected behaviour?

So, the final three bits I’m struggling with now are:

• Pi-hole behind the reverse proxy
• Portainer behind the reverse proxy
• Accessing Vaultwarden over the internet (because as soon as I leave my house, if the vault hasn’t synced then I don’t have access to all my passwords) - unless anybody has a better suggestion?

Portainer - I have no idea how I do it, because I use it to manage my containers, so don’t have the config for Portainer in Portainer (obviously). So if I screw up the config, how am I getting back in to Portainer to fix it?

And the far more troubling one is Pi-hole. I just cannot get that thing working behind the reverse proxy.

I’ve followed a few different guides (though none of them are recent), and the below is the latest docker-compose I have. It will bring up the login page, but when I login it keeps returning me back to the login page - it won’t go to the main admin page.

version: "3.7"

services:
  pihole:
    container_name: pihole
    image: pihole/pihole:latest
    restart: unless-stopped
    networks:
      - medianet
      - npm_network
    ports:
      - 8008:80
      - 53:53/tcp
      - 53:53/udp
    environment:
      - TZ=Europe/London
      - WEBPASSWORD=xxxxxxxxxx
      - FTLCONF_LOCAL_IPV4=192.168.1.116
      - WEBTHEME=default-auto
      - DNSMASQ_LISTENING=ALL
      - VIRTUAL_HOST=pihole.mydomain.com
    volumes:
      - /path/to/pihole:/etc/pihole
      - /path/to/pihole/dnsmasq.d:/etc/dnsmasq.d
    cap_add:
      - NET_ADMIN
    labels:
      - traefik.enable=true
      - traefik.http.routers.pihole.entrypoints=http
      - traefik.http.routers.pihole.rule=Host(`pihole.mydomain.com`)
      - traefik.http.middlewares.pihole-https-redirect.redirectscheme.scheme=https
      - traefik.http.routers.pihole.middlewares=pihole-https-redirect
      - traefik.http.middlewares.pihole-addprefix.addprefix.prefix=/admin
      - traefik.http.routers.pihole.middlewares=pihole-addprefix
      - traefik.http.routers.pihole-secure.entrypoints=https
      - traefik.http.routers.pihole-secure.rule=Host(`pihole.mydomain.com`)
      - traefik.http.routers.pihole-secure.tls=true
      - traefik.http.routers.pihole-secure.service=pihole
      - traefik.http.services.pihole.loadbalancer.server.port=80

networks:
  medianet:
    external: true
  npm_network:
    external: true

Update from this morning.

So far I’ve got the Traefik dashboard and my Homepage container using subdomains. Pi-hole is still an issue that I need to figure out.

I’ve decided to start on the SSL certificates and am following a couple of guides. Once I have those in place, I’ll start moving more containers behind subdomains.

I might have to expose my NAS IP to the internet and link it via Cloudflare, because I use an ExpressVPN on my Mac at all times, and when it’s turned on I can’t access ***any ***of my subdomains - this is obviously because ExpressVPN use their own DNS entries and don’t use the ones I’ve set. That will probably prevent me from using Vaultwarden (which is the whole purpose for all of this in the first place) because if I’m on the VPN I won’t be able to access the Vaultwarden container.

Unless anyone knows of a workaround for that?

Next steps:

• Get SSL working
• Figure out how to access subdomains whilst on the VPN (or get a DDNS account, create an A record in Cloudflare and point it at the DDNS IP, and open up ports 80 and 443)
• Spin up a Vaultwarden container via a subdomain
• Put all my other services behind subdomains
• Figure out how to get Pi-hole working via Traefik and subdomain
• Figure out how to get Tailscale access to my containers when not on my LAN

Figured it out. It’s a weird setting on Netgear routers whereby you have to also update the MAC address. All been working well for the last few hours and getting queries running through Pi-hole.

I’ve also got my Homepage container setup at http://home.mydomain.com and configured Traefik a little further so it’s now accessible from http://traefik.mydomain.com (no port).

For the past few hours I’ve been struggling with getting Pi-hole behind Traefik and accessible using http://pihole.mydomain.com. Only works if I stick /admin on the end, which defeats the object of using a subdomain. Found a forum post suggesting to use Traefik’s addPrefix after declaring the Host as pihole.mydomain.com, which works great for accessing the login screen, but when you enter the password it just loops back to the login screen.

Also tried a few other things that ultimately broke the Pi-hole container and took out my entire connection, as everything is dependent on Pi-hole for DNS! So need to figure out some kind of resiliency/backup for that (my router is using the NAS IP as it’s primary and only DNS server).

So, some progress. I’ve set Pi-hope back to IP:port and I’m gonna focus on getting other containers behind Traefik and leave Pi-hole till last. Then and only then will I look at SSL certificates (unless it’s advised to do it earlier?)

Any pointers on any of the above would be appreciated! And thanks again for getting me this far.

I’ve just added in a macvlan network to my Pi-hole compose as well, not sure if it’s making any difference or not.

I wasn’t planning on using Pi-hole for DHCP - I have a LOT of reserved addresses on my network and I don’t fancy having to move them all over. My hope had been to use Pi-hole for DNS but keep the DHCP reservation with the router.

I’ve manually updated the DNS on my Mac to 192.168.1.116 and I can now access the Traefik dashboard via http://traefik.mydomain.com:8080 (so, getting there). So some kind of issue with the DNS on my router I think - caching maybe?

Actually, no I don’t see anything coming through.

So the IP address of my router is 192.168.1.1, IP of my NAS is 192.168.1.116.

Checked the DNS on my Mac and it’s 192.168.1.1. Checked the DNS on my NAS and it’s 192.168.1.1. I changed the DNS in my router to 192.168.1.116.

Have I missed a step somewhere?

OK so made a start with this. Spun up a Pi-hole container, added mydomain.com as an A record in Local DNS, and created a CNAME for traefik.mydomain.com to point to mydomain.com.

In Cloudflare, I removed the mydomain.com A record and the www CNAME record.

Doing an nslookup on mydomain.com I get

Non-authoritative answer:
*** Can't find mydomain.com: No answer

Which I guess is to be expected.

However, when I then navigate to http://traefik.mydomain.com in my browser, I’m met with a Cloudflare error page: https://imgur.com/XhKOywo.

Below is the docker-compose of my traefik container:

traefik:
    container_name: traefik
    image: traefik:latest
    restart: unless-stopped
    networks:
      - medianet
    ports:
      - 80:80
      - 443:443
    expose:
      - 8080
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - /volume1/docker/traefik:/etc/traefik
      - /volume1/docker/traefik/access.log:/logs/access.log
      - /volume1/docker/traefik/traefik.log:/logs/traefik.log
      - /volume1/docker/traefik/acme/acme.json:/acme.json
    environment:
      - TZ=Europe/London
    labels:
      - traefik.enable=true
      - traefik.http.routers.traefik.rule=Host(`$TRAEFIK_DASHBOARD_HOST`) && (PathPrefix(`/api`) || PathPrefix(`/dashboard`))
      - traefik.http.routers.traefik.service=api@internal
      - traefik.http.routers.traefik.entrypoints=traefik

My traefik.yml is also nice and basic at this point:

global:
  sendAnonymousUsage: false

entryPoints:
  web:
    address: ":80"
  traefik:
    address: "8080"

api:
  dashboard: true
  insecure: true

providers:
  docker:
    endpoint: "unix:///var/run/docker.sock"
    watch: true
    exposedByDefault: false

log:
  filePath: traefik.log
  level: DEBUG

accessLog:
  filePath: access.log
  bufferingSize: 100

Any ideas what’s going wrong? I’m unclear on why the domain is still routing to Cloudflare.

Absolute superstar, thanks for your help so far. I’ll make a start on some of this tomorrow and see how far I get — either with Traefik or NPM.

Do I need to do anything with the domain itself on Cloudflare at the moment? Or do I just leave it with its current A record pointing at an IP address (it was done as part of the setup in Cloudflare so I have no idea what that IP address is).

Obviously that domain in reality will just sit there doing nothing.

Thanks. I realise they’re all pretty basic questions. But brace yourself: more are on their way!

So… no, I don’t want to give external access - I’m not running any services that anyone would want/need access to - other than perhaps my Jellyfin server, but not sure I even want anyone accessing that. So let’s assume for right now, no access to the outside world. Therefore, no port forwarding required.

So to get access to my internal network from the domain, do I simply setup local DNS records in something like Pi-hole, to point mydomain.com to the internal IP or my NAS? Kind of like a network-wide equivalent of modding the /etc/hosts file on my machine?

Perhaps a(nother) silly question but, what’s to stop me doing that now with a completely random domain name? Is there some kind of authentication I’d need to go through to prove that mydomain.com is, in fact, mine? Or does it simply not matter since it’s internal only?

If I’ve understood correctly, then, I don’t need Cloudflare at all in my setup if there’s no external access? Nothing to proxy, nothing to protect?

Assuming I get all of the above working and traffic routing to my containers, how would I then go about setting up SSL? Can that be done through Traefik rather than Cloudflare? Even if the domain isn’t external?

Thanks. There’s definitely stuff in here I want to do, I just need to figure out the order of play and break it down a bit.

As per reply to another comment.

Do I have to port forward 80 and 443 no matter what? Ideally I don’t want to forward anything.

Do I need DDNS in here somewhere, i.e. create a DDNS and link it to my NAS, create an A record in Cloudflare to point my domain to the external IP of the DDNS? Is that how I get into my NAS from the domain without worrying about the IP changing? How do I then prevent anybody accessing the NAS admin on port 5000/5001, as well as anything else except the containers I expose via Traefik?

Interesting, I’ve never considered Cloudflare Tunnels. Thanks.

However I do remember seeing this video the other day, that suggests perhaps it’s not always the best solution? Not sure this applies here, though: https://www.youtube.com/watch?v=oqy3krzmSMA.

Would love an up-to-date tutorial on how to do this without a domain name. I don’t own one but would still much prefer to use https://jellyfin.myserver.home than http://192.168.1.200:8096.

Thanks. Yeah I originally had plain ol’ NGINX (granted, without the additional networking I have now) and using

location /xyz {
$proxy_pass http://a.b.c:123
}

(I forget the exact syntax now) was pretty easy to route stuff around.

But that was all done under an initial

server :80 {
...
}

section, which might be the part I’m initially struggling with. It just doesn’t seem as intuitive in NPM despite me seeing lots of people preferring it over plain NGINX and Traefik.

NPM is why I have the macvlan and bridge networks in there in the first place, because ports 80 and 443 are in use on the Synology NAS. Otherwise, I could revert back to running a script that frees up those ports on boot-up and use NPM within the same network as all my other containers.

Thanks. Yeah, .local was just an example for the sake of posting something on here. In reality, I would be using ds1618plus.home. But good to know .local is reserved!