I imagine this has to do with the ffmpeg update, but I'm not sure what's wrong or how to fix it. All streams are just giving "No frames have been received, check error logs"
Logs seem to indicate ffmpeg is busted. I'll attach logs and config.
#Frigate 15.0-beta1, all streams broken
sullen mason
log is truncated, but I suspect it gets the necessary bits across. If there's anything else I can provide to help out, let me know. Also if there's any general suggestions to improve my config, let me know, lol. There's a 1070 in the server running Frigate, but it doesn't look like Frigate's using it. Not sure it's even necessary, tbh.
wheat horizon
connection is refused 2024-11-04 17:08:43.352361382 [2024-11-04 17:08:43] ffmpeg.north_front.audio ERROR : Error opening input file rtsp://127.0.0.1:8554/north_front_sub.
doesn't seem related to ffmpeg updates, is go2rtc running?
did you update to HA core beta? if so that might be the issue
sullen mason
I didn't update HA in any way yet
How can I check if go2rtc is running?
I'm running HA in a container on Debian. Should I go ahead and update the WebRTC Camera addon through HACS?
wheat horizon
no, that is unrelated, I am talking about HA core
check go2rtc logs in the frigate ui
sullen mason
It definitely is having issues.
Not injecting WebRTC candidates into go2rtc config as it has been set manually
The go2rtc service exited with code 1 (by signal 0)
Preparing new go2rtc config...
Rinse and repeat
2024-11-04 17:40:05.791132822 curl: (7) Failed to connect to 127.0.0.1 port 1984: Connection refused
The go2rtc service is not responding to ping, restarting...
Those two also appear here and there
wheat horizon
if ha is running try stopping it then restarting frigate and see if it sticks
sullen mason
As in, leave HA down while Frigate boots up?
wheat horizon
yes
wheat horizon
might need to enable trace logs for go2rtc
sullen mason
Starting HA back up. How can I enable trace logs?
sullen mason
Well, frigate didn't complain about the config on boot, so I reckon that might've been the right config for it, but the go2rtc logs don't look any different
wheat horizon
frigate does not validate go2rtc config
sullen mason
If frigate is running on 192.168.1.90, shouldn't 192.168.1.90:1984 bring me to the go2rtc front end?
wheat horizon
sorry it is level not default
wheat horizon
If frigate is running on 192.168.1.90, shouldn't 192.168.1.90:1984 bring me to t...
only if you have mapped the port in docker
but of course that requires go2rtc to be running which it is not
wheat horizon
it must be failing on parsing the config
can you get a shell in frigate and run cat /dev/shm/go2rtc.yaml
sullen mason
Doesn't exist in /dev/shm/
Oh wait, I need a shell in the actual container don't I?
wheat horizon
yes
sullen mason
Need to remember how to do that. One sec, lol.
wheat horizon
but it is /dev/shm/go2rtc.yaml
sullen mason
Okay, I got into shell in the container, but that file doesn't exist.
I need to step away for dinner, I appreciate your time, Nick, I'll be back in a bit to resume tinkering, if you're still here then. o/
wheat horizon
Okay, I got into shell in the container, but that file doesn't exist.
that would be an issue
it should be created when frigate starts up, so go2rtc is not even starting
sullen mason
If it wasn't starting at all, wouldn't it not be building go2rtc-logs.txt?
sullen mason
Do you have any recommendations for what I can do to chase it down? Or should I just wipe the frigate container and rebuild it? I don't care about the existing DB, I want to wipe the current storage drive it's using anyway 'cause it somehow had stuff saved from 5 months ago, though the retention policy is 5 days. I was gonna give it a "clean slate" anyhow. So I can backup my config, backup my compose.yml, and just let it rebuild things from scratch. Thoughts?
wheat horizon
well, nothing in frigate is going to affect this
how exactly is frigate installed?
sullen mason
In an otherwise empty LXC in Proxmox. So is HomeAssistant.
("otherwise empty" meaning a debian LXC with nothing else on it).
wheat horizon
gotcha, so running in docker in debian LXC
sullen mason
Correct, yeah
sullen mason
You know, now that I think about it, I'm not sure I've tried restarting the frigate LXC, lol. Let me try that in case any ports are stuck in use or something.
It is strange that it's been running fine this way for a year or so (when I converted everything to Proxmox), and it died on me when I updated to the beta.
Rebooting the LXC did not work
outputs of lsof:
docker-pr 630 root 4u IPv4 2023059195 0t0 TCP *:8555 (LISTEN)
docker-pr 637 root 4u IPv6 2023044628 0t0 TCP *:8555 (LISTEN)
docker-pr 651 root 4u IPv4 2023049863 0t0 UDP *:8555
docker-pr 658 root 4u IPv6 2023059198 0t0 UDP *:8555
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
docker-pr 675 root 4u IPv4 2023059207 0t0 TCP *:8554 (LISTEN)
docker-pr 682 root 4u IPv6 2023046849 0t0 TCP *:8554 (LISTEN)
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
docker-pr 724 root 4u IPv4 2023041424 0t0 TCP *:1984 (LISTEN)
docker-pr 730 root 4u IPv6 2023041427 0t0 TCP *:1984 (LISTEN)```appears something else is listening on the port
try to figure out what those processes are
sullen mason
Hmmm, all I'm getting is that they're docker processes, and docker ps -a only lists Frigate and portainer (forgot that was installed - it's using portainer's default port, 9443)
wheat horizon
so what are the go2rtc logs now?
nothing should have the port if it is not running
sullen mason
Same as they were the other day, no change.
wheat horizon
get a shell in the frigate container and run python3 /usr/local/go2rtc/create_config.py
sullen mason
Traceback (most recent call last):
File "/usr/local/go2rtc/create_config.py", line 168, in <module>
ffmpeg_cmd = f"exec:{parse_preset_hardware_acceleration_encode(ffmpeg_path, config.get('ffmpeg', {}).get('hwaccel_args'), input, '-rtsp_transport tcp -f rtsp {output}')}"
File "/opt/frigate/frigate/ffmpeg_presets.py", line 227, in parse_preset_hardware_acceleration_encode
return arg_map["default"].format(input, output)
IndexError: Replacement index 2 out of range for positional args tuple```oh, I wonder why that wasn't showing up in the logs
this is fixed in the next beta
but for now you should just set hwaccel_args globally
sullen mason
I had trouble getting hwaccel set up in the past, I can't remember what trouble I ran into. I'll look into it, but if it's giving me trouble again, could I run a dev build to get 'round it?
Welp, a quick hwaccel_args: preset-nvidia-h264 seems to have got things semi-working, lol. The homepage with all the thumbnails of video feeds is still busted (no frames have been received), clicking on them opens up a full view that is functional, strangely. Feeds are also coming up in HA now.
new logs from frigate
wheat horizon
you can also just set hwaccel_args to [] to disable it. Woudl likely be good to get it working though if you have the hardware