- Solved
- Annoyance
I use unRAID 7.2.2 and discovered an issue with transcoding a video with Emby server (in a docker) to Emby for Android/Emby vor Android TV. After a while the playback stops. This happens when QSV (Quicksync) is activated. When I deactivate it in Emby and only use VAAPI this will not happen.
So far I understand this was an issue with older kernel versions used by unRAID, but shouldn't be an issue with 7.x.x.
The Emby log looks like this, when the issue occurs:
20:54:26.110 [hevc_qsv @ 0x11aebc0] Error during QSV decoding.: unknown error (-21)
20:54:26.110 Error while decoding stream #0:0: Unknown error occurred
20:54:26.211 [hevc_qsv @ 0x11aebc0] Error during QSV decoding.: unknown error (-21)
20:54:26.211 Error while decoding stream #0:0: Unknown error occurred
20:54:26.314 [hevc_qsv @ 0x11aebc0] Error during QSV decoding.: unknown error (-21)
In the unRAID log file:
[ 1590.375486] i915 0000:00:02.0: [drm] Resetting vcs0 for preemption time out
[ 1590.375565] i915 0000:00:02.0: [drm] ffmpeg[129266] context reset due to GPU hang
[ 1590.431441] i915 0000:00:02.0: [drm] GPU HANG: ecode 12:4:28fffffd, in ffmpeg [129266]
[ 1596.519451] i915 0000:00:02.0: [drm] Resetting vcs0 for preemption time out
[ 1596.519533] i915 0000:00:02.0: [drm] ffmpeg[129266] context reset due to GPU hang
[ 1596.556176] i915 0000:00:02.0: [drm] GPU HANG: ecode 12:4:53fffeff, in ffmpeg [129266]
[ 1602.599417] i915 0000:00:02.0: [drm] Resetting vcs0 for preemption time out
intel_gpu_top is installed.
I also found a hint to where the problem lies:
https://github.com/jellyfin/jellyfin/issues/11062
It points to the kernel (Jellyfin is based on emby).
NOTE:
Fixed in 7.3.0 Beta 2