• 2 Posts
  • 17 Comments
Joined 2 years ago
cake
Cake day: October 18th, 2023

help-circle
  • Depends on your threat model, but you’re probably fairly secure from remote unauthorized access right now.

    Given that I’m American, I would put the *arr stack behind a dedicated VPN container like gluetun and set Gluetun up using a “no logs” VPN.

    For remote access, Tailscale can probably get around that double NAT. If you have it on your devices as well as your server, you won’t necessarily need to expose anything publicly.

    If that’s not an option, you could set up an external VPS to run a reverse proxy (Caddy perhaps) and use the Tailscale connection to connect the VPS to your home server. There are fully self hosted ways to do this (Headscale comes to mind), but Tailscale is how I personally would solve this.





  • It’s doable. I personally run my Jellyfin instance publicly available and there’s maybe 3 people who use it regularly. With my internet connection, WAN side users are limited to about 720p but I’ve had the 3 of us all playing different media at the same time on occasion. The main limiting factors on the number of simultaneously active users is how much upload bandwidth you have and how quickly you can transcode video files. Any 10 year old box will be able to handle 1 or 2 users at a time provided it doesn’t need to do a bunch of transcoding. If your building a box, would use a 11th or 12 gen Intel or if you must go AMD, have a graphics card to handle the transcoding. The “build a box” route can probably handle 4 or 5 simultaneous users, possibly more depending on your hardware choices. The main limiting factor in that case would be your upload.





  • When I was first playing with NC I was using a RPi3 with an external SSD for a drive. Performance was pretty good, but as soon as I tried the same setup in a VM, the performance tanked. The only way I found to avoid the performance penalty was a manual install like it was bare metal, which I didn’t really want to do. My experience with such setups is that they tend to be brittle.

    My understanding was that the performance penalty was caused by the chain of VMs. Proxmox --> Ubuntu VM --> Docker. I don’t know enough about it to say for sure.


  • My NextCloud is running on an old desktop that’s been repurposed into a server. The server is running Proxmox, and NC is running in docker directly on Proxmox using the nextcloud-aio image.

    Found that had better performance than running it in a VM and was less headaches than the other install options.

    I keep thinking about moving it to dedicated hardware, say some sort of mini pc, but it hasn’t been a high priority for me.