MTU LUG Wiki

Minutes 2025-10-23: Difference between revisions

Page Discussion

Minutes 2025-10-23 (view source)

Revision as of 03:45, 24 October 2025

549 bytes added ,  24 October
m
no edit summary
No edit summary
mNo edit summary
 
# attendanceAttendance dropped off this week
## alwaysAlways happens around this time of year
## Same thing happened at the RedTeam meeting which happened the hour before
# Lots of people run Jellyfin servers in their homelabs
# Jellyfin [https://jellyfin.org/] by Allen!
## Over-the-top media services
## If you’ve used Netflix (and other streaming services) you’re probably used to to navigating these
## Hardware transcoding
### Typically used to downscale media from the base resolution to clients (e.g. 4K to 1080p)
### Dane cameo for anecdote about audio?
#### Dane busy this week :(
#### Rozsa show aboutwith jazz
### x264 most supported but largest filetype
### AV1 smallest filetype but rarely supported (hardware decoding-wise) and computationally intensive
## Our jellyfinJellyfin server
### Dell R730
### NVIDIA 1060 and 1070
### RAID 1 MDADMmdadm on two 500GB root disk SSDs
#### ext4 as filesystem
### 14x 1.8TB 2.5” HDDs in a RAIDz2 (“RAID60”) ZFS pool with 16TB of total space
#### probablyProbably overkill for our use
#### Can download blender media at 4K res and do local LLMs for AI research which take a couple hundred gigs each
## Containers
### Podman can be used to manage services and allow updating without needing to update entire server or take down everything (e.g. like updating debian and needing to reboot to update kernel, and take down all running services)
### Our hope is to use for jellyfinJellyfin and the frontend/backends for LLMs
## Jellyfin and NVIDIA setup
### Jellyfin hardware transcoding usage
### hadHad to patch NVIDIA driver with a script off githubGitHub to enable more than 2 transcodes per GPU [https://github.com/keylase/nvidia-patch]
#### NVIDIA moment [https://developer.nvidia.com/video-encode-decode-support-matrix#Encoder]
#### NVIDIA moment
## rFFmpeg [https://github.com/joshuaboniface/rffmpeg]
## RFFMPEG
### remoteRemote ffmpegFFmpeg over a network, allows us to add GPUs to other servers and the jellyfinJellyfin server to use them remotely if we ever need
## *arr suite
### justJust a suite of tools that take files and rename them for more easy indexing via Jellyfin
#### Can be used in combination with our WIP automatic Blender movie downloader (coming #soon hopefully) to get auto-added to jellyfinJellyfin
## Also allows auto-downloading media if you hook it up to things like a torrent tracker
### usuallyUsually used for people running pirated media, not relevant for our case
## JellyseerJellyseerr [https://github.com/seerr-team/seerr] manages Jellyfin accounts and media requests/storage/quality
## Blender Open StudioMovies! [https://studio.blender.org/films!/]
### By Blender StudiosStudio
### Blender is a FOSS 3D modelling software
### They make Public Domain short films to showcase Blender’s current capabilities
#### Akin to Pixar’s short films
### They even used to allow you to download all the 3d3D assets used in the film
#### Were also Public Domain, so you could use them in your own projects for whatever you want
### Our hope is too ‘mirror’ the films (so they’re not only on YouTube and other proprietary video hosting sites)
#### Like our Linux mirrors server forthat hosts open source Linux programs, but this one is for open media
## SelfhostingSelf-hosting Jellyfin at home is also possible, and is what most people do
## AssigningHanding out Jellyfin accounts to usersLUG members
## Time to stress test the GPUs
## Had to install a Stream Limit plugin [https://github.com/JellyboxAD/Jellyfin.Plugin.StreamLimit] on Jellyfin as a preventative measure
[[Category:Meeting Minutes]]
Retrieved from "https://lug.mtu.edu/wiki/Special:MobileDiff/7987"