Server Version#:plexmediaserver_1.14.1.5488-cc260c476_amd64.deb, Ubuntu 18.10
Player Version#:
After installing this version, my CPU usage goes to 100% even though nothing is going on.
If I uninstall this version, and reinstall the prior version, it all works without the high CPU usage.
Any ideas what’s wrong?
Not one without seeing the log files.
Can you please recreate and submit the DEBUG ZIP file you download from the server?
Be advised. 1.14.1 is now frozen development.
1.15.0 is about to become public,
1.15.2 is about to be PlexPass
Prior version is reinstalled right now.
Will wait for 1.15.x forthcoming release to reinstall and retest.
Will create new topic and include .zip file.
I learned we’re investigating one other case of this.
If you are up to recreating, would you mind starting a wireshark capture just before launching PMS?
Doing so will confirm or clear SSDP (which uses that lib extensively)
Chuck,
Thanks for the follow up. I’ll wait for 1.15.x to retry, and I’ll do a Wireshark then, and a .zip file, if the problem still exists. Please give me specific instructions for what I need to do in WS, as I’m not an experienced WS user.
Sounds like it’s doing movie previews and thumbnails
Preview thumbnails is also a possibility.
The easy test of this is to let the server sit idle for a minute.
sudo du -ms /var/lib/plexmediaserver
Wait another few minutes, and repeat.
If the size is growing, it’s doing thumbnails.
Chuck, I’m pretty sure that it wasn’t working on preview thumbnails. Letting PMS idle made no difference, it was just grinding away at 100% CPU… and replacing PMS with the prior version solved that.
Is this in a container or VM of some type?
Chuck, no container or VM… just a straight Ubuntu .deb install.
this makes no sense. 1.14.1 was built on the old system. I am seeing issues on the new build system but particular to 1.15.x
which Ubuntu version?
Chuck, it’s 18.10
Makes no sense to me but I will update my 18.10 VM and see what I can find if anything.
Hi @heynnema, are you by any chance running an OpenWrt router on your network with UPnP enabled (or any home-grown system with miniupnpd)? 100% CPU usage (for one core in multi-core CPUs) has been seen by quite a few people with the 1.14.1.5488 and later builds of PMS on at least Windows, macOS and linux (and derivatives).
Try turning off UPnP (no need to turn off NAT-PMP, just UPnP) on your router and restart PMS. You should see the CPU usage return to normal if this is the same issue. Going back to 1.14.5470 resolved the issue in every case and UPnP was able to be turned back on again.
Hello Gary,
Your problem description is similar to my problem… one CPU, out of 4, is driven to 100%.
The problem is not seen with PMS 1.13.8.5395, that I have installed now.
It sounds like you’ve heard other reports about this problem with 1.14.x.
I have dd-wrt running as a bridge on a Cisco E2000 for some A/V equipment, but UPnP is disabled. UPnP is also disabled on my main Netgear 6400v2 router. I don’t believe that I have any miniupnpd running anywhere. I also have two Netgear extenders, EX6100 and EX7000, and I don’t believe they have any UPnP options.
Here what UPnP software is running on my host…
||/ Name Version Architecture Description
++±==============-============-============-=================================
ii libgupnp-1.0-4 1.0.3-1 amd64 GObject-based library for UPnP
ii libgupnp-igd-1 0.2.5-2 amd64 library to handle UPnP IGD port m
ii libminiupnpc10 1.9.20140610 amd64 UPnP IGD client lightweight libra
ii libminiupnpc17 2.1-1 amd64 UPnP IGD client lightweight libra
ii libnet-upnp-pe 1.4.4-1 all Perl extensions for UPnP
ii libupnp6:amd64 1:1.6.24-4 amd64 Portable SDK for UPnP Devices, ve
ii upnp-router-co 0.2+r119~ubu amd64 UPnP compliant routers manager
I have just run 1.13.8, 1.14.1, and 1.15.0 in Ubuntu 18.10 VM.
I am seeing no abhorrent behavior.
It must also be said I have a pfSense handling UPnP for port management.
Chuck,
Let me take the time to put 1.14.1 back on, I’ll watch the CPU, and I’ll get a .zip to you.
Chuck,
I reinstalled 1.14.1, CPU went straight to 100%. I’ve attached the .zip file. That should make things easier for you.
Made it trivial.
Mar 06, 2019 12:07:09.208 [0x7f68177fe700] ERROR - [Notify] Failed to add watch for /home/alheynneman/Pictures/Christmas Wallpapers (13: Permission denied)
Mar 06, 2019 12:07:09.208 [0x7f68177fe700] DEBUG - [Notify] Caught exception on directory.
Remedy: Increase the default table value to something more appropriate. Let’s start with something more reasonable. You can edit the value later and increase (in powers of 2) as appropriate.
sudo echo fs.inotify.max_user_watches=65536 >> /etc/sysctl.conf
sudo sysctl -p
Chuck,
Thanks for your time!
In /etc/sysctl.conf I already had fs.inotify.max_user_watches=1048576. I increased it by 50%, retried PMS, and it still had 100% CPU. So… I doubled the original value, same thing…
sudo sysctl -w fs.inotify.max_user_watches=2097152
I’ll attach the latest .zip for you.
