Having the same problem. Running Intel i7-8700 on Windows 10, using integrated Intel graphics with quicksync. With hardware transcoding enabled as soon as a video starts that needs transcoding the entire Plex Media Server app crashes and I need to restart the Plex Media Server on my computer. With hardware transcoding disabled, all transcodes work flawlessly.
Been like this for a few months now and not seeing any kind of talk of resolution from Plex on the issue. I’d really love to be able to turn on my hardware transcoding again, much less load on the CPU itself when I was able to use it.
Well, at least now I know why I was getting intermittent crashes. I noticed that H.264 files would play, but any H.265 file caused the entire server to crash until I turned off hardware acceleration. Now both file types play. I’m running PMS on an Intel Core i7-4770K with an Nvidia GTX 570 (that has no h.265 acceleration) on Windows 10 1903.
The Plex Team has always been extremely responsive. Remember, they probably use this software at home too. They’ll get it fixed ASAP I’m sure.
I would like any one getting the crashes and running Plex Media Server as a service or scheduled task or via RDP / Headless to stop doing that so we can concentrate on any new crashes. The issue with crashes in the intel driver when running headless / as service goes back over a year
With regards to recent crashes - we have one crash that appears to have come in between version 1.18.0.1944-f2cae8d6b and version 1.18.1.1966-10fc2f8d3. These are the crashes that manifest themselves in the server logs with this logged error before crashing ERROR - [FFMPEG] - Error setting child device handle: -xxx where xxx is a number that varies
We have another crash that I am trying to establish when it started or see if it is an intel driver issue
I have feedback from the development team that there is a suspected intel driver bug on Apollo Lake systems
And we have the RDP/Service/headless issue - which we should eliminate by not running in this mode whilst trying to find the cause of the recent crashes
And do you have any updates for those of us NOT using QuickSync for transcoding, i.e on Xeon platforms, especially on Windows where Transcoding spawns more threads when trying to seek on a file?
Thanks for sending me the crash dmp file for the crash at 06:59 UTC on Nov 7
It looks like the crashes that we now know were caused by the intel driver - see matrix here
Successfully tested with a remote Xbox One client that rolling Plex back to this exact version fixes any and all transcoding and buffer issues.
Updating to any version newer than this immediately breaks transcoding and therefore the client experiences buffering.
With this version, the transcoder also correctly buffers the 30-second chunks I have it set to, then sleeps, whereas on newer versions it’ll consistently sit at near max CPU use.
To help identify the cause of the crashes that came in between versions 1.18.0.1944-f2cae8d6b
and 1.18.1.1966-10fc2f8d3 , I am producing server development builds to be tried out. I will send you links by Private Message.
If you have disabled the graphics card, please re-enable and confirm the crashes are still occurring and then install the versions I will send you. Note that if the version number is lower than what is installed, you would need to do an uninstall first through Add or Remove programs
I am making the builds available to users that I have seen evidence from and I confirmed the issue
I’m having the same issue. My Plex Media Server (PMS) seemed to start crashing whenever using hardware encoding after updating to any PMS Version 1.18.x. By crashing, I mean the PMS tray icon disappears on the PC it is running on, and PMS stops responding. Restarting PMS manually works, so long as no HW decoder streams are started. I am not running ‘headless’. All versions of 1.18.x seem to have the issue. I’m currently running Version 1.18.2.2041. If I disable hardware encoding, the issue goes away. If I uninstall, and fall back to the last version 1.17.x, hardware encoding will work once again. My setup is using Windows 10 1909, an Intel i7 4790, with 28GB RAM, and an nVidia GT 1030.
So if your server is crashing also on 1.18.0.1944-f2cae8d6b and you are not running Plex Media Server as headless or as a service and you are not running on intel GPU drivers that have been seen to crash some Xeon E3-xx systems - listed here - then I would like to see dmp files and server logs after a crash
Here is a log file that I created right after a PMS crash that seemed to occur due to enabling Hardware Encoding. I disabled the feature, and PMS did not crash afterwards. I have not tried uploading logs before, so let me know if this is not correct. And, thanks for your assistance.
OK. So, why do the versions of 1.18.x I have installed coincide with hardware transcoding causing my PMS to crash? If I enable it, PMS crashes. I disable it, PMS does not crash. Or, if I downgrade to the last version of PMS 1.17.x and enable hardware transcoding PMS does not seem to crash… Now, I guess you are getting more specific as to the version numbers. I noticed the issue once I knew that I had started the 1.18.x versions. It may be possible that it was the 1.18.1.x version that started the crashing. But, that is still the issue I see happening. I bring this up and shared my logs in the hope this gets resolved. Thanks for you help on this.
I presume it is because you have not tried 1.18.0.1944-f2cae8d6b
I have already stated which version this bug came in with and that is under investigation with help from a number of users trying out different test builds.
