NickM-27
commented
1 year ago you're getting a lot of timeouts from the camera & go2rtc, I'd suggest updating to the latest version https://docs.frigate.video/configuration/advanced#custom-go2rtc-version
Closed Spectre5 closed 1 year ago
NickM-27
commented
1 year ago you're getting a lot of timeouts from the camera & go2rtc, I'd suggest updating to the latest version https://docs.frigate.video/configuration/advanced#custom-go2rtc-version
Spectre5
commented
1 year ago Thanks, I'll look into the custom go2rtc version. Do you know what might be the cause of this? The system isn't taxed, the camera is brand new, and everything is hardwired (with short CAT 6 cables). Also, at he very end of my long post, I've included some log output without using go2rtc. That log indicates timeouts too, as you've noted, but wouldn't that indicate an issue separate from go2rtc since it wasn't in use there? And that log indicates the follow a number of times as an error: "Timestamps are unset in a packet for stream 0. This is deprecated and will stop working in the future. Fix your code to set the timestamps properly"
NickM-27
commented
1 year ago "Timestamps are unset in a packet for stream 0. This is deprecated and will stop working in the future. Fix your code to set the timestamps properly"
this is a warning, not causing the stream to crash. The stream is crashing because the connection is timing out.
The fact that both subtype=0 and subtype=2 means both streams are timing out from the camera.
Spectre5
commented
1 year ago Ok, I'll try out the custom go2rtc. Is the instructions different when using docker compose? I don't have a /config directory. I've tried restarted Frigate with that go2rtc executable in the /media directory, /config directory (created it), and next to the docker compose YAML file, but in each case the version reported in the log is still 1.2.0.
NickM-27
commented
1 year ago you do have a /config directory, it is whatever is mapped to /config in your docker compose. You may have the config file mapped specifically, in which case the docker compose should be edited to map the entire folder instead of just the config file
Spectre5
commented
1 year ago Gotcha, just realized that now myself. I'll edit the docker compose to map it, try it out today, and report back. Thanks for the help.
NickM-27
commented
1 year ago I'd also suggest adding
logger:
logs:
frigate.record: debugto the frigate config which will give helpful info on the recordings
Spectre5
commented
1 year ago Great, the updated go2rtc is active and I've added that debug logging for frigate.record. I'll let it run awhile...
Spectre5
commented
1 year ago Using a new go2rtc as suggested by @NickM-27 seems to have fixed my issue. Out of curiosity, is there a reason that frigate does not include a newer release of go2rtc (past 1.2.0)?
NickM-27
commented
1 year ago Out of curiosity, is there a reason that frigate does not include a newer release of go2rtc (past 1.2.0)?
because newer versions weren't out when the current frigate version was released
Describe the problem you are having
I'm getting nearly non-stop warnings similar to "WARNING : Unable to keep up with recording segments in cache for camera1. Keeping the 5 most recent segments out of 6 and discarding the rest..." as well as "Discarding a corrupt recording segment: /tmp/cache/camera1-20230830031158.mp4".
Until I get a good and stable setup going, I'm working with only 1 camera (the recommended Loryta IPC-T5442TM-AS with the suggested settings at the bottom of the camera setup page). The storage is entirely local on a NVME drive, so my write speed should be as good as it gets. The camera is hardwired to the same switch as the computed hosting docker.
I'm running Frigate through docker on a dedicated LXC container in proxmox on a Lenovo M910Q (i7-6700T w/ 32GB RAM - the container given 4 cores, 8GB RAM, and 1 GB swap) and the google coral is setup and working well. The System page currently indicates less than 2% of CPU and less than 0.3% of memory for each of ffmpeg, Capture, and Detect (0 skipped). The storage shows 3.5 GB / 65 GB and the memory shows 7.6 MB / 1.05 GB for /dev/shm and 2.6 MB / 1000 MB for /tmp/cache. I list all of this to make it clear the system is not at all taxed and the storage writing should be as fast as it can be (NVMe)
I've tried clearing database and all media files as well as restarting docker (and the host), but the issue comes back within hours.
Version
0.12.1-367d724
Frigate config file
Relevant log output
FFprobe output from your camera
Frigate stats
Operating system
Proxmox
Install method
Docker Compose (in LXC Container)
Coral version
USB
Network connection
Wired
Camera make and model
Loryta IPC-T5442TM-AS
Any other information that may be helpful
I assume it is unrelated, but I am also running the Home Assistant Integration as well as the proxy addon.
The go2rtc log contains the following:
I've also tried to adjust the configuration to NOT use go2rtc. In this case, I get different errors in the frigate.log file: