I tried “all” settings. I tried reading the logs and norhing helped. I had many months without issues. How do you guys debug it? The logs aren’t too helpful for me and there are too many settings for me to play with. I’ve got an intel GPU with eirher QSV or VAAPI. Disabling does not work either. CPU is normal, GPU is normal. RAM is not full.
Edit:
- one drive was failing. SMART showed a dying drive. I replaced it and everything is smooth again. That is why I couldn’t find anything within the logs. And until today I didn’t know how to read io properly and check for smart status.
Things to watch out for
- cpu usage
- gpu usage
- right gpu tools. Properly passed to container. Properly used by jellyfin.
- proper transcoding settings with the gpu within jellyfin
- ram
- storage of drives
- io write/read speed
- jellyfin folders on ssd, media on hdd
- healt / smart status of drives
- lan bottleneck


Just start listening to dubstep and you’ll stop noticing 😆.
Maybe run lm-sensors and make sure the CPU/GPU isn’t being thermothrottled? I’d usually look at dmesg and look for red stuff. Any hardware issues are usually pretty obvious.
Try other apps. If you youtube or VLC behaves the same, the problem may be outside of jellyfin. If not, it narrows it down.
If could even be the server not being able to transcode in realtime. Try watching a file known to already be in a suitable format. It should direct stream and be much less load on the server. I’ve seen server encode CPU saturation and it does kinda look the same as client decode stutter. If it’s the server, you’ll probably see the same stutter from another device such as a phone.
A disk was failing, I updated my post
Ahh. That’s usually among the red stuff in dmesg. I glad to hear you solved it, but a failing hard drive is a pricey thing to endure these days.
The dubstep joke was a good one