Server Version#: 1.14.1.5488 (Public channel)
Player Version#: Several
Hi everyone,
I’m currently experiencing a very annoying phenomenon on my Plex server and I can’t seem to find anyone with a similar issue.
I’m running the server on a Debian system (Stretch v9.8, Linux v4.9.144-3.1), which is installed on an Intel NUC Baby Canyon NUC7i3BNK.
The whole system crashes after someone has been playing a video via transcoder (about 4-5 users regularly play something from remote).
It doesn’t happen every time, so I think it is a combination of some parameters like the quality selected, the kind of player or if subtitles are involved or not. I also could reproduce the crash in my local network with 3 different players.
So far I could not find a pattern, except for the transcoding part. Before every crash, some video was transcoding.
The strange thing is, that the system will completely shut down for about 15 minutes and it can’t be started again during that time. Overheating is not an issue and the power source is also fine.
So from that point of view, it would be an issue with the OS or hardware.
But on the other hand, we have the fact that transcoding was involved in every crash, which would let us think it is a PMS bug.
The issue first occured a few months back and I didn’t think any of it, because it was happening maybe once every few weeks.
In the last 3-4 weeks it started happening several times a week.
Before that, the server was running perfectly fine for over a year.
Does anyone have an idea on how this can be solved or is facing a similar issue?
I’m new to this forum, so please advise if you need more info or specific logs from my side.
It definitely sounds like a temperature issue, what are the CPU and Mboard temps?
The i3-7100u has a passmark of 3778 which would be good for 1 1080p 20-25 Mbit transcode. If you have debug logging turned on, reproduce the crash and then zip up the logs, post them here. In Plex Media Player, Settings > Troubleshooting > download logs will get you the file.
Thanks for the answer!
I will do a stress test and let you know the result as soon as I have the time for it.
Until now I’ve only ever checked the temperature of the chassis by hand and didn’t use any system sensors.
However, it would be very strange if this is a temperature issue, since I haven’t had any problems until a few months ago.
The amount of users or the media type (mostly 1080p x265 files) did not change since then and the NUCs location has got sufficient air flow.
The NUC itself lets you put its fans and throttling in “automatic” mode or to increase manually.
It is very good at showing you all temps and reporting then to the kernel.
I run my NUC in full auto mode. I can monitor from the Gnome system-monitor plug-in just for sanity sake. A ‘server’ install doesn’t have the GUI so this is where letting the NUC itself manage performance works best.
@ChuckPa:
Thanks for the tip. I guess these settings are to be found in the BIOS menu?
@pl_5309:
I reproduced the crash once again while monitoring the temperature.
I did so by playing some videos on my mobile and setting the quality to original, while using subtitles.
The first video didn’t trigger the crash (1080p x256 .mkv file, SRT subtitles), but the second one did (1080p x256 .mkv file, PGS subtitles).
As you can see in the screenshot below, the temperature reached around 71 degrees. Intel specifies a maximum operating temperature of 100 degrees.
While playing the first video, temp was around 70 degrees too, but it was playing fine for about 15 minutes, which is why I switched to the second one (with different subtitle format).
Noteable: I’ve upgraded PMS to v1.15.1.791 this morning.
Yes. They are found in the (semi) permanent “BIOS”. Technically it’s “Firmware” now because it’s largely static and far beyond the old meaning (“Basic Input Output Services”) because it’s not the least bit ‘basic’
Is your media on an external USB drive? There are two slow disk messages and client not responding, not sure which caused which. I didn’t see any indication of the Server crashing. These appear to be all remote android clients, have you checked for issues with the client.
@pl_5309:
My media is on a NAS, which is attached to the same 1 Gbps switch as the PMS is.
The NAS contains WD Red drives, so I don’t think the disk performance is an issue.
The log stops at the time of the system crash (file with “.1.log”), so we only know what is happening before.
I’ve triggered the crash with my Android phone to fill the log, but it happens with all sorts of clients (e.g. Samsung TV, Chrome or Chromecast).
Alright guys, it’s definitely a CPU temp issue.
I’ve been running some stress tests and the NUC shut down at around the same temperature.
This means the values of the used monitoring package are terribly wrong. BIOS reports an idle temp of nearly 60°, monitoring just around 40°… I suspect the CPU is at the maximum 100°, when the system shuts down.
Now, is there a way to make the transcoder consume less CPU power?
I’ve already set it to the lowest settings (prefer higher speed encoding and ultra fast x264 preset).
Are there other options to consider? Or am I understanding the settings wrong?
