jordandrako

I never narrowed it down completely, but I did bandaid the problem. I believe the issue was caused by one of my reolink doorbells with a poor wifi signal causing the ffmpeg process to crash and hang, and for some reason it kept eating ram. I changed from rtsp to flv (look at reolink specific config on the frigate docs), which decreased the crashing severity, and limited the amount of ram frigate can use so if a memory leak does happen, it doesn't take all the server's ram and crash it.

I woke up this morning with my server not working and unable to access the web gui or ssh. I had to hard reset to get it back up. I was monitoring the system logs and it froze up again. I waited a couple minutes and the logs started showing again and I noticed some out of memory errors as well as it killing some ffmpeg processes. I believe it's either frigate or plex, both are set to use hardware transcoding using a GTX-1080. They have both been running fine until last night. I can't narrow it down, can somebody help me figure what container is eating the ram, and how I might solve it? Edit: I've narrowed it down to frigate, I think. I'm not sure why frigate started crashing my server now after running fine for weeks though. I'll bring my logs over to the frigate repo. If anyone here has run into similar ffmpeg crashes in frigate on unraid that would still be helpful. mycelium-diagnostics-20240116-0959.zip