Frigate + Nvidia on new setup #21079
Replies: 3 comments 6 replies
|
Based on the error logs showing The key problems I can identify: 1. LXC Container ConfigurationYour LXC container needs proper device passthrough for Nvidia. Based on similar setups, you need to add these lines to your LXC container configuration file (typically 2. Docker Compose ConfigurationFor the services:
frigate:
runtime: nvidia
environment:
- NVIDIA_VISIBLE_DEVICES=all
- NVIDIA_DRIVER_CAPABILITIES=all3. Hardware Acceleration ConfigurationFor Nvidia GPUs using the TensorRT image, the correct ffmpeg configuration is(3): ffmpeg:
hwaccel_args: preset-nvidia4. Detector ConfigurationFor the ONNX detector with Nvidia GPU in the The Recommended next steps:
📚 Sources:
Have another question? Just tag @inkeep. |
|
Your logs don't contain any errors, in what way is it not working? |
Hi 343816941 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 1173, index: 19, mask: {20, 44, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set. Frigate starts, runs for a few seconds while cameras getting loaded and reboots. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Hi, just moved my setup to new pc with nvidia and struggling for the last few days with set it up.....
Frigate starts but is keep reseting itself in loop.......... ;(
Installation on proxmox in LXC container
Thanks in advance for any support!!!
Logs.....
My compose.yml
My configuration.yml:
All reactions