Which version of the nVidia drivers do you have installed?
418.30 is the required minimum. It’s not that new but is the min for HW transcoding to operate correctly.
The normal Plex logs (DEBUG, not VERBOSE) capturing the start of a session are of great help.
I just updated today to 440.36. Plex Media Server Logs_2019-12-06_13-57-45.zip (2.5 MB)
I’m coming in late, but i agree with others, i THOUGHT it was wifi only, but it’s happening on my wired roku as well now.
I haven’t tried to disable HD decoding. I’ll just sit back and watch from this point on, i’m not really contributing, but i’m here to say thanks for looking into this. My issues are exactly on par with what’s been discussed thusfar.
Hey @ChuckPa, just checking if there was any update on this. Thanks!
I have not been able to spend any time on it (sorry).
I’ve been up to my proverbial eye balls with TMDB and QNAP.
What’s the common element, Confined to Roku or other players (web) end early too?
No worries man, you do a lot of work that is greatly appreciated! I have only had this happen on roku and web. Whats up with TMDB? Ive been noticing TVDB changing a lot of stuff recently that temporarily broke my sonarr setup. That was annoying.
TheTVDB changed their API (v3 is now live)
It was supposed to be fully backward compatible with the v2 API but isn’t quite.
We saw where we can make a change to avoid that and did.
Unexpected fallout, TMDB HTTPCookies contents became invalid.
The resolution, for those impacted (very few) is to erase the directory contents and restart PMS.
Regarding web and roku problem, a thought… Consider updating to current.
There have been changes in the bundled Plex/Web and server.
I don’t know if these are related but (not being an app guy) it can’t hurt?
Well thanks for all your hard work keeping it all working! Ill make sure to get everything up to date and see how that goes. My wife said this happens on her ios device as well btw.
So it seems it stopped doing this for me. I made 2 changes, and i’m not sure how unrelated they are, but bare with me.
Now, #2 i had enabled a while ago, and i have no idea why i disabled it. The reason behind enabling it was this; Since the streams are ending early, i figured it would be easier to find where it ended early again by using the thumbnails. My library is rather large, and it took about 4 days to re-generate them all, but last night i watched maybe 5 things, and none of them stopped early.
It could all be co-incidence, but it seems to have fixed my issue. No idea if it appiles to others or even makes sense? I’m about to turn HW transcoding back on.
This is consistent with my experience. Try turning hw transcoding back on and see if the problem comes back. And if it does, snag some logs as a sacrifice for our Lord and savior, the great and powerful @ChuckPa.
Ugh, after a few days, this came back!!! nothing was changed, and it’s doing it again. So weird!
Just a random note, but I started a show on Version 1.18.4.2171 with full hw transcoding to see how it works and its not doing hw decoding. Just thought that was strange.
May I have some logs please? 
I may be good but I can’t work in the dark 
Sorry for delay (holiday “stuff”)
Happy holidays Chuck! Hope you had a well deserved break! Thanks for all the hard work! I’ll grab some logs sometime tomorrow. In the meantime I saw in console that it said there was no available hardware decoding accelerator available.
Here are logs where it doesnt hardware decode: Plex Media Server Logs_2019-12-30_11-45-53.zip (3.4 MB) and here are logs right after when i restarted my server and played the same file and it did do full hardware transcoding: Plex Media Server Logs_2019-12-30_11-48-54.zip (2.7 MB)
Out of memory errors.
Dec 30, 2019 11:44:02.756 [0x7f7e14ff9700] DEBUG - Request: [127.0.0.1:49264 (Loopback)] GET /:/metadata/updateProgressMessage?message=Scanning%20House%2FSeason%204 (23 live) GZIP Signed-in Token (LrrrAc)
Dec 30, 2019 11:44:02.756 [0x7f7df8ff9700] DEBUG - Activity: updated activity 279ca4e5-13c0-4cd4-bbc3-34170d8a76aa - completed 90.7% - Scanning Shows
Dec 30, 2019 11:44:02.756 [0x7f7e74f4b700] DEBUG - Completed: [127.0.0.1:49264] 200 GET /:/metadata/updateProgressMessage?message=Scanning%20House%2FSeason%204 (23 live) GZIP 0ms 166 bytes
Dec 30, 2019 11:44:02.757 [0x7f7e74f4b700] DEBUG - Completed: [127.0.0.1:49262] 200 PUT /activities/279ca4e5-13c0-4cd4-bbc3-34170d8a76aa?percentComplete=%2B0%2E0865384638&subtitle=Season%204 (23 live) GZIP 1ms 166 bytes
Dec 30, 2019 11:44:02.760 [0x7f7decff9700] ERROR - [FFMPEG] - cu->cuCtxCreate(&hwctx->cuda_ctx, 4, cu_device) failed
Dec 30, 2019 11:44:02.760 [0x7f7decff9700] ERROR - [FFMPEG] - -> CUDA_ERROR_OUT_OF_MEMORY: out of memory
Dec 30, 2019 11:44:02.760 [0x7f7decff9700] ERROR - [FFMPEG] -
Dec 30, 2019 11:44:02.760 [0x7f7decff9700] DEBUG - Codecs: hardware transcoding: opening hw device failed - probably not supported by this system, error: Unknown error occurred
Dec 30, 2019 11:44:02.760 [0x7f7decff9700] DEBUG - Codecs: testing hevc (decoder) with hwdevice vaapi
Dec 30, 2019 11:44:02.760 [0x7f7decff9700] DEBUG - Codecs: hardware transcoding: testing API vaapi
Dec 30, 2019 11:44:02.761 [0x7f7decff9700] DEBUG - Codecs: hardware transcoding: opening hw device failed - probably not supported by this system, error: Generic error in an external library
Dec 30, 2019 11:44:02.761 [0x7f7decff9700] DEBUG - Codecs: testing hevc (decoder) with hwdevice nvdec
Dec 30, 2019 11:44:02.761 [0x7f7decff9700] DEBUG - Codecs: hardware transcoding: testing API nvdec
I hope that means something to you.
I don’t know nVidia.
How many are you transcoding at once?
That was only one video. The buffer is set to only 60 seconds and I’ve never had that issue before even with 5-6 videos all transcoding together.
I am having this problem as well, and have had the problem for at least a month. I had upgraded PMS to the latest (1.18.3.2156) in the hopes that it would be fixed, but to no avail. I am running PMS on Windows 7 with hardware decoding disabled, using the “Make my CPU hurt” setting. Streaming to two different Roku clients (2019 Streaming Stick+ and 2016 Premiere+) has shown the problem (neither Roku is direct playing due to bandwidth limitations, since I am streaming uncompressed BD rips).
If you could restart your server and get a stream to crash, end, skip, or otherwise cause this issue and post the logs, that would be super helpful!
So I currently have HW transcoding turned off, and also the “kill stream when paused after x minutes” option turned off because I swear to god even with HW transcoding turned off I was still seeing some files end early on some devices, and only turning off that kill stream option fixed it.
The kill stream option was the culprit behind the fairly recent issues with Chromecasts. The problems with HW transcoding was causing stuttering and early ending depending on the file regardless of client. Sometimes everything would play fine. Turning off both these features makes everything run smoothly. I can only conclude that these two features, which had both been working pretty good prior to sometime during last summer, are still broken.
I don’t know what made HW transcoding, and kill stream after X minutes features become so bad when they had been so good, but I sure wish we could go back to that.
