How do I securely host Jellyfin? (Part 2)
How do I securely host Jellyfin? (Part 2)
I am making this post in good faith
In my last post I asked about securely hosting Jellyfin given my specific setup. A lot of people misunderstood my situation, which caused the whole thread to turn into a mess, and I didn't get the help I needed.
I am very new to selfhosting, which means I don't know everything. Instead of telling me that I don't know something, please help me learn and understand. I am here asking for help, even if I am not very good at it, which I apologize for.
With that said, let me reoutline my situation:
I use my ISP's default router, and the router is owned by Amazon. I am not the one managing the router, so I have no control over it. That alone means I have significant reason not to trust my own home network, and it means I employ the use of ProtonVPN to hide my traffic from my ISP and I require the use of encryption even over the LAN for privacy reasons. That is my threat model, so please respect that, even if you don't agree with it. If you don't agree with it, and don't have any help to give, please bring your knowledge elsewhere, as your assistance is not required here. Thank you for being respectful!
Due to financial reasons, I can only use the free tier of ProtonVPN, and I want to avoid costs where I can. That means I can only host on the hardware I have, which is a Raspberry Pi 5, and I want to avoid the cost of buying a domain or using a third party provider.
I want to access Jellyfin from multiple devices, such as my phone, laptop, and computer, which means I'm not going to host Jellyfin on-device. I have to host it on a server, which is, in this case, the Raspberry Pi.
With that, I already have a plan for protecting the server itself, which I outlined in the other post, by installing securecore on it. Securing the server is a different project, and not what I am asking for help for here.
I want help encrypting the Jellyfin traffic in transit. Since I always have ProtonVPN enabled, and Android devices only have one VPN slot enabled, I cannot use something such as Tailscale for encryption. There is some hope in doing some manual ProtonVPN configurations, but I don't know how that would work, so someone may be able to help with that.
All Jellyfin clients I have used (on Linux and Android) do not accept self-signed certificates. You can test this yourself by configuring Jellyfin to only accept HTTPS requests, using a self-signed certificate (without a domain), and trying to access Jellyfin from a client. This is a known limitation. I wouldn't want to use self-signed certificates anyways, since an unknown intruder on the network could perform a MITM attack to decrypt traffic (or the router itself, however unlikely).
Even if I don't trust my network, I can still verify the security and authenticity of the software I use in many, many ways. This is not the topic of this post, but I am mentioning it just in case.
Finally, I want to mention that ProtonVPN in its free tier does not allow LAN connections. The only other VPN providers I would consider are Mullvad VPN or IVPN, both of which are paid. I don't intend to get rid of ProtonVPN, and again that is not the topic of this post.
Please keep things on-topic, and be respectful. Again, I am here to learn, which is why I am asking for help. I don't know everything, so please keep that in mind. What are my options for encrypting Jellyfin traffic in transit, while prioritizing privacy and security?
How about creating your own LAN within the untrusted network?
Something like an inexpensive OpenWRT router would do fine. Connect all your devices and the server to the router. They are now on a trusted network. Set up Wireguard on the OpenWRT router to connect to Proton so that your outbound traffic from all your devices is secured.
I was looking for this. Op seems to be obsessed with "zero trust", so creating a trusted area for this stuff would be an easy win.
Exactly! I did that for a couple years until I found a cheap modem to replace the ISP modem. It didn't do any routing, so there was no weird NAT issue, it just converted the DSL signal to Ethernet with a WAN IP.
I didn't have to change any network settings on my LAN when I switched, or when I moved to another place with a different ISP. I had that same router for years, even after I got a dedicated AP for my house.
I work for an ISP, and this is a common practice among my peers
I have done this before as well when living in a dorm where wifi was shit so i did my own little setup in my room so I could stream to Crome cast etc on my own trusted lan. Get a small router with support for wire Guard vpn (i love mikrotik for this) and you have an easy way to tunnel out for all your devices.
Hey, this is off topic from the original post but could you tell me what device specifically you have used?
I am going to be moving into a dorm soon and was looking to set up my own VLAN or ehatever you need for a private network because I don't want to mess with the dorm router. I had a look at a Mikrotik switch (CRS310) but was unsure whether the fan noise would be too loud if I am staying in the same room and more importantly, whether this even allows me to do what I want to do
Edit: I misused the word dorm. It is a shared appartment rented with a couple of other students.
A lot of people have suggested Tailscale and it's basically the perfect solution to all your requirements.
You keep saying you need ProtonVPN which means you can't use Tailscale, but Tailscale actually supports setting up an exit node which is what you need. Put Protonvpn on the Raspberry Pi, then set it up as an exit node for your tailnet. There's a lot of people talking about how they did this online. It looks like they even have native support for bypassing the manual setup if you use Mullvad.
As long as every client has the ability to use Tailscale (I.e. no weird TVs or anything) this seems like it checks all your boxes. And since everything is E2EE from Tailscale, TLS is redundant and you can just use HTTP.
I'll just add my 2¢
Tailscale is incredibly powerful and they do a lot of work to make their systems intelligible, but the learning curve is still pretty steep. But still a great option.
One thing that I do, though it may not be as secure as a reverse proxy is just using tailscale funnel to expose my jellyfin instance.
I’d like to learn a self-hosted SSO but time is my least abundant resource at the moment.
Hi again.
How about the following idea:
Set up ProtonVPN on the raspberry pi.
On all other devices (or at least those you want to use Jellyfin on), switch from using Proton to using Wireguard. Unlike your phone, the raspberry pi has no trouble running multiple VPNs. I think the ProtonVPN limitations in regard to not allowing split tunneling don't apply here, since all outgoing traffic will still go via Proton.
Essentially, the Pi would function as a proxy for all of your traffic, "and also" host Jellyfin. You would still connect to http://192.168.20.10:8096/ (or whatever) on your devices, but that address would only resolve to anything when you are connected to the pi via Wireguard. No HTTPs, but "HTTP over Wireguard", if you will.
Nots that this requires you trusting the pi to the same degree that you trust your phone.
For your static devices (PC, TV) this should solve the problem. Devices which you take with you, like your phone, unfortunately will loose internet connectivity when you leave your home until you switch off Wireguard, and switch on Proton, and not be able to connect to Jellyfin when you return home, until you switch them back.
Essentially, you would have a "home" VPN and a "on the go" VPN, though you never need to connect to both. There might be ways to automate this based on WiFi SSID on Android, but I have not looked into it.
The Pros:
- this should meet all your requirements. No additional expenses, no domain, no dynDNS; no selfsigned certificate or custom CA; traffic is never unencrypted; works on all common devices.
- Wireguard is sufficiently lightweight to not bog down the pi, normally
- this is actually well within the intended use-cases for Wireguard, so no "black magic" required in configuring it
- if you ever do decide to get a domain, you can configure everything to always be connected to your pi via Wireguard, even on the go! Not required though.
The Cons:
- when you are new to selfhosting, Wireguard is a bit daunting to set up. It is not the easiest to debug (don't worry, it's easy to tell IF it is working, but not always WHY it isn't working). Some manual route handling is probably also required on the pi. It should definitely be doable though, but might turn this Jellyfin thing from a weekend project to a 2 week project...
- I have no experience with how well the pi runs Jellyfin. If the answer is "barely", then adding multiple concurrent Wireguard sessions might be a bad experience. Though in this case, you could only switch Proton to Wireguard whenever you want to watch Jellyfin.
- the manual switching might be annoying, but that is the price to pay here, so to speak
Edit: someone else already mentioned setting up your own trusted network with a second router. IMO that is the better, more hassle-free option IF you are willing to shell out the money. My suggestion is the "free" version of that, essentially 😄
Hi again.
Hi there!
Set up ProtonVPN on the raspberry pi.
I'm actually surprised nobody suggested simply using the Pi with OpenWrt as my own router. Though, that would make it hard to host Jellyfin.
Nots that this requires you trusting the pi to the same degree that you trust your phone.
For the most part, I trust the security of my Pi. I can hold it in my hand and see every line of code, after all!
Devices which you take with you, like your phone, unfortunately will loose internet connectivity when you leave your home until you switch off Wireguard, and switch on Proton, and not be able to connect to Jellyfin when you return home, until you switch them back.
I plan to post a tutorial about how to securely host Jellyfin. Another user gave a solution to this problem that I absolutely love, and I'll showcase it there. I don't want to spoil it :)
Could you explain Wireguard vs. Tailscale in this scenario?
Thank you all so much for your help! This is likely the solution I will go with, combined with another one, so again thank you so much!
P.S. I don't care if you wrap an ethernet cord around her finger, get going!
Tailscale is just a bunch of extra fancy stuff on top of Wireguard. If you don't need the fancy stuff, using raw Wireguard can be more lightweight, but might require more networking knowledge.
The biggest thing Tailscale brings you the table is NAT traversal. On top of that it uses direct Wireguard tunnels as necessary instead of creating a mesh like you usually would if you were using raw Wireguard. It also offers convenient bits of sugar like internal DNS, and it handles key exchanges for you so it's just generally easier to configure. When you do raw Wireguard you're doing all the config yourself, which could be a pro or a con depending on your needs—and you'll be editing config files, unlike Tailscale which has a GUI for most things. It also supports some more detailed security options like ACLs and I think SSO, while Wireguard is reliant on your existing firewall for that.
Here's what Tailscale has to say about it: https://tailscale.com/compare/wireguard
I've messed around with Tailscale myself, but ultimately settled on running Wireguard. The reason I do that though is because I trust my LAN, and I only run Wireguard at the edge. Tailscale really wants to be run on every node, which in turn is something that raw Wireguard theoretically can do but would be onerous to maintain. If I didn't trust my LAN, I'd probably switch to Tailscale.
I’m actually surprised nobody suggested simply using the Pi with OpenWrt as my own router. Though, that would make it hard to host Jellyfin.
A brief internet search shows that surprisingly, hosting Jellyfin on OpenWRT should work.... No idea how well though. Come to think of it, having OpenWRT on the pi might make it a lot easier to configure, with graphical settings available and so on.
Could you explain Wireguard vs. Tailscale in this scenario?
I've never used tailscale, I'm afraid. Normally I would say: just use whatever seems easier to set up on your device/network; however, note that tailscale needs a "coordinate server". No actual traffic ever goes through it, it just facilitates key exchanges and the like (from what I understand), but regardless, it's a server outside your control which is involved in some way. You can selfhost this server, but that is additional work, of course...
Thank you all so much for your help! This is likely the solution I will go with, combined with another one, so again thank you so much!
Glad I could help, after being so unhelpful yesterday :)
P.S. I don’t care if you wrap an ethernet cord around her finger, get going!
Eh... Marriage is not really common in either of our families. We agreed to go sign the papers if there ever is a tax reason, lol. Sorry if that's a bit unromantic :D Nice rings though ^^
I read the old thread and now this one.
As I understand it, you want to create connection between clients on your lan, but you don't trust your lan, so it's like having a raspberry pi server and some client both on the coffee shop network and you want them to communicate securely?
Tailscale is what you want. Easy setup, free, and allows exactly this to happen.
I use tailscale for exactly this purpose. And with the added benefit of bring able to watch media and manage the device remotely and easily
ProtonVPN in its free tier does not allow LAN connections
This is the limiting factor. In order to get around this, you'll have to put your Jellyfin server on the Internet. Hopefully you can enable port forwarding. If not, you have painted yourself into a corner.
If you cannot use self-signed or internal CA certs, you will also need a domain name, and something like Let's Encrypt to issue certs for that domain.
you’ll have to put your Jellyfin server on the Internet.
Don't.
Yeah, you shouldn't, but OP seems determined to hamstring themselves and do everything as convoluted as possible.
Although not ideal, I would be willing to pay for ProtonVPN (or another) if that's what is required. If I did have LAN connections, what are my options? Eventually I will get a more trustworthy router, but I still don't want to trust it by sending data in plaintext, even if I can control it and enable port forwarding.
Already answered in your previous post: https://lemm.ee/post/60855169/19569046
Again, Tailscale.
Since I always have ProtonVPN enabled, and Android devices only have one VPN slot enabled, I cannot use something such as Tailscale for encryption.
I previously proffered some information in the first thread.
But there's something I wish to clarify about self-signed certificates, for the benefit of everyone. Irrespective of whichever certificate store that an app uses -- either its own or the one maintained by the OS -- the CA Browser Forum, which maintains the standards for public certificates, prohibits issuance of TLS certificates for reserved IPv4 or IPv6 addresses. See Section 4.2.2.
This is because those addresses will resolve to different machines on different networks. Whereas a certificate for a global-scope IP address is fine because it should resolve to the same destination. If certificate authorities won't issue certs for private IP addresses, there's a good chance that apps won't tolerate such certs either. Nor should they, for precisely the reason given above.
A proper self-signed cert -- either for a domain name or a global-scope IP address -- does not create any MITM issues as long as the certificate was manually confirmed the first time and added to the trust store, either in-app or in the OS. Thereafter, only a bona fide MITM attack would raise an alarm, the same as if a MITM attacker tries to impersonate any other domain name. SSH is the most similar, where trust-on-first-connection is the norm, not the outlier.
There are safe ways to use self-signed certificate. People should not discard that option so wontonly.
It sounds like the clients do not have the ability to manually trust a self-signed cert.
@catloaf @litchralee As they shouldn't trust a self-signed cert. If you can run Jellyfin, you can run an internal CA and DNS server. Create a .internal domain in your DNS server and create certs based on it. Then just roll out your root ca cert.
Wow, there isn’t a single solution in here with the obvious answer?
You’ll need a domain name. It doesn’t need to be paid - you can use DuckDNS. Note that whoever hosts your DNS needs to support dynamic DNS. I use Cloudflare for this for free (not their other services) even though I bought my domains from Namecheap.
Then, you can either set up Let’s Encrypt on device and have it generate certs in a location Jellyfin knows about (not sure what this entails exactly, as I don’t use this approach) or you can do what I do:
- Set up a reverse proxy - I use Traefik but there are a few other solid options - and configure it to use Let’s Encrypt and your domain name.
- Your reverse proxy should have ports 443 and 80 exposed, but should upgrade http requests to https.
- Add Jellyfin as a service and route in your reverse proxy’s config.
On your router, forward port 443 to the outbound secure port from your PI (which for simplicity’s sake should also be port 443). You likely also need to forward port 80 in order to verify Let’s Encrypt.
If you want to use Jellyfin while on your network and your router doesn’t support NAT loopback requests, then you can use the server’s IP address and expose Jellyfin’s HTTP ports (e.g., 8080) - just make sure to not forward those ports from the router. You’ll have local unencrypted transfers if you do this, though.
Make sure you have secure passwords in Jellyfin. Note that you are vulnerable to a Jellyfin or Traefik vulnerability if one is found, so make sure to keep your software updated.
If you use Docker, I can share some config info with you on how to set this all up with Traefik, Jellyfin, and a dynamic dns services all up with docker-compose services.
To be totally honest I didn’t read your entire post, but just from your intro I think we are in similar situations. ISP router, low costs, using only the hardware you have around. I’ve solved a lot of stuff with Tailscale. None of my services are public facing and instead I connect to them over Tailscale (could be replaced with wireguard).
The wall I’m hitting you or maybe others could help with, is accessing my services from sub domains of a single Tailscale address rather than having to type port numbers for everything. I know this involves a reverse proxy and DNS (I use PiHole for that), but I’m stuck trying to configure the two in a way that actually works. Once I finally ditch iOS for good I’ll probably just sync a hosts file between all my devices using Syncthing to help streamline the DNS situation.
Once I finally ditch iOS for good
I had that feeling for all too long. It's so refreshing to break free. Word of advice: make sure to switch over your Signal account to make your new phone as an owner
You planning on GrapheneOS?
Honestly I want a Linux phone, but the scene needs to mature a bit. I’d also like a physical keyboard, so I’m even more limited in my options. LilyGo just released something I’d try, but it sold out almost instantly. Good call on the signal transfer, though I wish better platforms were catching on. Having to use a phone number to sign up kind of defeats the purpose in my opinion. Graphene and Postmark are on my short list of things to try if I end up on an android device.
And I'm in a similar place as you though I've managed to get a bit further. I'm using docker and caddy sidecar with tailscale. I've started with nextcloud aio that had all neatly explained here and I've built on top of that. I've added other services like immich, jellyfin. I'm also testing local reverse proxy with the same domain (*.ts.net) as through the tailscale (to get their certificates), so that I can access them locally without tailscale. I want to use it mostly locally and only occasionally remotely. You might also learn something from this tutorial
OP, I have been facing the same situation as you in this community recently. This was not the case when I first joined Lemmy but the behaviour around these parts has started to resemble Reddit more and more. But we'll leave it at that.
I think I have a solution for you if you're willing to spend $2-$3 a month - set up a VPS and run a Wireguard server on it. Run clients on your devices and the raspberry pi and connect to it.
As for your LAN: from the discussion you linked, it seems that Jellyfin will use the CAs present in the OS trust store. That's not very hard to do on Linux but I guess if you have to do it on Android you'd have some more trouble. In either case, using a reverse-proxy (I like HAProxy but I use it at work and it might be more enterprise than you need, for beginners Caddy is usually easier) will fix the trouble you're having with your own CA and self-signed certs.
I am interested in the attack vector you mentioned; could you elaborate on the MITM attack?
Unfortunately, if you don't have control over your network, you cannot force a DNS server for your devices unless you can set it yourself for every individual client. If I assume that you can do that, then:
- Set up DNS server on Pi
- Set up CA on Pi
- Create root CRT, CSR and server certs from it (bare-minimim setup)
- Copy over this stuff to Jellyfin image/VM, and copy root cert to clients trust store.
- Run reverse proxy in front of Jellyfin and configure the correct IP address of the reverse proxy with an A record in your DNS server.
- Configure reverse-proxy with server/application cert.
- Use RethinkDNS on Android to pass everything through the wireguard server hosted on the VPS, and set private DNS to the DNS server hosted on the Pi.
I think that should do it. This turned out more complicated than I imagined (it's more of a brain dump at this point), feel free to ask if it is overwhelming.
OP, I have been facing the same situation as you in this community recently. This was not the case when I first joined Lemmy but the behaviour around these parts has started to resemble Reddit more and more. But we’ll leave it at that.
I've noticed that behavior is split between communities. Lemmy gets a bit weird because communities are usually hyper-specialized, and sometimes instances themselves cultivate different cultures (e.g. lemmy.ml is usually for privacy enthusiasts, since that's where c/privacy is hosted). That, with the addition of specific idols for each community (e.g. Louis Rossmann for the selfhosted community) affects how each community behaves. That's my theory, anyways.
I am interested in the attack vector you mentioned; could you elaborate on the MITM attack?
Basically the "this website is not secure" popup you see in your browser is sometimes due to the website using a self-signed cert. There's no way to verify that that cert is from the website itself or from an attacker trying to inject their own cert, since there's no CA attached to the cert. If an attacker injects their own self-signed cert, they can use that to decrypt your HTTPS traffic (since your browser will be encrypting using their cert) and then forward your traffic along to the real website so that from your perspective (minus the warning screen) nothing is wrong. I'm oversimplifying this, but that's basically how it works.
Unfortunately, if you don’t have control over your network, you cannot force a DNS server for your devices unless you can set it yourself for every individual client.
I forgot to mention in this post, but because of browser fingerprinting reasons I don't want to use a custom DNS. Thanks for the suggestion though!
I remember you were worried about your ISP messing things up for you, hence the VPN. I would recommend creating a "Virtual Machine" that does all of your downloading to whatever hard drive you're using. That VM can have proton installed. Then, on your regular computer (not within the VM), you can host Jellyfin with no VPN involved, making it accessible at 192.168.0.xx.
I think this hits your goals without needing to expose Jellyfin to the Internet. Plus it has minimal technical complexity. Your downloading traffic will be VPN protected, but Jellyfin will still be accessible to your local network.
edit: You can set up a password for Jellyfin, protecting it from your internal threats.edit2: You can use letsencrypt to create a certificate that picky clients will accept. Buy a domain, any domain, and configure the "A record" to point to 192.168.0.xx (your Jellyfin IP). Then tell your client to go to whatever domain you get, like "luigiliterallydidnothingwrongplzfree.com", then the client will have to use the internet to ask DNS what the IP address is, but after that, it will just use your local network.
edit3: Since you just have the raspberry PI, instead of using a Virtual Machine, you could have 2 separate SD cards. One only has the downloader and VPN installed, the other only has Jellyfin installed (no VPN). Then swap as needed.
One thing that I haven't seen anyone mention yet, in this post or the last one, how to you plan on aquiring videos for your server? If you plan on torrenting, you just have to pay for a vpn, Proton doesn't allow you to make p2p connections like that on a free account
I've been able to use Proton for torrenting, although at abysmal speeds. I don't acquire many new videos, so this isn't an issue quite yet. When I have more money I will absolutely be switching to Mullvad VPN.
Don't use Mullvad for torrenting. They're a great VPN but they had to remove port forwarding so you'll be unable to torrent properly. AirVPN is an alternative that still has port forwarding available.
Here's an idea: on your android device use something like Insular to create a work profile, that way you get its own VPN slot, add your selfhosted-related apps there along with Tailscale. You can keep ProtonVPN on for your other apps, while using TS for your "LAN away from home" stuff. Since Tailscale already encrypts all traffic, you don't have to worry about HTTPS, certs, et al.
THIS
While I would make the modification to use Android's Private Space instead of a work profile (or Shelter instead of Insular), this was such an obvious solution, and I feel stupid for not seeing it. I might use Wireguard instead of Tailscale, I don't know yet, but thank you! Consider yourself an outside the box thinker!
We all got hung up on trying to fix Proton, when Android was the issue here!
and Android devices only have one VPN slot enabled, I cannot use something such as Tailscale for encryption.
I solved a similar situation with a tailscale subnet router . a tailscale subnet router is a tailscale node that exposes the non-tailscale network to the tailscale network. This way I am able to access one of my routers (and its PBX) from all tailscale nodes. The android phone has only tailscale as a VPN. If i pay for mulvad I can have the rest of the traffic go over a mullvad node.
doesn't really help you here though, unless you install protonVPN on the pi and add that as a tailscale exit node.
Edit: Okay, I saw your other post, ignore this answer. It won't work.
Just to give you another way of doing it, I propose using "a third party provider" for your DNS, which you said you didn't want, but since I think it could still work, I tell you how it would work:
Duckdns is a free provider for DNS and let's you create standard certificates via let's encrypt without exposing the rpi.
You can register for free and just input your local IP for the raspberry e.g. at charger8283.duckdns.org
Since the IP is local, no one outside your network can access it, but because the URL is registered globally, you can get a certificate using nginx proxy manager.
This would result in https traffic, that never leaves your local network and is also free.
Maybe self host your own VPN on a VPS and connect the jellyfin server as a client as well as any other devices you want to see that jellyfin server as other clients and configure the VPN server to not override your default routing and to allow clients to see each other? In my head I don't think that would conflict with your protonVPN connection.
Your traffic would be encrypted between devices so I wouldn't say https is nessesary and thus no certs needed.
The rubs that occur to me are that I'm not sure you can do this on a free tier VPS which is the only option I see given your financial limitations. And your devices all need to be able to connect to said VPN.
Edit: Slightly less worse English.
