I run a cluster of Plex servers using Docker distributed over multiple bare-metal servers. Since users have been upgrading to 1.15 I’ve been seeing a very large spike in memory usage that just keeps spiking the longer the processes live. The average memory usage used to be around the 300MB RSS but the last few weeks I’ve seen it go up as high as 3GB before my cgroups kill it for OOM reasons. This is an overview of processes sorted by memory usage of one of the servers.
Is this a known bug? If not can I do something get it to be known?
I am noticing the same thing with the Linux x86_64 builds. It coincided with my upgrade from 1.14.1.5488 to 1.15.1.79 last month. I am running it on a PC media center also running KODI, and I have noticed that the memory utilization gets very high and causes the UI to become very sluggish, even unresponsive at times. A restart of the PMS service restores the system to normal operations.
I’ll attempt to troubleshoot this a bit further in order to possibly provide more information.
We are investigating an issue of high memory usage on Linux systems that have DVR set up. I’d like to determine if this is the same issue or not.
Do you have DVR set up on the systems that are experiencing this issue? If you are willing to delete your DVR, please do so and restart PMS, and then see if you still have the problem. Otherwise, if you don’t want to delete your DVR, please do a manual guide refresh (under Live TV & DVR Settings) and see if that causes a spike in memory that does not go down after the refresh is done.
I’m having the same problem on my Synology 415+ NAS with 8 GB RAM. I have this server running for 4 years and since v1.15, the Plex process goes up to 5 GB RAM and kills the whole thing. I have to reboot once it consumes all free resources.
I’m not using DVR, only one movie and one TV shows library.
I’m am definitely seeing the issue associated with DVR. I’m running 1.15.4.919 in docker on Linux. Memory usage spikes during guide refreshes, and is not released when finished. My system with 8 GB of RAM starts using the Swap partition after about 2-3 days without restarting Plex. I deleted my DVR, and after a week now, Plex hasn’t gone above about 2% RAM usage.
I have the same issue on my plex server running in a ubuntu proxmox container. The avg memory used by the container was 600 MB until around Feb 13th. The memory usage now slowly climbs up to 2.5 GB in about 4-5 days and then it gets reset back to 600MB.
I’m using the DVR and refreshing the guide did not cause any memory usage spikes, but it lowered the usage from 1.5 GB to 800 MB after it finished.
I have noticed this on my xigmanas container as well, I thought it was just myself but I had to tune the system variables to free up another 8gb of ram for plex so my system would stop running out of RAM and using swap. Version 1.15.3.876
I am also running Plex Server on Ubuntu 18.04.2 in a Docker container with dvr/live tv setup. Mine starts at 1.2 gb and will slowly climb to max out my ram at 32 gb over the course of 5 days or so, until it crashes and restarts ever since i upgraded from 1.14 to 1.15. I currently have 1.15.4.993 and nothing has changed. Disabling dvr doesn’t make a difference either.
Adding fuel to this one, i noticed my ram usage and swap usage go up over the past few weeks, eventually tracked it down to the plex process. I don’t have any fancy graphs for this but its using 3-5GB’s idle and 3GB of swap space just for plex.
And now I cant download 1.14.1 any longer…this sucks. I been staying on that and then noticed last 2 days it crashed each day and sure enough it was on 1.15.4 now.
I had plex running without issues for months, but sense the last couple of PMS i have had random crashes every couple of days. I will keep an eye on RAM, as it could be building until it crashes.
