New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Issue]: Playback ends prematurely when transcoding #6957
Comments
Here is another run with debugging turned on.
|
What client are you using that causes this? |
androidtv 0.12.3 |
On the client I get messages like these when it ends.
|
I'm having the same issue when playing unsupported media on my Chromecast Ultra. It seems like Jellyfin creates a single .mkv file in the /transcodes/ directory, and when the transcode/remux is complete, TranscodingJobHelper deletes the file which causes playback to stop.
|
I have this issue too! The transcode ends and is then deleted |
I've seen this issue a bunch when transcoding to progressive mkv on the android tv client. For me, it usually happens between 1-5 minutes after transcoding completes Edit: |
So might there be a temporary fix to this? |
I've been having this same issue on iOS and Roku clients after updating my server to 10.8.0 and enabling NVENC hardware acceleration. From reading this, I assume it has more to do with the latter than the former. |
I seem to be having the same issue now that I have upgraded to the Jellyfin server 10.8.0 release. When playing the same file, it seems to happen at the same point in time, meaning it's repeatable. I know it's not that the file is bad because I'm not getting any indication in the ffmpeg logs of any errors, and it also seems to happen with files that previously worked on 10.7.0 with several clients. It occurs with the Web player on Firefox on the desktop, as well as in the Jellyfin Android mobile app. Of the files I have tested, it only seems to be happening to files that use ASS subtitles (which ExoPlayer doesn't seem to support very well yet, so they always get burned into the video stream), whereas files that require transcoding but don't have subtitles don't seem to have a problem, which is odd. It also seems to think the file has finished playing successfully, as the web interface reports the video as "watched", even though it jumps from about halfway through a 24-minute episode straight into the next episode. What's also odd is that although I don't see this issue when using ExoPlayer on either the Android mobile or TV app, I do experience the app pausing and loading forever after a few minutes if left alone and playing the same files. My phone is a Pixel 6 with Android 12, my TV is a Hisense HiSmart A4. My host environment is a Debian 11 machine (from an OpenMediaVault 6 image, to be precise), with a native install, and hardware acceleration turned off (though I have turned it on with NVENC and a GeForce 750 Ti and no throttling, to compare, and it made no difference). |
So with #7911 merged, would this be fixed in the nightly? |
Most of my library gets transcended because people watch on web browsers and now nobody can watch to the end of anything. Rolling back until this is fixed. guess its not 10.8 stable just yet. |
I've run into a similar issue (not 100% sure it's the same one): movies on the Android TV client would stop after a few minutes of playing (as if it was buffering, but it would never recover), even though transcoding would continue (and exceed 1x rate). Tested the latest commit from the 10.8.z branch today (specifically fd4ffc6) and it seems to not happen anymore. |
I have been running a build I made from 10.8.z since 6/14 and it has been working for me |
What is the difference between 10.8.z and the master branch? Is it possible for the 2 branches to be merged? |
Got exactly the same issue on Macos Monterey with Jellyfin 10.8 installed throught homebrew. Playback just stops. |
I feel like this issue needs to be re-opened as it's clear that it's not been resolved. I've not been able to use my Jellyfin at all the past week and I've resorted to going back to Plex. |
No issues for me since 10.8.1 so I'm guessing this is something else, mine has been working flawlessly |
And I've had the issue since 10.8.1 for all clients. |
Right so that's my point, I had the issue until 10.8.1 and you've had it since, it's likely a different or new issue affecting different users |
Not sure you see my previous comments or not, I never specified it was this issue directly. Others are reporting this issue is not fixed for them specifically, I am merely asking whether mine is attached to this. No one confirmed nor denied. |
im also having this issue on chromecasting to my chromecast ultra. always stops playback after apr 10 min.
|
I'm seeing the same issue. My logs are effectively identical to the ones posted by @y2kboy23 above. EDIT: Since this is a problem that happens intermittently (for me), I decided to check the logs for playback sessions that did not terminate prematurely. It looks like the A quick search of the code base shows that the log is generated here in the |
Doesn't this indicate that it isn't this issue? Since this issues is about TranscodingJobHelper deleting transcoded file after transcode job is finished but before playback is finished resulting in ending playback prematurely and the issues you're describing is about transcoding job ends prematurely. PS. I've had this issues since 10.8.0 but 10.8.1 fixed it for me |
@SlowNStupid I just realized that I've been running |
Ah sorry, my wording was bad. I meant that the original issues (the issues described in OP) was fixed in 10.8.1 for me. |
Yeah, no, the transcoding kill timer on 10.8.1 is definitely way too aggressive. I don't exactly have a basis for comparison, but when transcoding Hi10p down to normal H264 for my chromecast it gets about 6000 frames in and then terminates for absolutely no reason. Log attached. It works fine on smaller files, for what its worth.
|
I got the exact same problem. Any updates? I get this error after 15mins of playing and then the playback stops.
I tried 10.8.0, 10.8.1 and 10.8.3, all the same problem.
Any updates on this? Why is this issue closed? I had to switch to Plex because Jellyfin is unusable for me cause its deleting unfinished transcodes. |
This exact same thing is happening for me. 10.8.5. Android client, MacOS Jellyfin Mobile client, and Firefox 105.0.1 (MacOS) web client. |
If someone comes over this issue, I want to add to @zx900930 suggestion: I still received hangs/GPU crashes sometimes with h265 (h264 works fine) and alder lake. This does not happen often, but basically the whole VM is not usable any more in this state. I even tried the latest drm-tip kernel and a fresh compilation of the newest intel-media-driver. It seems that simply the drivers are still immature for the new hybrid CPUs (12thgen+). You can see this by the amount of h265 issues in the intel-media-driver repository on Github. What you can do, however, is actually set i915.enable_guc from 2 to 3. This will fully enable the GuC itself, responsible for power management. Now this won't prevent crashes, but now the GPU will be able to recover from them and restart itself and the driver. This is so fast that the user doesn't notice that (ffmpeg is killed and a new process simply spawns, continuing the transcode). Whats also interesting is, once it crashes for the second or third time, no more crashes are happening for me. |
Same issue here. It seems that whenever the server transcodes in mkv instead of ts, it deletes too rapidly the transcoded content. |
I can confirm that disabling "Throttle Transcodes" fixed the issue for me. With the setting enabled, Jellyfin pauses the transcode but never resumes it. The ffmpeg Log ends like this:
|
Hi all, I'm having this same issue with the Docker image of 10.8.10, the latest version at the time of writing. This only affects live TV. Throttling was already disabled prior to finding this thread. The container is hosted on a Kaby Lake NUC running Ubuntu 22.04 and has QSV enabled. I tried VAAPI and even disabling hardware acceleration for comparison, but no difference. The latest intel-media-va-driver-non-free driver is installed. intel_gpu_top shows no activity when running a stream Interestingly enough, the stream will restart after stalling out when being played from the Fire TV app but will stall and crash when using Chromium or Firefox. See the log messages below. The second message repeats until the stream is restarted on Fire TV.
It's very possible I'm missing what the fix is, but I don't think the kernel mods suggested above for Alder Lake are applicable and throttling is already disabled like I mentioned |
I'm experiencing this issue lately, with both transcoding and remuxing. In only certain cases, the server finishes transcoding or remuxing but deletes the output file before playback is done, causing playback to stop immediately. This is especially noticeable with the web client. Does anyone know of a solution? Right now my workaround is to use a client on my phone (specifially Findroid) to finish the video, but obviously this isn't optimal. |
I'm having this issue too with 10.8.10, transcoding Nvidia NVENC enabled, reproducing on Webos, it stops every few minutes, don't know why |
Reporting the same issues with VAAPI transcode. Stream pauses with "source error" every 5-10 minutes. When I press the Play button, playback continues. This is all there is in logs:
Also I get a bunch of "Slow HTTP Response from" messages. Tho I'm local with a great bandwidth. Checked resources: no OOM, no overheating etc. |
Hi, I solved disabling the power savings setting of windows 11 (or delaying it enough, like 3 hours). |
Same issue here for me. x265 playback and it cuts out with just about a minute left usually. The fix for me is to go back to using 264, but it's a shame :( Running on local hardware using VAAPI. Throttling not enabled with lots of disk space (+2 terabytes). |
Please head to our chat rooms or our forum for troubleshooting help before posting here: https://jellyfin.org/contact |
Please describe your bug
It appears the TranscodingJobHelper is deleting the stream files before playback completes causing playback to stop.
TranscodingJobHelper reports that ffmpeg transcoding completed then 52 seconds later deletes the stream file while its still being played.
Jellyfin Version
Other
if other:
10.8.0
Environment
Jellyfin logs
FFmpeg logs
Please attach any browser or client logs here
No response
Please attach any screenshots here
No response
Code of Conduct
The text was updated successfully, but these errors were encountered: