Server Version#: 1.15.1.707
Player Version#: Web / 3.83.1
This was going to be a call for help, but thanks to the “your topic is similar to…” bit to the right of the editor, I was able to find a recent thread and solve my issue, but I still want to document this issue as it seems to be affecting several people recently.
Starting around a month ago, I noticed that my server was consuming ~20% CPU usage on my 4790k at all hours while idling. I let it slide for maybe a week, but when I saw it wasn’t going away I turned off the server until I could find time to troubleshoot.
That troubleshooting started a couple days ago, and it was a bit hair-pulling. I fully wiped everything related to PMS a couple times, but even on a fresh install, it would still happen within 20 seconds of the server starting up. I had found threads from a few years ago talking about disabling some of the advanced scheduled tasks, but none of them affected this CPU usage.
Then I saw a fewrecent threads that advised disabling UPNP on my router, and sure enough that has fixed the issue.
Attached are the logs from a 100% fresh PMS install. Started the server, got the 20% CPU usage, disabled UPNP in router, restarted server, and all is well. Hopefully this issue can be sorted soon.
Hi @Tonystew42, thanks for sharing your experience with this one. Sadly there’s a growing number of us waiting for it to get fixed.
What router hardware/software are you running? Everyone experiencing the issue so far has been running OpenWrt with the miniupnpd daemon on it. I think you may be the first user with PMS on a Windows machine, though.
Edit - Forgot to mention, I also have dedicated port-forwarding set up for Plex on ports 32400, 1900, 3005, 5353, 8324, 32410, 32412, 32413, 32414, and 32469.
Hey guys!
I’m experiencing the same issue - latest Plex Media Server, running Windows 10, constantly utilizes ~20% CPU when miniupnpd is turned on. My router is running latest OpenWRT as well. I’ve faced already with this issue at Linux before, fixed that by turning off miniupnpd daemon.
We’ve been making some decent headway with diagnosing the issue over here:
The thread goes on a bit, but tl;dr it looks like a bug in PMS’s parsing of SSDP replies from UPnP replies on the LAN and it’s been raised with engineering. Release notes for 1.14.1.5488 note “(SSDP) Better checking for invalid IP addresses in location field of SSDP XML”, which is when the high CPU behaviour first manifested.
Stumbled onto this experiencing the same problems. Running Plex on an Ubuntu VM (I really should move to docker) with an OpenWRT router. I had enabled the UPnP module in the Gui (formally LEDE). And that blew my CPU out where the VM would consume a core constantly.
Fiddled with some settings and believe I have come up with a work around.
Within LuCI under Services → UPnP → MiniUPnP Settings
Start UPnP and NAT-PMP service = Enabled
Enable UPnP functionality = Enabled
Enable NAT-PMP functality = Disabled
It was disabling one of UPnP or NAT-PMP that worked for me. If both are enabled my CPU spikes.
