Closed kirsch33 closed 1 year ago
Wyze cams cause many problems due to their poor implementation of streaming (regardless of rtsp firmware / type). I'd highly suggest not using the wyze cam with go2rtc or at least not using the wyze cam restream as the source for frigate cameras section
Wyze cams cause many problems due to their poor implementation of streaming (regardless of rtsp firmware / type). I'd highly suggest not using the wyze cam with go2rtc or at least not using the wyze cam restream as the source for frigate cameras section
but ultimately it is either go2rtc or frigate (not sure which) that is suffering from the runaway memory usage regarding the ffmpeg stream and causing the host to crash (no doubt caused from a faulty stream on a cheap camera), but is there anyway to identify when this issue is occurring and implement a failsafe?
we have failsafes in place for many stream conditions but these conditions only do so much. At a certain point the failsafe is literally just to stop running the camera. Could be done but that logic doesn't exist currently.
Frigate kept crashing for me as well until I stopped using the built in go2rtc and switched back to running go2rtc in its own docker container. I'm not using Wyze cams but I am using Xiafang cameras (the camera Wyze is cloned from) flashed in dafang hacks.
Most likely it's just that go2rtc was running a newer version. Many users have found updating the internal go2rtc to fix the issue
Are there directions on updating the internal go2rtc?
That was fast, thanks!
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
Describe the problem you are having
so i have recently added 2 wireless cameras to my config recently, a cheap Wyze Pan v1 with Dafang firmware and a new Amcrest link here. My other cameras are hardwired Amcrest that i've not ever previously had any issues with
since adding these cameras, ive noticed a steady stream of ffmpeg errors in the frigate logs. it seems that most of the time these errors are inconsequential and recoverable since the streams are available 90% of the time (occasionally have to reload a couple times) but otherwise i would only know about these errors from always keeping an eye on the log file.
however, about every 24-36 hours i have been noticing the entire host that Frigate is on will lock up entirely for about 5 minutes or so (which affects almost every other service on my network) and i can trace back the error to ffmpeg out of memory:
this is a huge issue because it interrupts every VM that this host runs. the VM that runs Frigate has been given 48GB memory so I dont think its an actual resource issue.
i have tried using a custom go2rtc version and the results are the same. i currently use 1.5.0
Version
0.13.0-ca7790f
Frigate config file
Relevant log output
go2rtc logs only show an i/o timeout around the same time as the logs above
Operating system
Proxmox
Install method
Docker Compose
Coral version
PCIe