Closed kchiem closed 1 year ago
Is this still a problem?
It was.. but since you're the first response, and it's been over a year, I've since switched over to linux for this. Has something changed such that this wouldn't be a problem anymore? If so, I can spend the effort to fire up a new VM and test it.
It's been over a year, a lot has happened both in the FreeBSD VM subsystem, and to a lesser extent in the drivers. FreeBSD 12.0 has also been replaced by 12.1. There has been a few memory leaks that have been fixed at one point, https://github.com/FreeBSDDesktop/kms-drm/commit/636503003d16314fae1ffa3d15a9ab04e616c12f comes to mind.
Ok, I'll try and test this in the next couple of days.
Running FreeBSD 12.0 on an Intel NUC8i5. With plex server using hardware transcode, I'm seeing wired memory climb at about 2 gb/hour while transcoding. Restarting plex doesn't free the memory. Eventually, free memory runs low and the system starts swapping, grinding to a halt. Before that happens, while memory is low, I also get:
[drm] GPU HANG: ecode 9:1:0xabfffff4, in Plex Transcoder [100772], reason: No progress on bcs0, action: reset
Relevant boot dmesg: