Are these options going to require installation of specialist software and then entering of special configuration parameters on the client computer as rather than just using any standard browser on any internet connected computer and typing yourjellyfindomainname.com ?
You have to port forward a port and setup dynamic dns, for 99% this is a insurmountable difficulty.
The person doing screensharing runs sunshine.exe
they tell their hostname from the watcher, example.com
the watcher runs
if not paired already Moonlight.exe pair example.lan — > (then a number appears, they tell the number to screensharer, who then types this number in https://localhost:47990/pin )
once paired (pairing key exchange is permanent until manually revoked by default) the watcher runs Moonlight.exe --display-mode windowed --absolute-mouse --resolution 1920x1080 stream example.lan desktop
A screen appears on their device, showing the other device, the latency and quality are excellent, you can watch youtube videos, play doom, it’s almost like it’s your own computer, you can fullscreen the window if you want.
Me and my friends, we use an extra computer with sunshine on it. We all connect at the same time to this extra computer and share it, show stuff to each other. It works really great
As for connectivity, there is no STUN/TURN coordination, either you open a port forward on your router or you use a VPN like zero tier, tailscale, hamachi, softether etc…
Here is the latest moonlight client
https://github.com/moonlight-stream/moonlight-qt/releases/download/v6.1.0/MoonlightSetup-6.1.0.exe
https://github.com/moonlight-stream/moonlight-qt/releases/download/v6.1.0/MoonlightPortable-x64-6.1.0.zip
And here is the latest sunshine server
https://github.com/LizardByte/Sunshine/releases/download/v2025.924.154138/Sunshine-Windows-AMD64-installer.exe
https://github.com/LizardByte/Sunshine/releases/download/v2025.924.154138/Sunshine-Windows-AMD64-portable.zip
I use sunshine/moonlight there is nothing as good as that.
Both the server and client run in portable mode
By default it will fullscreen and use relative mouse but use this command line to one-click start moonlight
Moonlight.exe --display-mode windowed --absolute-mouse --resolution 1920x1080 stream yourserver.lan desktop
Only little hiccup, the client need to pair with server before connecting, the client will give you a number and you type it in the sunshine management interface
Check out zoneminder, it’s the defacto open cctv server.
As for anything that offers a service for external access, that’s just setting up dynamic dns and setting up one port forward, I don’t understand why so many people struggle with that that they pay a third party a monthly fee to do it for them.
I got two of those for 100$ USD for the purpose of hosting openwrt in proxmox LXC containers. One thing I noticed is they have no cooling. I put a 10 GBe mellanox card in it plus a very low end radeon gpu and it gets quite hot in there. My recommendation, instead of trying to embiggen it as much as possible, by putting 2 more sticks of ram and the biggest cpu, I would recommend just buying another. The performance boost per dollar isn’t as much as the performance capacity of a second, third or 4th machine.
That sounds interesting but my services are mostly in lxc container with dynamically allocated dhcp addresses (I do not wish to maintain a list of static IPs)
And that makes it hard to start them on demand based on incoming connection like you would an httpd process daemon.
My services are only refferred to by hostnames, for instance vaultwarden.lan, but this address does not resolve when the LXC is offline. I guess I could make a static DNS entries in the local dns server and point those to a LXC container orchestrator to stop and start them as they are accessed.
But also, just leaving them all running and ready to answer queries is, so much more convenient if I have the RAM for it.
But I’m sure there are cases where this is the way to go, but I don’t know how to start LXC containers based on sockets without creating static leases and static dns entries, I really like how IP address and DNS names are completely automated on my network !
I have a G3 and, this might be counter intuitive, but buy an older model that uses DDR3 ram, old ram like that is obsolete and no one wants it, but more ram is more important in self-hosting than fast ram. Of course that limits how much you can do at once, but if you only have 8gb of ram instead of 32 or 64, the lack of ram is going to be very limiting in the sense that you’ll have to run very few things at one time and that becomes cumbersome to manage services “on demand”, starting them and shutting them down to save on ram, instead of letting them all be ready and idle.
I had something similar when I used XMBC (I think it’s called kody now) and it seems it was because there was some confusion with the scrapper to imdb and it needed some manual override file. You had to create a .nfo file or something like that, that told the scrapper the right imdb number or other exact identifier for it. I wouldn’t be surprised if the people at plex have some way to know of all the custom fixes people do and with that, when you have some issue recognizing some content, they can figure out what most people do.
I would be surprised if jellyfin has any way to to know what the users are doing so that wouldn’t work
What do you mean with glitching UI ? Is it more than just not responding while scanning ?
I haven’t installed jellyfin, I’m never installing plex, I would just use samba shares over vpn instead of that.
I wanted to know what’s the problem with jellyfin ?
Is it more than the weekend-destroying linux-jankiness ?
Because I can deal with that
Actually they are a reduction in complexity, yes I am not using most of the features, they run in L2, but their backbone runs off a x86 single board computer and they run a mostly hardware agnostic OS (Sonic).
This is what I mean by a reduction in complexity, it’s basically running debian os with pcie switchdev interfaces on a PC. It’s familiar and stable, not locked in to proprietary hardware, they’re cheap and plentiful
What are the 3 biggest shortcomings of Jellyfin ?
I get what you mean, I only use L3 top-of-rack data center switches, what a bunch of amateur peasants !
You can do data scrubbing with PAR2 or filesystem level with btrfs on top of LVM (or even in a traditionnal partition)
I think you can ram cache with bcachefs or a ramdrive, and unless you’re in a VM then your file system driver would already do file caching in ram ?
Well, it taints the kernel, probably runs some background processes, it should be removable ?
Except having slightly better deduplication, I don’t the see what justifies the extra complexity and living under the bad aura of Oracle. LVM does almost everything ZFS does, it’s just less abstracted, which I like actually because I want to know on what hard drive my stuff is, not some mushy file cloud that either all works or is all gone.
Is there any way to remove ZFS and Ceph, they cause errors and taint the kernel
Enable hugepage allocation, it will deduplicate memory chunks and save you lotsaram Especially good with an hypervisor desktop
When I last tried to setup openwrt+iPXE I kept ending up on their forum but I don’t get what are they actually for ? Is that just a netbootable disk imaging thing ? Live OS netbooting ? Installer netbooting ?
Is it like iVentoy the non-open source netbootable version of ventoy ?