Plex CPU hang

server-linux

#1

I'm still trying to narrow down the source of the problem but there is a CPU wait hang issue in kernel 4.12.12 that Plex frequently triggers.
You lose a CPU to 100% wa time and the hung Plex Media Service process becomes a zombie.


#2

a) when it hangs again , we’ll need a a crash dump.

sudo kill -11 `pidof 'Plex Media Server'`

b) as soon as it restarts get a full set of log files (Settings - Server - Help - Download Logs) and attach the ZIP file here with your next post.

until then, I need considerably more information about your machine.

a) distro and version
b) storage configuration ( media local or network )
c)transcoding?

See what I’m asking? What’s relevant?


#3

Gentoo
Local media on an ext4 volume; LVM raid5 backing.
There should have been no transcoding; it hung the media server.

The machine was heavily loaded; firefox playing a video via plex; makemkv ripping a bluray; handbrake encoding videos; QEmu running Windows; a console doing the Gentoo never-ending-build. (reNice values set appropriately.)
(I do not overclock; CPU temps were warm but <40 C).

The process cannot be killed. I can’t reboot either; I have to kill power to restart.
Firefox will eventually lock up as well.
I am currently running on 4.12.5, i915 IGU with no issues so far.
I want to reintroduce the nvidia driver (970 GTX) to make certain that is stable on 4.12.5 next.

All of the behavior added up suggest either an nvidia module bug, hardware failure, or a kernel bug (as opposed to Plex) and the stability on 4.12.5 suggest it’s not my hardware (unless it’s the 970 that’s failing).
The kernel charfed a fault to the log when it happened.
In retrospect I should have kept those logs but I was frustrated trying to get the system working again.
Once I have my GPU working with a stable setup again I’ll go back to 4.12.12 and get the log.

So I don’t think this is Plex’s problem to solve but thought you might want to know an issue is out there that could rear its head soon in the main-stream distros.


#4

What do you find in the system log (journalctl -xe) or /var/log/messages with dmesg ? there must be something in there to give you a clue


#5

The issue was resolved upstream; I’m running v4.14.13 now without any hangs.


#6

Thanks for letting me know. I will pass that along