2024-09-15, 04:03 PM
Hi all,
using Jellyfin since the very beginning 5y ago now, and it's the first time I've a problem like this. Using 10.9.11 in an lxc container on proxmox, working flawlessly since a couple of years with qsv hardware transcoding.
Watched a movie yesterday evening with no issues. Went watching a video this morning and I found it quite sluggish. I also begin gettings alerts from checkmk of high memory and high CPU usage.
By loggin in I see tens of jellyfin threads in htop. Also a bunch of "thread pool starvation".
The high CPU doesn't stop if I stop playing.
After restarting the container, same things happen, as soon as I play something everything crawls.
I enable debug logging and I copied the logs here from the start of the playback: https://pastebin.com/PqCDk4bJ
I don't see anything weird but I can't compare from the log of the 'normal' condition.
I avoided to copy the hundreds of "AuthenticationScheme: "CustomAuthentication" was successfully authenticated", the "<item name> has no parental rating set" and a couple of very long sql queries (long in terms of characters. Execution time was udner 50ms, getting worse as the CPU was getting pegged) that were polluting the logs.
Any hints on what might cause this?
using Jellyfin since the very beginning 5y ago now, and it's the first time I've a problem like this. Using 10.9.11 in an lxc container on proxmox, working flawlessly since a couple of years with qsv hardware transcoding.
Watched a movie yesterday evening with no issues. Went watching a video this morning and I found it quite sluggish. I also begin gettings alerts from checkmk of high memory and high CPU usage.
By loggin in I see tens of jellyfin threads in htop. Also a bunch of "thread pool starvation".
The high CPU doesn't stop if I stop playing.
After restarting the container, same things happen, as soon as I play something everything crawls.
I enable debug logging and I copied the logs here from the start of the playback: https://pastebin.com/PqCDk4bJ
I don't see anything weird but I can't compare from the log of the 'normal' condition.
I avoided to copy the hundreds of "AuthenticationScheme: "CustomAuthentication" was successfully authenticated", the "<item name> has no parental rating set" and a couple of very long sql queries (long in terms of characters. Execution time was udner 50ms, getting worse as the CPU was getting pegged) that were polluting the logs.
Any hints on what might cause this?