[Support]: ffprobe execution does not specify rtsp_transport, resulting in frigate failing to complete startup for cameras not responding to UDP #22906
Replies: 5 comments 5 replies
-
|
This is a known behavior where Frigate uses OpenCV (not ffprobe) to determine stream dimensions, and OpenCV uses ffmpeg backend which defaults to UDP for RTSP transport. (1) A workaround that has been confirmed to work is setting the environment variable on your container: This forces OpenCV to use TCP when opening RTSP streams, which resolves the hanging issue for cameras that don't respond to UDP. (1) For your Helm chart configuration, add this to your environment: env:
OPENCV_FFMPEG_CAPTURE_OPTIONS: "rtsp_transport;tcp"Alternatively, you can use go2rtc to restream your camera and have Frigate connect to the local restream instead. This approach reduces direct connections to your camera and can help avoid protocol compatibility issues. (2) go2rtc:
streams:
doorbell: rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASS}@reolink-doorbell:554/Preview_01_sub
cameras:
doorbell:
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/doorbell
roles:
- detectFor Reolink cameras specifically, they should be connected via RTSP, and the 📚 Sources: Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
This behavior is explicitly called out in the release notes and breaking changes for Frigate 0.17. You should explicitly define a |
Beta Was this translation helpful? Give feedback.
-
|
@hawkeye217 The behavior mentioned in the release doesn't seem to reference the transport mode, nor the outcome of an indefinite hang. Setting Something I'll note is that having an indefinite hang here that will go for 10+ minutes is a poor failure mode, as it gives few clues to users the underlying cause of why the container doesn't seem to launch properly. Even if the symptom can be resolved through env settings, I'd still argue an indefinite hang is a bug? |
Beta Was this translation helpful? Give feedback.
-
|
Adding the following to helm yaml values resolves this issue for me: As noted above, I'd still somewhat argue that an indefinite hang with little indication of why is still a bug. That said, I don't feel super strongly about this. Fine with maintainers closing this if the the indefinite hang issue is known or being addressed elsewhere. |
Beta Was this translation helpful? Give feedback.
-
|
I've put up a PR to add a log message and clarity to the docs for any users who may come across this issue in the future. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Checklist
Describe the problem you are having
Upon launching the frigate container with the config included, an
ffprobeprocess is launched.ffprobeis not being set with-rtsp_transport tcp. This results in it indefinitely hanging for my cameras, which don't listen for UDP. This in turn seems to be preventing frigate from starting up properly. I've confirmed theffprobeprocess runs in the container without restarting or halting for at least 10 minutes.The ffprobe process which the container is running is:
Running that command manually in the container reproduces the indefinite hang. By modifying the command and passing
-rtsp_transport tcp, the probe completes successfully.Steps to reproduce
Version
0.17.1@sha256:1724960349dad0bd2ae8ec884171a6fd5755a4dc242a0e66cadbda9c0e85c99b
In which browser(s) are you experiencing the issue with?
Any
Frigate config file
docker-compose file or Docker CLI command
Relevant Frigate log output
Relevant go2rtc log output
Included in container logs.Operating system
Debian
Install method
Docker CLI
Network connection
Wired
Camera make and model
Reolink Doorbell WiFi
Screenshots of the Frigate UI's System metrics pages
UI does not start.
Any other information that may be helpful
No response
Beta Was this translation helpful? Give feedback.
All reactions