Selfhosted

If you're only trying to use Jellyfin at home, you don't need any reverse proxy or domain. All you need is for both devices to be on the same network, and for the Raspberry Pi to have a fixed internal IP address (through your router settings).

On the Shield, you just give the Jellyfin app that IP address and port number (10.0.0.X:8096) to connect and you're good to go.

First of all, is that all correct or have I misunderstood something?

There's a couple things you've got a bit wrong:

I think I’m correct in saying that mysubdomain.mydomain.com is actually an IP address and a public port, so something like 123.456.7.8:443, then Cloudflare - which is the reverse proxy - gets involved (somehow? how?) to say “ah, 123.456.7.8:443, you obviously want to go to funkless.raspberry.pi:NGINX (or rather something like 987.654.3.2:443)” and then NGINX - which is the proxy-proxy, not a reverse-proxy - goes (somehow? how?) “ah, 987.654.3.2:443, you obviously want to go to 987.654.3.2:8096 which is jellyfin”)

I'm not sure what Cloudflare product you're using, but I use it as a DNS server for my domain. If you're doing the same thing - you'll have configured A records and such if so - then what's happening is this:

1. You request subdomain.mydomain.com. Your device needs the IP to connect to, so it asks Cloudflare for the IP address. Think of this like calling information to find a phone number.
2. Then your device initiates a request to the IP address it gets back. This is where TLS gets used, and encrypts your connection to that IP address. It also includes the domain requested in a header for the request.
3. Nginx (which is a reverse proxy, meaning it handles incoming rather than outgoing connections) receives the connection and looks at the domain header. Then it looks in its configuration for the IP and port it should connect to, and forwards the request

However, if you're using some other thing at Cloudflare to make a VPN this might be entirely wrong.

How does mysubdomain.mydomain.com know it’s me and not some random or bot?

Unless you've implemented some kind of filtering or authentication in Nginx, it doesn't. I got around this by configuring HAProxy - which is like Nginx - to only allow requests from my local network except for specific domains that I want to be public.

Is this step “port forwarding” or “opening ports” or “exposing ports” or either or both? (I don’t understand these terms)

Exposing or opening ports is something you do with a firewall. The purpose of Nginx is to make it so you only have to open 1-2 ports, and Nginx will handle redirecting traffic based on its configuration.

If my browser when accessing mysubdomain.mydomain.com is always going to port 80/443, does it need to be told it’s going to talk to cloudflare - if so how? - and does cloudflare need to be told it’s going to talk to NGINX on my local machine - if so how?

If you're using Cloudflare like I described above, you will only need to tell Cloudflare the public IP address of your Nginx server. Generally you do this by telling your domain registrar (where you buy domain.com) to use Cloudflare's "nameservers" and then configure Cloudflare to point to your public IP address.

How do I tell NGINX to switch from local:443 to local:8096 (assuming I’ve understood this correctly)

You edit the Nginx config to add something like this:

server {
    server_name subdomain1.example.com;
    location / {
        proxy_pass       http://hostname1:port1;
    }
 }

Then, when Nginx receives a connection request for subdomain1.example.com for any location, it will proxy it to the configured hostname (or IP address) and port.

Is there a difference between an SSL cert and a public and private key - are they three things, two things or one thing?

There are two parts to an SSL cert: A public key and a private key. How SSL works is... complicated, but suffice to say the public key is shared with the connection, and the private key is hidden on the server. You can encrypt data with either one, and only the matching key can decrypt it. This allows both sides to trust the connection and for nobody else to see the data.

Doesn’t a VPN add an extra step of fuckery to this and how do I tell the VPN to allow all this traffic switching without blocking it and without showing the world what I’m doing?

The Internet is like an ogre: It has layers. HTTP and DNS are on one layer, VPNs are a different layer. HTTP and DNS traffic can travel over the Internet, or your local network or over the VPN.

If you're just setting up a local Jellyfin server, you technically don't need Cloudflare. Your home router will probably let you hard-code a DNS entry for a local IP address, which will keep all of that traffic on your local network. And if you do that right you won't even need SSL.

Gluetun just looks like a text document to me (compose.yml) - how do I know it’s actually protecting me?

I'm not familiar with how Gluetun works, but it's not just compose.yml. When you start it with docker-compose run it will download and extract the code to run Gluetun, and configure networking and other things.

Lots of people contributed really good answers, so I don't have anything valuable to add to their answers. But I wanted to point out for your detailed question, you include what you have done, what is your understanding and what are your shortcomings clearly. As opposed to a lot of posts with vague, detail-challenged narratives, that's a top notch post.

And the community delivered by giving good answers, so go community!

Also, you didn't just ghost after the initial post and interacted.with the people who graciously donated their time, so another bonus point there, as well.

As others have mentioned (and also explained in quite some detail) you're trying to bite off a lot at once. First, for Jellyfin locally you can ignore most of that.

And if you really want to learn the ins and outs of all that (and I can recommend it, it's useful), then I suggest you start with some simple web app. Something like note taking or maybe even something trivial like a whoami service, which basically just echos some information it was sent back to you. That's super useful because you know that it is unlikely to be broken, so you can focus on the networking/port forwarding issues. And once you've got that working and have a rough feeling how this all works you can go on to more complex setups that actually do something useful.

A lot of this is being complicated for you by not understanding networking fundamentals. I'd suggest looking into a Network+ certification which will cover all of these basics like DNS. You don't have to actually get the cert, just going through the motions on learning the material should help a lot.

You seem to be close on grokking the whole picture and just need some of the basics that are hard to pick up from just doing things at home. A lot of work has been done to try abstract that away from consumers in order to make things easier which is making it harder for you.

Look, this is a large puzzle you're trying to solve all at once. I'll try to answer at least some of it. I'd advise you take these things step by step. DM me if you need some more help, I may have time to help you figure things out.

I paid for and installed mullvad (app) but it crashes a lot (for over a minute every 20 seconds), so it looks like I need to configure something like gluetun to do it instead.

Check the error logs and see what's wrong with it instead. How is it crashing? Did you check stdout and stderr (use docker attach or check the compose logs)?

If I want to watch them on my TV I need to connect something to my TV that talks to the raspberry pi, so I have an NVIDIA shield with Jellyfin installed on it - but in order for the NVIDIA-Jellyfin to connect to the RaspberryPi-Jellyfin it needs to go through the internet (if this is not the case, how does one point the NVIDIA-Jellyfin at the Raspberry Pi jellyfin?)

Technically not. You can use the Jellyfin web UI to stream directly from the RPi. You may need the shield if the RPi does not have enough resources for streaming, but I'd try it out first. Try to get the IP the Raspberry is listening on on your local network and put that in a web browser on a computer first. IF you get the web UI and can watch stuff, then try a web browser on your TV, or cast your computer to the TV or something. As long as you have a web browser you should be fine.

First of all, is that all correct or have I misunderstood something?

You should look a bit into how the internet, DNS and IP addresses work on the public internet and private networks. You can absolutely set it up so that traffic from your local network hitting your domain never leaves your home, while if you try the same from somewhere else, you get an encrypted connection to your home. You're a bit all over the place with these terms so it's hard to give you a straight answer.

How does mysubdomain.mydomain.com know it’s me and not some random or bot?

If the question is whether how the domain routes to your IP, look up how DNS works. If you are asking how to make sure you can access your domain while others can't look up the topic of authentication (basically anything from a username/password to a VPN and network rules).

How do I tell Cloudflare to switch from web:443 to local:443 (assuming I’ve understood this correctly)

If I remember correctly, Cloudflare forwards HTTP/S traffic only, so don't worry about the ports, that's all it will do. About the domains, you need to have a fixed public IP address for that, and you have to give Cloudflare by setting a DNS A record for an IPv4 address and/or an AAAA record for an IPv6 address.

So something like this: A myhost.mydomain.com 123.234.312.45

Is this step “port forwarding” or “opening ports” or “exposing ports” or either or both?

Nope. Port forwarding is making sure that your router knows what machine should answer when something on the Internet comes knocking. So if the RPi port 8096 is "forwarded" to the router, then if something from the internet connects to the router's 8096 port, it will get to your RPi instead of something else. Opening ports has to deal with firewalls. Firewalls drop all connections on all ports that are not open, for security reasons. By opening a port you are telling the firewall what entities outside your device can connect to a service like Jellyfin listening on that port. Exposing ports is Docker terminology, it is the same as port forwarding except instead of "moving" a port from your machine to your router you "move" a port from a container to your machine.

If my browser when accessing mysubdomain.mydomain.com is always going to port 80/443, does it need to be told it’s going to talk to cloudflare - if so how? - and does cloudflare need to be told it’s going to talk to NGINX on my local machine - if so how?

The DNS server you are hosting the domain from will propagate that info through the DNS network. Look up how DNS works for more info. If your domain is managed by Cloudflare, it should "just work". Cloudflare knows it talks to your router by you setting up a DNS record in their UI that points to your router, where your RPi's port should be forwarded, which directs traffic to your RPi, on which your NGINX should be listening and directing traffic to your services.

How do I tell NGINX to switch from local:443 to local:8096 (assuming I’ve understood this correctly)

Look up NGINX virtual servers and config file syntax. You need to configure a virtual server listening on 443 with a proxy_pass block to 8096.

Is there a difference between an SSL cert and a public and private key - are they three things, two things or one thing?

Yes, SSL certs are the "public keys" of an X509 pair, while what you know as "public and private keys" are RSA or ED25519 key pairs. The former is usually used to make sure that the server you are accessing is indeed who it claims to be and not a fake copy, it's what drives HTTPS and the little lock icon in your browser. RSA or ED25519 keys are used for authentication as in instead of a username and password, you give a public key to a service, then you can use a private key to encrypt a message to auth yourself. One service you might know that it uses it is SSH.

Doesn’t a VPN add an extra step of fuckery to this and how do I tell the VPN to allow all this traffic switching without blocking it and without showing the world what I’m doing?

A VPN like Mullvad is used for your outgoing traffic. All traffic is encrypted, the reason you want a VPN is not so that others can't see your messages, it's so that your ISP and the other people forwarding your messages don't know who you're talking to (they'll only know you're talking to your VPN), and so that the people you're talking to don't know who you are (they are talking to your VPN). You need this so your ISP doesn't see you going to pirate sites, and so that other pirates, and copyright trolls acting as pirates don't know who you are when you talk to them and exchange files using torrents.

Gluetun just looks like a text document to me (compose.yml) - how do I know it’s actually protecting me?

I don't know shit about Gluetun, sorry.

From https://nginxproxymanager.com/ : "Add port forwarding for port 80 and 443 to the server hosting this project. I assume this means to tell NGINX that traffic is coming in on port 80 and 443 and it should take that traffic and send it to 8096 (Jellyfin) and 5000 (ombi) - but how?

Again, look up virtual servers in NGINX configuration. You need a virtual server listening on 80 and 443 proxying traffic to 8096 and 5000, separating on hostnames I guess.

Also from that site: “Configure your domain name details to point to your home, either with a static ip or a service like DuckDNS or Amazon Route53” - I assume this is what Cloudflare is for instead of Duck or Amazon? I also assume it means "tell Cloudflare to take traffic on port 80 and 443 and send it to NGINX’s 80 and 443 as per the previous bullet) - but how?

Add a DNS A record.

Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I've seen in this thread:

[Thread #458 for this sub, first seen 28th Jan 2024, 18:55] [FAQ] [Full list] [Contact] [Source code]

Tons of good responses here. I’m surprised that nobody has brought up Tailscale though. It’s def the easiest vpn solution I have found. It’s got some great documentation and how to projects to get a home lab running and it’s got its own domain system baked in most of it being zero configuration. You can access mullvad vpn exit nodes straight from it, and set up those domains with ssl super easy e.g.

sudo tailscale serve —https=443 localhost:8096

That single command would allow any other devices connected to your Tailscale account to reach your Jellyfin using the domain “{serverhostname}.[tail-scale].ts.net” complete with a private reverse proxy and ssl cert.

There are a few things to click around in tailscale on but it’s a extremely easy to use free application that has made my self hosted life significantly easier due to my system living behind multiple firewalls that I sadly have no control over.

If you get a reverse proxy setup all you need is port 80 and 443 and configure it it'll expose the services that you want to be exposed through the subdomain Personally I've got Traefik service sitting on my media server and anything I want to expose goes through it .it has the details for the connection to cloudflare and so long as I direct it properly both on the container side and Traefik it'll run as expected. The idea is if you go to say jellyfin.example.com cloudflare will direct that at at your reverse proxy(nginx in this case) which then redirects to the right machine/container because you entered from "jellyfin" .

The VPN gluten it is another container that will have the login details to your provider .

I'm still working my way through the self hosted rabbit hole myself, however I used a combination of Google and this sitehttps://www.smarthomebeginner.com/traefik-docker-compose-guide-2022/ The entire site not just the specific article linked . As well as https://trash-guides.info/