Strong suggestion for Tailscale here. It is incredibly easy to use and very easy to set up with multiple users. Opening ports directly to the internet is a thing of the past for me now, ever since I started.

So the reason you’d want a reverse proxy is because it handles security and would do a much better job of it than an exposed jellyfin port.

Public FQDN -> your home IP -> your router allows 443/whatever to your reverse proxy -> it handles SSL and being hit by the internet (look into nginx security and even fail2ban) -> proxy serves up whatever insecure site/app you’d like.

How Dangerous is it to just open the port for it on my router and access it like that?

It would probably be fine

Lets say i want to access jellyfin from Kodi on my xbox or something outside my network, the vpn solution wouldnt work for this i would think.

Depending how you set things up it will. Connect your computer to your VPN and it will have access. Set up a router as a VPN gateway (basically it would connect to the public/untrusted network and makes another network that routes its traffic through the VPN, as a bonus this also prevents snooping on the part of the untrusted network operator). Something like this could also be done in a more permanent way on remote networks, a VPN gateway to an IP range elsewhere. Only issue there would likely be for family members when they travel.

My issue with reverse proxies, and why im asking, is it seems less secure? I mean Im well aware that an IP is easy to get, i guess. But how likely is someone to look for something on my network specifically? With reverse proxies it seems like i would be broadcasting my server to the internet in a way its easier to happen across, than someone being interested in a random residential IP.

You seem to not exactly understand what a reverse proxy is, and seem to be confusing it with a domain name or something, while also not fully understanding how those work either.

A reverse proxy is a piece of software designed to accept HTTP(s) requests and “proxy” that request to the proper server. Reverse proxies are usually hardened against attacks better than application servers, and it is usually a good idea to put one in front of any application you are exposing to the internet. There is no “broadcasting” involved, basically you would port forward to the proxy as a middle-man of sorts instead of to jellyfin directly. It is also easy to set up SSL on the reverse proxy, regardless of whether you choose to encrypt your internal traffic between the reverse proxy and backend service. Nginx, Traefik, and Caddy are all popular options for reverse proxies and relatively easy to get set up.

With a domain name you purchase some name and can have that point at your IP address so you can use a memorable name rather than some random numbers. It also lets you update where the domain points so that when your IP changes (which does happen on most residential connections, even if only after something like an extended power/router outage).

Domain names also doesn’t involve announcing your IP anywhere other than your DNS provider, instead DNS is a convoluted system that basically lets someone querying for what the IP address associated with, say, example.com can figure out where your DNS server is and ask it what IP it should talk to.

When combined this a reverse proxy can be used to decide what to do with traffic based on which domain name (example.com vs jellyfin.example.com) someone is accessing, all on the same IP. You can do all sorts of neat things for every different domain your reverse proxy handles, like requiring authentication, using IP whitelists, and more.

I run a minecraft server for friends on my main computer anyway, and i know tons of people do that, theoretically thats the same level of danger as opening my network for jellyfin specifically.

Wouldn’t it be nice to just be able to tell your friends to connect to minecraft.example.com, and not even have to worry about telling them a port or anything? That’s another neat thing you can do when you have control over DNS. You can even make a page for your minecraft server that works in a web browser with instructions, links to things (mod downloads, discord, etc), a community forum, or whatever else you want.

VPN isnt an option because of this xbox stuff i mentioned and people in my family who have 0 chance of understanding it regardless.

So what is the better option, going through this reverse proxy ( which im actually also unsure would work with kodi) or rawdog the server on my network. I guess leaving the server exposed? or every device even.

I would absolutely run a reverse proxy. I would suggest buying a domain name (Usually like ~$15 a year or something, depending on the domain and registrar you go with. I am a fan of Cloudflare).

Regardless what you go with, best of luck in your self-hosing journey! LMK if you have any questions about reverse proxies or domains/DNS.

Edited to add: Attackers are totally interested in random residential IPs. They are constantly running port scans to find vulnerable software/devices/whatever that they can attack. Sometimes it’s just to add them to a botnet, other times it can be a vector to scan and decide to further exploit your network. Some of the biggest DDoS attacks ever have been via insecure cameras or other devices that people have just port forwarded to put them on the internet. In some cases a reverse proxy could have saved them. Also, “security through obscurity” can work (just shouldn’t be relied on). Unless someone specifically thinks to check whether your server responds to camera.example.com, they would never even know you are exposing a potentially vulnerable device.

Just put jellyfin and tailscale on an old windows computer and it’s setup in less than 10 minutes.

Not only is it likely, it’s pretty much guaranteed that a bot will scrap your network if it’s opened to the public.

Someone may have commented this already but my recommendation is to set up an overlay network like tailscale or twingate.

Doesn’t require you to open any ports on your firewall, and Tailscale at least is very performant since it uses Wireguard as it’s underlying protocol. (I have yet to test Twingate but I’ve heard positive things.)

It will require a little more setup per device but it’s honestly incredibly simple and more than secure enough for a home network.

Tailscale also has something called a subnet router which you can use to get incompatible devices onto the tailnet.

I use a reverse proxy so I can just use a hostname and not need a port. I run Jellyfin that way no problem, function-wise.

Additionally, not having a domain won’t necessarily protect you since you do have people out there scanning for ports and when they see 8096, they’re going to immediately know it’s a Jellyfin/Emby server and any vulnerabilities associated with those. If you use a reverse proxy, they only see 443 which is…pretty much every other site on the internet. That’s security through obscurity, I know, but it will help mitigate some of the easier attacks.

I’ll say that everything I have to have a port open for (mostly game servers) gets targeted by the internet at large despite the fact that I’ve published the address and port absolutely nowhere online and only shared it with close friends. I almost never get anyone trying to log in to my other services.