mirror of
https://github.com/jellyfin/jellyfin.git
synced 2026-01-23 23:20:51 +01:00
[Issue]: Nightly docker container constantly OOM #5435
Labels
No labels
area:database
awaiting-feedback
backend
blocked
breaking change: web api
bug
build
ci
confirmed
discussion needed
dotnet future
downstream
duplicate
EFjellyfin.db
enhancement
feature
future
github-actions
good first issue
hdr
help wanted
invalid
investigation
librarydb
live-tv
lyrics
media playback
music
needs testing
nuget
performance
platform
pull-request
question
regression
release critical
requires-web
roadmap
security
security
stale
support
syncplay
ui & ux
upstream
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference: starred/jellyfin#5435
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @Joly0 on GitHub (Feb 5, 2024).
Please describe your bug
I am running the nightly docker image from lsio, which is based on the latest nightly build and jellyfin (i think its the ffprobe process) is constantly OOM my whole server.
This happens mostly only when i scan my media library. There is nothing really uncommon, just a normal movie library with about 150 movies. Nothing more.
Jellyfin Version
Other
if other:
10.9.0
Environment
Jellyfin logs
No response
FFmpeg logs
No response
Please attach any browser or client logs here
No response
Please attach any screenshots here
No response
Code of Conduct
@joshuaboniface commented on GitHub (Feb 5, 2024):
Without significantly more information about your system and Jellyfin instance, it's impossible to say why this might be happening. Please provide as much detail as possible.
Also, do note that unstable builds have been unreliable for quite some time now, with builds regularly failing as we work on our improved CI; we don't recommend running the unstable builds right now.
@Joly0 commented on GitHub (Feb 5, 2024):
I have tried to add some more information, but sadly i dont have much more.
Here is some part of my syslog, maybe thats any helpful:
Otherwise yes, i have noticed, that nightly builds are quite unstable, but i havent made a backup before going from stable to nightly and now i cant return without setting up everything from scratch, which i cant
@Joly0 commented on GitHub (Feb 5, 2024):
To add some more information about my system: I have an Ryzen 9 7950x with 64GB of RAM running Unraid/Slackware as the host os. Jellyfin is run inside a docker container, provided by linuxserver.io. This container is using the same executables as the official unstable builds, so same ffmpeg, same jellyfin-web, etc.
@Joly0 commented on GitHub (Feb 5, 2024):
I have tried to limit the ram usage of the jellyfin container to 8GB, but that didnt work, because of cgroup limitations of the unraid host. So jellyfin is eating it up all. Also cpu usage is spiking very high
@Joly0 commented on GitHub (Feb 8, 2024):
Btw, is it possible to downgrade from the 10.9.0 unstable version to the latest 10.8.x stable version? I had tried it, had some issues and couldnt proceed. I have read, that there is no option to downgrade again and i would have to start from scratch. Is it somehow possible atleast to remain the users, as i dont want to recreate all my users with their settings, etc.
I dont mind all the other jellyfin settings, but users is annoying, as i would have to give everyone a new password then aswell
@felix920506 commented on GitHub (Feb 8, 2024):
@Joly0 No, there is no way to downgrade from unstable to 10.8.x
@Joly0 commented on GitHub (Feb 8, 2024):
Ok, was able to downgrade and keep my users :)
@solidsnake1298 commented on GitHub (Feb 10, 2024):
The LSIO nightly definitely behaves differently than the official unstable image. The number of ffmpeg processes is supposed to be limited to 2x the CPU threads, IIRC. The official unstable image behaves correctly, 16 ffmpeg processes on my 8 thread CPU, and overall system memory usage peaked around 4.4GB. The LSIO nightly spawned over 100 ffmpeg processes at one point, and memory usage continuously climbed to over 12GB before I stopped the scan.
@Joly0 commented on GitHub (Feb 13, 2024):
Hey @solidsnake1298 i have asked on the linuxserver.io discord server, and this is what they said:
Can you say something about that?
@solidsnake1298 commented on GitHub (Feb 15, 2024):
Double checked that the LSIO image correctly detected CPU core/thread count (it did). So not sure what else to say beyond that SOMETHING is broken. Because in addition to the ffmpeg OOM issue, the Books library type in the LSIO image is straight up broken. No metadata providers and it doesn't track playback progress like the official image does.
@aptalca commented on GitHub (Feb 16, 2024):
Linuxserver dev here. . .
We pretty much install jellyfin debs from the jellyfin apt repo in a fairly vanilla ubuntu jammy image, and start the server as an unprivileged user with the data folders set to /config. We do no modifications of upstream code. We don't touch anything related to jellyfin spawning ffmpeg/ffprobe processes, and we don't touch anything related to jellyfin's library handling.
Our image is as close as you can get to running official jellyfin binaries on an Ubuntu Jammy machine or VM.
I understand upstream devs often refrain from troubleshooting issues in downstream projects (which is often fair).
@Joly0 I recommend that you install jellyfin unstable from the jellyfin apt repo in an ubuntu jammy VM (instructions in our Dockerfile), reproduce the issues so you can submit the bug reports to jellyfin without any relation to the linuxserver docker image.