Seems like they could do that already ascertain this with the two already-provided reports, but OK. 
Well, now I am seeing a new crash. It no longer sticks to the 33% but goes straight to Roku. I am hoping it is the same bug in a different set of clothes.
I’ve had it occur in less than 20 minutes so thinking it’s the same one.
Also enjoying the spectre here of two corporations worth multi-millions if not billions of dollars having to rely on me to report not only once, not only twice, but three freaking times here for the same issue.
Looks like the same?
Faulting application name: Plex Transcoder.exe, version: 0.0.0.0, time stamp: 0x6322df1a
Faulting module name: nvcuvid.dll, version: 7.17.15.2647, time stamp: 0x635830fa
Exception code: 0xc0000005
I am definitely seeing the Roku Client crash now 3 times, and only one of them surfaced with the above error. Pertinent times are in ET:
- 11:56am
- 12:21pm
- 1:07pm
Only the 11:56am has emitted to the Application Event Log
20GB file inbound
That is the same - a transcoder crash in the NVIDIA driver
If the Plex for Roku app is crashing and exiting out to the Roku home screen then we will need to investigate that - not as an nvidia issue but as a Plex for Roku issue until we know why it is happening
To investigate the Plex for Roku crashes what would be needed is to enable network logging on the roku app and to have the server logs and details of the Plex for Roku app environment: Roku device ID, Roku version, Plex account signed by the Plex for Roku app and time(s) of the Roku Plex app crashes
To enable network remote logging on the Plex for Roku app - you need to get into Settings on the app, and then select Advanced and select Logging and change it from Disabled to Remote
So please ensure when launching the Plex for Roku app that you have remote logging enabled. Local logging would not help here since the app is exiting. The setting should be remembered once set - but please double check that it is enabled when starting to stream DVR Live TV and expecting to get the app crashing
When it happens, note down the time and get the server logs and I will look into it and refer to the roku plex team - would need the roku device identifier and plex.tv account used
With regards to the diagnostics I received, the server logs covered the period
Oct 29, 2022 from 12:11:07 to 13:16:00
Application Event Log shows these nvidia module transcoder crashes
Oct 29, 2022 at 11:56:59
Oct 13, 2022 at 01:05:52
Oct 08, 2022 at 21:02:36
Oct 02, 2022 at 19:31:34
There was one Trancoder crash that was not in the nvidia module - it was
Oct 16, 2022 at 18:43:51
So for the 29th October NVIDIA module crash in the transcoder at 11:56:59 - would need the dump file - which ought to have got saved by Windows into the %LocalAppData%\CrashDumps assuming you still have the regedit outlined here Crash Logs: Plex Media Server | Plex Support
Also whilst I do have the transcoder sessions files going back to 11:33 am - so should cover the time of the nvidia dll crash - but without the server logs covering the time period, i cannot identify which of the sets of the transcoder temp files are relevant
The crash happened at 11:59:56.
The server logs were captured at 13:16 and only cover period 12:11:07 to 13:16:00
So to summarize:
For the Plex for Roku crashes - would need app remote logging enabled and server logs initially and roku specific details (ID / Account / Time)
For the transcoder crashes, need the dump file and server logs captured very soon after and the event logs and the transcoder sessions directory zips created / retained with the 4 specific registry settings
To clear up the transcoder sessions files - suggest deleting old directories manually or disabling the special settings and restarting PMS and then re-enabling the settings and getting the diagnostics set
1 Like
Thank you very much for your thorough details, update, and steps @sa2000.
Funny enough I did make a capture of logs shortly after the crash but deleted it as I thought getting a new capture would include it. I did not realize that download logs means download logs for the past hour. It would seem that the interface would denote this.
I have updated the store location with DMP files and server logs.
Please let me know if you have everything you need for this and then we can focus on the Roku crashes which appear to be a different matter (and should probably have its own thread).
Created here: Roku Crashes
Thanks for adding dumps and logs
Got the logs covering the time of the 11:56:59 crash
Assuming you have not made any changes to drivers or Plex Media Server version since the 29th October crash, I will send you a message asking for some binaries/dll’s to be zipped and sent to me and details of the Nvidia drivers/firmware in use
1 Like
Hi @sa2000 would any of these settings result in pixelated transcoding for live tv streams? I did delete them all and have reset the server since then, but am now experiencing pixelated rendering for live TV streams as described here: Live TV HD Appears Pixelated
Thank you for any insight/assistance you can provide. 
This setting adds extra writing to the filesystem - saving the channel stream capture
TranscoderInputTracing Set the value to 1
The impact of the other three would be using up disc space and not freeing it up which may lead to running out of disc space
So the capture of the channel stream could slow down the transcoder process
Thank you for that update @sa2000. Note that I had to enter in all 4 of them, and then deleted them, and am now experiencing pixelated streaming when viewing 720p channels. Not sure it’s related but trying to track it down.
Wanted to check in on this issue to see if NVidia has found a way to weasel their way into another excuse not to fix the problem here.
Happy Holidays 
bump
Looking forward to NVidia’s imaginative reply.
Not as snarky/petulant bump
Bump
Sorry - nothing new from NVIDIA.
Suggest keep on updating to latest Plex Media Server version and latest Nvidia Graphics Cards drivers / firmware and may be the crashes would stop.
AKA literally what I have been doing since reporting this October '21.
Any comment/idea why NVidia would have us jump through hoops here and not actually fix the problem, @sa2000 ?
Thank you for any update you can provide to this issue, reported October 16th, 2021, which was 520 days ago:
