[Support] ich777 - Jellyfin AMD/Intel/Nvidia [DEPRECATED]


Recommended Posts

1 hour ago, luxinliang said:

But I've met some problems recently that I can't get the images of episodes. Metadata seems ok and the original jellyfin also works properly. Log is attached below, can you help me fix the problems? Thx a lot!

This is a well known issue with Emby and also Jellyfin, sometimes the pictures can't be fetched.

 

I think this is happening only for a few movies/series or am I wrong?

 

What do you mean exactly with original Jellyfin?

Link to comment
9 hours ago, ich777 said:

This is a well known issue with Emby and also Jellyfin, sometimes the pictures can't be fetched.

 

I think this is happening only for a few movies/series or am I wrong?

 

What do you mean exactly with original Jellyfin?

Thx for your reply. The original jellyfin means the docker image jellyfin/jellyfin.

 

In fact, the container based on your docker image can't get any episodes' images (not just a few) and keep throwing errors "MediaBrowser.MediaEncoding.Encoder.MediaEncoder: ffmpeg image extraction failed for file" and “MediaBrowser.Providers.TV.EpisodeMetadataService: Error in Screen Grabber”.

 

When using the docker image jellyfin/jellyfin with same version 10.7.6 don't come across this problem. I even try setting the same parameters and passing the same config, cache, media folders for both of them, while the original jellyfin works fine, your jellyfin still pops out the same errors.

 

Can you help me with that?

Link to comment
2 minutes ago, luxinliang said:

In fact, the container based on your docker image can't get any episodes' images (not just a few) and keep throwing errors "MediaBrowser.MediaEncoding.Encoder.MediaEncoder: ffmpeg image extraction failed for file" and “MediaBrowser.Providers.TV.EpisodeMetadataService: Error in Screen Grabber”.

This is really strange... I'm not experiencing this issue.

Can you try to delete the container entirely (but don't delete the folders for the container on the host) and then add it back with the button "Add Container" ont he bottom on the Docker page and choose the template from the drop down, so nothing has to be configured again.

 

Do you use a hardware transcoder?

 

This error indicates that something is wrong with the extraction from an image from the video file...

Link to comment
2 hours ago, ich777 said:

This is really strange... I'm not experiencing this issue.

Can you try to delete the container entirely (but don't delete the folders for the container on the host) and then add it back with the button "Add Container" ont he bottom on the Docker page and choose the template from the drop down, so nothing has to be configured again.

 

Do you use a hardware transcoder?

 

This error indicates that something is wrong with the extraction from an image from the video file...

I did try to re-add the container, but that didn't work.

And I also try using the QSV, VAAPI and none hardware transcoder, while ffmpeg still throw the same error in all of them.

Link to comment
2 hours ago, ich777 said:

This is really strange... I'm not experiencing this issue.

Can you try to delete the container entirely (but don't delete the folders for the container on the host) and then add it back with the button "Add Container" ont he bottom on the Docker page and choose the template from the drop down, so nothing has to be configured again.

 

Do you use a hardware transcoder?

 

This error indicates that something is wrong with the extraction from an image from the video file...

BTW my cpu is intel J4105 with UHD Graphics 600

Link to comment
2 hours ago, ich777 said:

This is really strange... I'm not experiencing this issue.

Can you try to delete the container entirely (but don't delete the folders for the container on the host) and then add it back with the button "Add Container" ont he bottom on the Docker page and choose the template from the drop down, so nothing has to be configured again.

 

Do you use a hardware transcoder?

 

This error indicates that something is wrong with the extraction from an image from the video file...

Hi again, I may find the solution.

 

I noticed that an error said the cache folder denided to write, so I deattached the cache folder which passed to the docker and deleted "--volume /mnt/cache/jellyfin-cache:/cache". Then everything is back to work, I can grab the images! After that, I also tried to modify the cache folder's permission or create another one, but none of them worked.

 

I think it is a bug, and somehow the original jellyfin don't come across this cache folder issue.

  • Like 1
Link to comment
41 minutes ago, luxinliang said:

I did try to re-add the container, but that didn't work.

What didn't work? What where the exact steps that you did? Hope you don't downloaded it from the CA App because that would overwrite you local template.

 

42 minutes ago, luxinliang said:

And I also try using the QSV, VAAPI and none hardware transcoder, while ffmpeg still throw the same error in all of them.

I tested it now on my system and I can't reproduce this issue.

Everything is generated just fine, found a few issues on the Jellyfin Github issue tracker with the same issue but they are all open ond not solved. Looks like to me that is specific to a few installations from users, even the official Jellyfin container has that issue.

 

41 minutes ago, luxinliang said:

BTW my cpu is intel J4105 with UHD Graphics 600

If you've enabled the iGPU VAAPI should always work, I would always recommend to use VAAPI instead of QuickSync.

A buddy from me has the same system with my Jellyfin container running and I've told him that he should keep am eye on it if he has the same error if he add files.

Link to comment
9 minutes ago, luxinliang said:

I think it is a bug, and somehow the original jellyfin don't come across this cache folder issue.

Do you use the same folder for both instances from Jellyfin on your system?

 

Have you reset the permissions on unRAID or changed disks or something like this?

Has the cache disk enough space on it?

 

As said above can't reproduce this issue, even not on my second test system.

 

Try to set the following permissions for the folder on unRAID:

chown -R 99:100 /mnt/cache/jellyfin-cache

chmod -R 777 /mnt/cache/jellyfin-cache

Link to comment
5 minutes ago, ich777 said:

What didn't work? What where the exact steps that you did? Hope you don't downloaded it from the CA App because that would overwrite you local template.

 

I tested it now on my system and I can't reproduce this issue.

Everything is generated just fine, found a few issues on the Jellyfin Github issue tracker with the same issue but they are all open ond not solved. Looks like to me that is specific to a few installations from users, even the official Jellyfin container has that issue.

 

If you've enabled the iGPU VAAPI should always work, I would always recommend to use VAAPI instead of QuickSync.

A buddy from me has the same system with my Jellyfin container running and I've told him that he should keep am eye on it if he has the same error if he add files.

Well, re-add the container can't eliminate the errors. Also both VAAPI and QSV transcoder work fine, just the images issue.

 

6 minutes ago, ich777 said:

Do you use the same folder for both instances from Jellyfin on your system?

 

Have you reset the permissions on unRAID or changed disks or something like this?

Has the cache disk enough space on it?

 

As said above can't reproduce this issue, even not on my second test system.

Yes, I did use the same folder for both of them, and after creating the cache folder in another disk it worked properly.

 

I think the whole thing is a simple access permissions thing and sorry for taking up your time.

Link to comment
37 minutes ago, luxinliang said:

Yes, I did use the same folder for both of them, and after creating the cache folder in another disk it worked properly.

Never use the same folder for two instances of anything if it's a cache or working directory...

 

Don't know exactly which permissions the original Jellyfin container uses but my containers never run as root by default.

 

So it is now working?

Link to comment
1 hour ago, ich777 said:

Never use the same folder for two instances of anything if it's a cache or working directory...

 

Don't know exactly which permissions the original Jellyfin container uses but my containers never run as root by default.

 

So it is now working?

Got it! It is working right now, thx for your help!

BTW why you recommend to use VAAPI transcoder instead of QSV?

Link to comment
23 minutes ago, luxinliang said:

BTW why you recommend to use VAAPI transcoder instead of QSV?

Because it's basically the same and has no real advantage/disadvantage.

 

The only disadvantage from QuickSync is that it's proprietary and if Intel decides to depricate certain CPU architectures, like it happened lately in the Media Drivers, then QuickSync isn't working anymore like it happened for my test sytem J3xxx series and I wondered why in my Container QuickSync doesn't work anymore after switching my Container to the newest version, where on my i5-10600 system it worked flawlessly now I'm back on VAAPI on both systems. ;)

Link to comment
14 minutes ago, ich777 said:

Because it's basically the same and has no real advantage/disadvantage.

 

The only disadvantage from QuickSync is that it's proprietary and if Intel decides to depricate certain CPU architectures, like it happened lately in the Media Drivers, then QuickSync isn't working anymore like it happened for my test sytem J3xxx series and I wondered why in my Container QuickSync doesn't work anymore after switching my Container to the newest version, where on my i5-10600 system it worked flawlessly now I'm back on VAAPI on both systems. ;)

I see, in my system, when using QSV to transcode hevc videos, it can achieve up to 30 fps more than VAAPI, but the cpu usage is also 10%-20% more.

Link to comment
15 minutes ago, luxinliang said:

I see, in my system, when using QSV to transcode hevc videos, it can achieve up to 30 fps more than VAAPI, but the cpu usage is also 10%-20% more.

This is very unlikely that you get such a huge perfomance boost with QSV.

 

Also keep in mind that even if you enable throttling in Jellyfin that it doesn't work because they disabled it in general because it caused trouble with HW acceleration.

 

Hardware transcoding is really hard to measure with applications like Emby/Jellyfin/Plex, in my oppinion the only real benchmark is how many simultaneous streams you can encode without stuttering.

Link to comment
5 minutes ago, ich777 said:

This is very unlikely that you get such a huge perfomance boost with QSV.

 

Also keep in mind that even if you enable throttling in Jellyfin that it doesn't work because they disabled it in general because it caused trouble with HW acceleration.

 

Hardware transcoding is really hard to measure with applications like Emby/Jellyfin/Plex, in my oppinion the only real benchmark is how many simultaneous streams you can encode without stuttering.

Ok, I'll watch it out, thx for sharing. Also, this is a great docker project😘.

  • Like 1
Link to comment

I seem to be having an issue starting the docker with the extra parameter after the last update.

--runtime=nvidia

When I update and run it the docker fails to start saying the --runtime=nvidia parameter is invalid. Any idea what would cause that?

P.S. Going short of reverting to the last version I did force download and reinstall the lastest docker image without clearing the appdata folder.

P.P.S. It was working wonderfully prior to the update and I do have the Nvidia Visible Devices parameter set using the Nvidia driver plugin.

Edited by xWeegix
P.P.S.
Link to comment
24 minutes ago, xWeegix said:

Any idea what would cause that?

Seems like you have a issue with the Nvidia Driver plugin.

Please make sure it's installed and working.

Try to uninstall it and reinstall it.

If you have it installed for the first time or have reinstalled it on a running unRAID instance always remember that you have to reboot or restart the Docker service.

 

26 minutes ago, xWeegix said:

P.P.S. It was working wonderfully prior to the update

Which update exactly? The last update from the stable container was about 11 days ago.

Link to comment

Ahh yeah so using by brain and some moderately intellegent googling it seems like my Nvidia driver installation broke and the container toolkit isn't working with docker atm. I just reinstalled the Nvidia driver plugin. I should've known when the GPU Statistics plugin was also not working but still installed.
As for the versions it was probably coincidental that I updated jellyfin from 10.7.5 and some other things where things ended up breaking.

 

Thanks for the quick reply.

  • Like 1
Link to comment
10 minutes ago, xWeegix said:

As for the versions it was probably coincidental that I updated jellyfin from 10.7.5 and some other things where things ended up breaking.

Please report back if it is working again...

Have you rebooted or something else?

 

Eventually the package Nvidia Driver package is corrupt, to check it click on the Download button on the plugin page.

Link to comment

I uninstalled Nvidia driver plugin, rebooted, reinstalled Nvidia driver plugin, restarted, then reinstalled GPU Statistics. Downloaded your jellyfin docker and it kept the old template so I added the --runtime=nvidia to extra parameters and the jellyfin docker now boots and nvidia transcoding works. I checked the transcode logs and the GPU stats plugin and found it is all working as expected. :D

 

Quote

Eventually the package Nvidia Driver package is corrupt, to check it click on the Download button on the plugin page.

I did end up checking the driver package with that method and it passed the checksum. I even changed the driver from the most current version v470.63.01 to v465.31 which I knew worked and that didn't fix anything. So the nvidia driver plugin install broke for some reason and a full reinstall of the Nvidia driver plugin was required.

Edited by xWeegix
  • Like 1
Link to comment
  • 3 weeks later...

Hi,

 

I modified the jellyfin-web package to add 60 and 40 mbit bitrate to the 4k transcode option in the qualityOptions.js file, build the deb and installed it in the docker container without error. After this I commited to a new docker image and run a new container with this image.

 

I.stuck at the jellyfin logo and the set up process doesnt start. There is no error in the container log.

 

Did I miss smth here ?

 

Maybe you can add more bitrates in your build in the future.

 

Thx & Greetings

Link to comment
54 minutes ago, Kekskruemel said:

I.stuck at the jellyfin logo and the set up process doesnt start. There is no error in the container log.

Delete the Browser cookies please and try it again.

 

54 minutes ago, Kekskruemel said:

Maybe you can add more bitrates in your build in the future.

I don't compile Jellyfin from source so I can't include more resolutions/bitrates.

I would recommend that you maybe bump and upvote this thread: Click

Link to comment
  • ich777 changed the title to [Support] ich777 - Jellyfin AMD/Intel/Nvidia [DEPRECATED]
  • ich777 locked this topic
Guest
This topic is now closed to further replies.