Regression: 2.00.21 change to VAAPI causes Transcode plugin failure, reverting to 2.00.20.1 restores functionality #795
Comments
|
Also somewhat related to what I can only assume is the |
|
Yeah the only thing that changed with the nodes are the container username and FFmpeg going from v5 to v6 so most likely driver related as you mentioned. From what I read there’s no change to FFmpeg argument syntax etc for v6. |
|
I went ahead and created #797 just to not cause conflicting info. |
|
i am having the same issue of failed QSV encoding.
|
|
Having same problems on Intel QSV running in LXC container latest 6.2 kernel - 20.20.1 work just fine - 20.21 brake things exactly as above. |
|
Same for me - QSV using docker and the 2.0.21 fails (error as above), reverting to 2.0.20 is fine. Not withstanding this small blip, I do like tdarr - just amazing! |
|
Which intel gen are you using? |
|
Intel GPU top returns 'Intel Coffeelake (Gen9)', though its an i5-8259U. Only been using tdarr for 2 weeks with a test install happening to use .20, so when latest didn't work thought it was me - and haven't kept any logs - sorry. However, I do remember googling the message from above '... filter 'Parsed_null_0' and...' |
|
Thanks, I've got an 8250u coming next week which is slightly behind yours but should be able to fix the issue. As I mentioned, the .21 Node changes were pretty much only FFmpeg v5 to v6 updates so will see what the problem is. |
For me, i am using 12th Gen i5-1235u |
|
I'm on 12th gen 1240p |
|
Issue is due to the last arguments Not 100% sure what the solution is. Running |
|
say with updated our version to latest but want to keep the ffmpeg v5 that is working for qsv. can we just replaced the "ffmpeg.exe" that from old version that we have backup? i run both docker version and windows version |
|
yes that should work fine |
|
Is there an update to this (or potential fix/workaround), I currently ran into this with 2.00.23. |
|
@Fribb do you need to encode 10 bit video? If not then you can disable |
Yes, I encode everything to 10bit. Though this happens with both my own plugin as well as the "Tdarr_Plugin_MC93_Migz1FFMPEG" plugin (but both use 10bit) |
|
Tracked down the issue and sent a PR. For anyone else that wants to upgrade sooner rather than later, copy the plugin local and replace |
Describe the bug
Updates made in v21 seem to have broken my forked custom version of bsh1_Boosh_FFMPEG_QSV_HEVC. As stated in seems changes made in most recent build have created problem for QSV transcode via bsh1_Boosh_FFMPEG_QSV_HEVC plugin. Issue confirmed in Log below shooing failed transcode with v21, then reverted Docker container back to v20.1 and successfully transcoded the SAME file on same server with NO other changes. Unfortunately ran out to time to troubleshoot tonight so restarted all failed queues with previous version. If need be I can assist in troubleshooting but may need to see if any files are still available for process when you are ready.
To Reproduce
Steps to reproduce the behavior: Update Docker container on working system with valid transcode jobs, and now jobs fail.
Expected behavior
Jobs continue to process....
Screenshots
NA
Please provide the following information:
Failed: (running latest aka v2.00.21)
SAME media succeeding on same system with previous Docker release: v 20.1
-Worker error [can be found on the 'Tdarr' tab by pressing the 'i' button on a failed item in the staged file section or in the transcode error section at the bottom]
The text was updated successfully, but these errors were encountered: