SOLVED: Jellyfin Server Hangs HW Transcoding

[stnd]

HI, hoping I can get some help or direction with my issue.

Setup is Proxmox (i5-12500 UHD770) -> Ubuntu LXC -> Jellyfin

When hardware transcoding and I try to navigate the video timeline the container CPU  and Memory will max out and hang the system requiring stopping the container and starting it again.

Transcoding Settings
https://imgur.com/a/J3eSGBS

Transcode Log
https://pastebin.com/62nqpArG

/usr/lib/jellyfin-ffmpeg/ffmpeg -init_hw_device vaapi=va
https://pastebin.com/vq60TrXN

What else can I provide to help with this.

[TheDreadPirate]

Couldn't you run Jellyfin as an LXC container directly on proxmox? Seems like an unnecessary virtualized layer.

1) You need to pass your GPU twice. Once from your Proxmox host to Ubuntu, then Ubuntu to LXC container.
2) Select Intel Quick Sync instead of Intel
3) Which image you using? The official jellyfin/jellyfin image? A third party image?

[stnd]

Couldn't you run Jellyfin as an LXC container directly on proxmox?  Seems like an unnecessary virtualized layer.

1) You need to pass your GPU twice.  Once from your Proxmox host to Ubuntu, then Ubuntu to LXC container.
2) Select Intel Quick Sync instead of Intel
3) Which image you using?  The official jellyfin/jellyfin image?  A third party image?

Thanks for responding.

Official Jellyfin is installed using install script in he LXC container running Ubuntu. The vgpu is passed through once. How could I figure out why my CPU and Memory usage goes to 100% and the LXC container hangs until I force stop it. right now it happened again after I tried changing bandwidth options.

[Revv23]

I have same questions a TDP.

Why run a container in a container?

did you switch to QSV?

[TheDreadPirate]

You need to pass the GPU twice. Proxmox > Ubuntu > LXC. Which one did you already do?

[stnd]

I have same questions a TDP.

Why run a container in a container?

did you switch to QSV?

You need to pass the GPU twice.  Proxmox > Ubuntu > LXC.  Which one did you already do?

I am sorry if I am confusing things. Where am I running a container in a container? 

I only have Ubuntu as a Linux container LXC. Once Ubuntu was installed, Jellyfin was installed using the automated script.

Code:

curl https://repo.jellyfin.org/install-debuntu.sh | sudo bash

This is the guide I followed https://www.michaelstinkerings.org/gpu-v...u-uhd-730/ and the transcoding seems to work, but if I mess with the stream and try to change the position in the timeline or bitrate it can cause full CPU/MEM usage and cause the Ubuntu LXC Container to become unresponsive requiring stopping and restarting.

[TheDreadPirate]

My bad, misunderstood your setup. How much RAM does your Ubuntu VM have?

[stnd]

My bad, misunderstood your setup.  How much RAM does your Ubuntu VM have?

Ubuntu LXC
CPU: 4
MEM: 4GB
SWAP: 1GB
DISK: 8GB

[TheDreadPirate]

Jellyfin writes transcodes to disk. Are you running out of disk space? You can change where Jellyfin writes transcodes to in Dashboard > Playback and have Jellyfin write them to a disk with more space.

[stnd]

Jellyfin writes transcodes to disk.  Are you running out of disk space?  You can change where Jellyfin writes transcodes to in Dashboard > Playback and have Jellyfin write them to a disk with more space.

Thank you. I upped the memory to 8GB and the disk space to 32GB and have been messing with the streams and so far so good. That must have been it and would explain why it maxed out resources and froze. I will use it today and if all is good, mark your post and thread solved.