It seems with every qnap firmware update that hardware transcoding breaks.
I have both Nvidia packages installed on my qnap (uninstalled and reinstalled them after today’s firmware upgrade).
Manual validation with the GPU enabled ffmpeg works great and I see the GPU usage. But the Plex transcoder just shi*s itself.
I updated the plex.sh script to include the Nvidia paths for LD_LIBRARY_PATH and PATH so that it’d use ffmpeg from the Nvidia packages, but then Plex just errored at startup.
What is the resolution to this problem? And why do I have to fight this Plex/qnap issue every month or two? I’ve paid them each a lot of money (lifetime Plex plus subscriber) but feel like I’ve wasted my money with both vendors every time this problem comes back.
I will need server DEBUG logs ZIP to see what you’re seeing and diagnose.
Do not edit Plex.sh. LD_LIBRARY_PATH is not used. RPATH is used.
It’s been working perfectly, without alteration, since its release several years ago.
Plex.sh checks the system for each of the following as it starts Plex.
– Viable GPU
– QTS drivers installed
– GPU card in “QTS mode”
When all conditions are satisifed, PMS is informed the GPU is present and can be used.
This is what our lab machine (TS-877 with GT-1050) is currently showing.
I enabled debug logging, stopped plex, purged the logs, started plex.
This apparently made Plex Web have a hissy. When I first started a playback of a 4k movie it was automatically transcoding it to some other (SD) resolution and there was plenty of GPU usage happening. Seemed promising!
But then I did a shift-reload in Plex tab and started the playback over. At that point I was then able to select the quality again. I picked some 1080p quality and it gave me the busy circle and then a blank screen in the player.
The “crash” I’d seen yesterday came back – seeing:
WARN - [Req#124b/Transcode/x34cak4377b3ej1cbaycwb60] Transcode runner appears to have died
in the logs and playback ceases.
I have the logs - they seem safe enough to attach here but figure I’d ask before doing so.
Hi! @ChuckPa
I have a issue in last firmware 5.1.5.2679 build 20240219 i upgrade it and also automatically its NvKernelDriver also update to last, i dont remember well version number but It Is new.
I check that plex reorganize gpu inside plex and in container → /dev/dri/renderD129 quadro p2000.
But when i doing a transcoding in plex docker it crash and fail error and finally CPU trascode it but not gpu.
I check to unistall NVIDIA GPU Driver 5.1.0.0822 and fresh install Nvidia gpu Drive and reboot but without luck. (I use qts mode)
Finally I revert firmware to 5.1.5 20240116 and NvKernelDriver 5.1.5.2645 and It is ok with pms-inc plex image and also with plex from linuxserver image
I think that its a fault from new NvKernelDriver and outdated drivers from 2022, but in last firmware dont trascode.
Aside from my local testing, one of my friends tried to stream a movie and I can see that he had to restart playback somewhere around 50 times.
Several crashes of the transcoder will be in the logs:
Plex Media Server.2.log:Feb 29, 2024 02:13:32.102 [140176536103736] DEBUG - Jobs: '/share/CACHEDEV1_DATA/.qpkg/PlexMediaServer/Plex Transcoder' exit code for process 1180 is 1 (failure)
Plex Media Server.2.log:Feb 29, 2024 02:13:49.576 [140176536103736] DEBUG - Jobs: '/share/CACHEDEV1_DATA/.qpkg/PlexMediaServer/Plex Transcoder' exit code for process 1272 is 69 (failure)
