Emby extrahiert nicht alle Kapitel-Thumbnails

[b0mb 35]

so here is the latest Logfile....

the problem is stil present.... sometimes all thumbs are extracted and sometimes only 3 of 25

maybe it would be worth a tray to force thumb extraction for a single movie but how to do?

embyserver.txt

[b0mb 35]

With latest Update the Problem is still present.

Btw trying Plex doesn´t have any trouble with extracting.... ^^

[Luke 37056]

@softworkz will take a look at this. Thanks.

[softworkz 3335]

On 10/14/2022 at 6:54 PM, b0mb said:

On 10/14/2022 at 5:53 PM, softworkz said:

/mnt/4K/!4K/

Where are your files stored, what kind of storage is this?

/mnt is where all my folders for the shares of my unRAID server are located

/mnt ist completetly passed through to the docker container

Can you please try to create a library inside the Docker container, e.g. at /test/mytestlib

Then add some files to it which didn't show proper extraction and see whether the problem reproduces there (logs please).

Thanks

[b0mb 35]

10 hours ago, softworkz said:

Can you please try to create a library inside the Docker container, e.g. at /test/mytestlib

Then add some files to it which didn't show proper extraction and see whether the problem reproduces there (logs please).

Thanks

i can only passthrouh a path from the host into the container as unraid creates an image file for the container which is limited to 10GB in my case

[softworkz 3335]

41 minutes ago, b0mb said:

i can only passthrouh a path from the host into the container as unraid creates an image file for the container which is limited to 10GB in my case

Shouldn't that be sufficient for 1 file?

[b0mb 35]

17 minutes ago, softworkz said:

Shouldn't that be sufficient for 1 file?

as the problem only belongs to 4k files and the 10GiG are shared with the other containers i am running simply no ....

[softworkz 3335]

11 minutes ago, b0mb said:

as the problem only belongs to 4k files and the 10GiG are shared with the other containers i am running simply no ....

Alternatively in a folder on the host file system (but not any unraid share) with Emby installed on the host system directly rather than docker...

Edited November 3, 2022 by softworkz

[b0mb 35]

1 minute ago, softworkz said:

Alternatively in a folder on the host file system (but not any unraid share) with Emby installed on the host system directly rather than docker...

i won´t evey try to install emby directly on my unRAID system as it´s not made for....

I´ve already opened a thread in another forum where are many users using unRAID and Emby....

I´ve also tried linuxserver.io Container and Jellyfin and all do have the same problem like emby which is appearing randomly.... Plex is working like a charm but as far as i´ve seen in the debug.log they don´t use ffmpeg for extracting the thumbs 

[softworkz 3335]

The point is: you're the only one reporting that issue. My suspicion is that it's related to I/O in some way, maybe related to seeking (random file access).
The only way to circle down such issues is differential diagnosis, which involves and requires trying out things that are different and opposite to what you have right now. 

That's all I have to offer in the context of Emby. As it seems pretty clear that the issue is not related to Emby at all,  you might want to perform some tests which involve accessing the same file in parallel from multiple sessions which also request different parts of that same file.

The problem COULD be something like that the mounted storage fails to perform seeking and the mount client is forced to start receiving the file from start on each seek. Old SMB implementations were sometimes suffering from this alone or when in combination with parallel access requests, etc etc.

[Teddyknuddel 73]

Also ich hatte ein ähnliches Problem - bei mir wurden die Thumbs aber erst gar nicht erstellt. Auslöser war, dass ich mit einem externen Editor, weiß den Namen leider nicht mehr, die Videobeschreibung, Bilder usw. bearbeitet hatte. Abhilfe war, dass ich alles im Videoverzeichnis bis auf das Video selbst gelöscht habe, und in ein neues Verzeichnis verschoben hatte. Dabei wurden die Kapitelbilder dann auch sauber neu eingelesen.

[softworkz 3335]

May sound similar, but it's unrelated.

[Teddyknuddel 73]

3 minutes ago, softworkz said:

May sound similar, but it's unrelated.

Und was hat man zu verlieren, wenn man meine Methode versucht?

[softworkz 3335]

3 minutes ago, Teddyknuddel said:

Und was hat man zu verlieren, wenn man meine Methode versucht?

Nothing to lose - but nothing to win.

You named what caused the problem in your case:

37 minutes ago, Teddyknuddel said:

Auslöser war, dass ich mit einem externen Editor, weiß den Namen leider nicht mehr, die Videobeschreibung, Bilder usw. bearbeitet hatte.

The OP didn't do that.

You said:

40 minutes ago, Teddyknuddel said:

bei mir wurden die Thumbs aber erst gar nicht erstellt

which wasn't the case for the OP.

I'm afraid, but I don't see any similarity.

[b0mb 35]

@softworkz

do you have any Explanation why this only happens to 4K Stuff?

[softworkz 3335]

5 minutes ago, b0mb said:

@softworkz

do you have any Explanation why this only happens to 4K Stuff?

It's unlikely that it's really due to 4k resolution. Much more likely it's due to the fact that 4k videos are generally larger in size.

[softworkz 3335]

...other peripheral causes are possible: 4k videos are often HEVC and HEVC often has larger periods without key-frames, which means that it might cause more seeking or seeking over larger distances.

Maybe, somewhere in the whole process, there's a parameter or value for which a threshold exists (maybe due to a bug or similar) and things go wrong once that threshold is exceeded. It's just a guess, though.

[b0mb 35]

my complete metadata is outsourced to a share. Would it be worth a try to delete this metadata and rebuild the database?

[softworkz 3335]

No. Such attempts are not effective at all. 

You need to make small and targeted changes and start form a completely different base.
Forget about your existing library - create a new one where everything is different with just 1-3 videos and on a different storage.

[softworkz 3335]

10 minutes ago, b0mb said:

my complete metadata is outsourced to a share

If you mean the main emby metadata folder, then configure your new library to store metadata inside the library - just as an example. Do everything differently for your new test library.

The desired result is - of course - that the problem won't happen there. In that case, you would have 

• a working case
  and 
• a non-working case

and you can assimilate the two subsequently until you find the culprit that makes the difference.

But when the suspicion is already about file IO and you create the new library just on another unraid volume, you're dropping chances for achieving a "working case" situation

[b0mb 35]

new library is not the problem but not inside the container ^^

[softworkz 3335]

Just now, b0mb said:

new library is not the problem but not inside the container

at least not on an unraid volume. Maybe a folder on the host system disk.

[b0mb 35]

nah..... that´s not a good idea... unraid isn´t designed for that

[pünktchen 1258]

17 minutes ago, softworkz said:

at least not on an unraid volume. Maybe a folder on the host system disk.

Or a usb stick maybe.

[softworkz 3335]

Just now, pünktchen said:

Or a usb stick maybe.

That's not a good baseline, as it's about processing 30-50 GB files.