I have started growing my 4K library now that I have a 4K tv. I have noticed that when playing 4K files I will have some random spikes of 100% CPU usage and crashes of the stream.
I do have one 4K HDR movie that will play through without crashing and I noticed it is the only one that doesn’t have Dolby TrueHD Audio. So I’m running a test right now, I watched the first half of a movie with TrueHD enabled and it crashed 3 times. Now I’m watching the second half with the audio switched to AC3 5.1, so far no crashes.
Logs are attached. I really hope I can get this fixed.
Looking at your log, it’s only transcoding the audio so there shouldn’t be much cpu usage. I don’t see any crashes of playback. When did it happen last?
If I recall it was about 3ish hours ago. It happened about 4 times in total today, and its happened numerous times over the last week.
However, I am playing a movie right now with TrueHD enabled and its almost over without a crash, I was trying to make it happen again to get another set of logs. But this is definitely the exception, normally it would have, at a minimum, stuttered and buffered numerous times by now.
We’re you adding a lot of content while watching and was your server scanning the files for info such as chapter thumbnails and so forth at the same time.
No I wasn’t adding anything at the time. I did update the new TV Show agent yesterday and refresh meta data, but I was having this issue before that anyway.
And yes, I am using CPU pinning, otherwise my entire UNRAID server becomes unreachable when it happens. I have 10 of 12 threads available to PLEX.
It just happened, the credits were getting ready to roll and it FINALLY crashed again. It happened at 5:11 its now 5:18. I also managed to snap a picture of the CPU monitor in UNRAID when it happened.
Alright are you doing pinning or are you doing it to isolated cpus? If you pin plex to multiple isolated cpus, it will only use one even if you have it assigned to multiple assuming you are using dockers that is. If you put plex in a VM on Ubuntu or something, you can assign it multiple isolated cpus.
Unraid is a resource hog. The mover or something is bogging down your pinned cpus. Pinning just means they can only use those cpus, not that other tasks can’t. Unraid is still using them for its needs. Isolated cpus are only used by the thing they are assigned to.
I mean it may not. However I was having similar issues and I just removed the pinning altogether and noticed improvement. Unraid did a decent job of sorting stuff out
Your logs show everything was fine just before this spike. The transcoder was running at >20x speed, which is right when only doing audio transcoding.
I can’t tell if it was PMS or something else on your computer that caused the cpu spike, but that did cause PMS to close the connection to the client and that’s why playback stopped. I’m still looking.
I’ve been digging through every app on my system I can think of and none of them should even be capable of doing that. Plex is the only thing on the system pinned to that number of threads, and it’s only those threads that spike.
Are you opposed to turning pinning off for a week and seeing if the issues resolves? Most people I have found really only use pinning in unraid with VMs in an isolated cpu environment. It would be an easy way to rule it out. It may not be plex running out of resources but unraid stalling on 2 cores even momentarily. Or remove a couple pins from plex.
You have the gpu for hardware acceleration, so you should be able to loosen up on the cpu pinning. I only have 4 cores, all unpinned with about 12 dockers and multiple plugins and no problem, 1800 movies, 5500+ tv shows. I use quick sync vs discrete gpu tho. 4-8 streams going at any given time typically. I5-7400
Which plex docker are you using. I tend to stick with linuxserver.
Im using Linuxserver as-well. I can certainly try disabling pinning and see how it goes. It was originally only enabled for an unrelated and resolved issue, and I just never saw a need to un-do it.
Ok, so it just happened again except now it was all cores (I disabled all pinning), and no-one was streaming anything with TrueHD, I was however watching a movie with DTS-HD MA 6.1 which I have never heard of, but according to Tautulli it was Direct Playing.
When this happens I cannot access the Plex WebUI until whatever it is settles down (usually 2-3min). While waiting I did run htop on the Unraid host and saw Plex Transcoder as the number 1 process.
I did manage to grab the Plex logs after it calmed down. I also grabbed the unraid logs if anyone here thinks they would be useful I can upload them.
