Server Version#: 1.41.8.9834
Player Version#: N/A
Logs: Debug logs available to upon request, there is a LOT of sensitive information even though the tokens are masked
Hardware:
CPU: 14700k Intel
NVMe SSD: 2TB - Only about 1TB is in use
RAM: 64GB @ 4000MHz
TL:DR - PMS server locks up at the top of the hour EVERY hour regardless of current usage or even lack of background tasks running.
My Plex Media server is using the linuxserver/plex docker image and has been in service for over 10+ years. Lately I’ve been having some weird issues with ALL clients disconnecting at the same time and found that the PMS service every hour on the hour locks up. The CPU is at 100% utilization on all cores and only one process is using over 100% CPU and that’s the PMS service using 2000% (ie all the cores).
So at 3pm, it will lock up for about 3-4 minutes. Then at 4pm, same thing and so on.
Doesn’t matter if there’s no usage or high usage of the server during this time.
The research I’ve done turns up nothing specific to my issue.
The CPU is completely used during this timeframe and only the PMS service. Not the transcoding services or scanners, or anything else. There are no background tasks running, the “console” view doesn’t show any obvious errors and is unusable while the server is locked up. There is no IO wait in top and everything is running fine until this issue strikes at the top of the hour.
I can provide the debug logs of the latest time this happened, I grabbed them AS SOON as the server was made available again. I use Netdata to monitor this and have charts showing this reliably happening every hour and can provide any stats from any time. Or even screenshots of my config or anything. This has been happening for a couple of months, and even after optimizing the DB in Plex, or ChuckPA’s script and even after cleaning all the statistics bloat; this issue persists
All this to say, I cannot for the life of me figure out what’s going on. Any help would be appreciated.
Weirdly enough, I have a lot of hardware headroom usually that I run 3 PMS servers in different docker containers and only one of the 3 has this issue. The others are affected as well due to the lack of CPU, but the are just slow during this time and not locked up. Nor is CPU on these two other PMS services averaging higher than 100% (1 core)
