Can't get DUAL NIC setup to work

[Eightball1990]

Hi,
I have Frigate installed as container in docker running on a Unraid Server.
This server has its main LAN connection on eth0, which is the motherboard network interface.
I added a PCI NIC to the server, assigned eth1 by unraid, which is connected to a POE switch interfacing with my cameras only.
All cameras have been assigned a static IP 10.1.1.x , i configured eth1 to have a static IP 10.1.1.1

when i try to setup a single camera , the yaml saves but i receive no frames;

please find attached the error logs.
frigate-logs.txt
Can anyone help me to determine what is wrong, please?

The frigate yaml configuration is as follows:
`mqtt:
enabled: false

cameras:
#Camera 1: Carport
Carport:
enabled: true
ffmpeg:
inputs:
- path: rtsp://[username]:[password]@10.1.1.65:554/rtsp # <----- The stream you want to use for detection
roles:
- detect
detect:
enabled: false # <---- disable detection until you have a working camera feed
width: 1280
height: 720

detectors:
coral1:
type: edgetpu
device: pci:0
coral2:
type: edgetpu
device: pci:1
detect:
enabled: true
version: 0.16-0
semantic_search:
enabled: false
model_size: small
face_recognition:
enabled: true
model_size: large
lpr:
enabled: true
classification:
bird:
enabled: false`

[inkeep-ai-assistant[bot]]

---

The issue is likely related to Docker networking rather than Frigate configuration. When running Frigate in a Docker container on Unraid with a dual-NIC setup, the container needs proper network access to reach cameras on the isolated network (10.1.1.x).

Your Frigate configuration looks correct for a basic camera setup(1). The "no frames received" error typically indicates the container cannot establish a network connection to the camera at 10.1.1.65.

For Docker installations, you'll need to ensure the container can route to your camera network. The standard Docker Compose configuration uses bridge networking(2):

services:
  frigate:
    container_name: frigate
    restart: unless-stopped
    image: ghcr.io/blakeblackshear/frigate:stable
    network_mode: bridge
    ports:
      - "8971:8971"
      - "8554:8554"

(2)

However, with your dual-NIC setup where cameras are on an isolated network (10.1.1.x via eth1), the container likely cannot reach that subnet through standard bridge networking.

Potential solutions:

1. Use host networking mode - This gives the container direct access to all host network interfaces including eth1. Change network_mode: bridge to network_mode: host in your Docker Compose configuration(2)

2. Create a custom Docker network - Configure Docker to use a macvlan or ipvlan network that bridges to eth1, allowing the container to communicate directly with the 10.1.1.x subnet

3. Add routing - Configure routing rules on your Unraid host to allow the Docker bridge network to reach the 10.1.1.x subnet through eth1

The first option (host networking) is typically the simplest for troubleshooting. Once you confirm the camera feed works, you can refine the networking setup if needed.

Check the Frigate Docker logs (docker logs frigate) for specific ffmpeg connection errors that will confirm whether this is a network reachability issue.

📚 Sources:

1. Getting Started
2. Installation

---

Have another question? Just tag @inkeep.

[Eightball1990]

changing network_mode from bridge to host in the frigate container settings doesn't solve the problem.

[NickM-27]

The best way to diagnose this is to check within the Unraid CLI if you can ping the camera, and then within the Frigate container

[cardinjt]

Not sure of the data you have entered in your copy for "PCI Network Interface Card". The address 10.1.1.0 with a /24 subnet mask appears to describe an entire subnet, not an individual device... I would have expected that you would have needed to specify the specific static IP Address "10.1.1.1" as the address for the NIC in this spot ? (but admit I am not familiar with your Unraid config screens). I did what you are trying to do under Debian 13, and roughly followed process similar to what is described here "https://devtutorial.io/how-to-configure-static-ip-address-on-debian-12-p3121.html" performing config of the 2nd static NIC address using the command line. In any case - after you setup the static address for the NIC - and the IP addresses "10.1.1.x" for the individual cameras... you should be able to access / ping the cameras from the Unraid CLI (or desktop if you have one there) ... After that configure Frigate docker container etc.

[Eightball1990]

Hi NickM and cardinjt, thank you both for making time to support me. I performd a ping to one of the cameras from the unraid CLI and see that it is not reachable, so i guess my network configuration for the NIC is not right yet:

regarding the "PCI Network Interface Card" configuration, i figured that i should set it up similarly as done for the main LAN interface eth0, but without specifying a default gateway.
this is my main LAN eth0 config:

and as posted before this is my camera subnet eth1 config:

What would you suggest that I try next?

some more network ip info:

[Eightball1990]

Hi, thanks for your help! I'm getting somewhere now!
By setting the IP addess for eth1 (my second NIC) to 10.1.1.1 and also specifying this als default gateway in the Dahua Camera TCP/IP settings of one of my cameras, i am able to ping that camera from from my unraid server!

However when i try to add that camera to Frigate it seems that it is not found, i see following errors in the system log:
Error opening input file rtsp://:@10.1.1.66:554/rtsp.
ERROR : Error opening input files: Server returned 404 Not Found
Attached my frigat system logs:
frigate-logs (1).txt.

my container network type is still set to host.

the container IP is the one of my server, maybe that is the problem?

Should i setup my container differently?
Also how can i ping my cameras from inside the container? if i open the console from right clicking the frigate logo i get a console that doesn't seem to recognise the command ping (sorry if i am doing somethign stupid, i'm not linux savvy):

[cardinjt]

You can try to install 'ping' inside your container, as apparently it is not provided by default. Or try some other networking tools like 'wget', or 'curl' from inside container, using your camera IP, to see if they provide expected response. If not, then you need to adjust container config so that the IP Address range of your cameras passes thru. Ask "chat gpt" or equivalent (maybe the frigate AI bot on the documentation pages), maybe an Unraid forum, for options.... I'm not knowledgeable of Unraid specifics to provide further insight.

[Eightball1990]

thanks cardinjt for the great help so far. I understand that you are not able to help out with unraid itself. I followed your suggestion to ask help to an AI bot; with the help of Gemini i believe i managed to pass both network interfaces to the frigate container. The frigate container now also has an IP address on the camera subnet (i.e. 10.1.1.2).

After following these instructions https://stackoverflow.com/questions/49463719/docker-ping-executable-file-not-found-in-path-unknown to install the ping utility , i was able to ping the camera from within the container (at least i think it is from within the container):

The errors in frigate
Error opening input file rtsp://:@10.1.1.66:554/rtsp.
ERROR : Error opening input files: Server returned 404 Not Found
still persist though...
Any idea why?

[Eightball1990]

I got it to work finally! (with the help of Gemini)
Turned out that also the URL structure for my dahua camera needed to be different than was suggested by default by frigate;
i had to use:
rtsp://admin:password@ip_address:554/cam/realmonitor?channel=1&subtype=0

so in summary, the steps i had to take to solve the original problem:

1. in unraid network settings configure eth1: set the IP addess for eth1 (i.e. the second NIC) to 10.1.1.1 / 24

2. in camera settings, set static ip addresses on camera in camera subnet range, e.g. for me 10.1.1.x, put default gateway to 10.1.1.1

3. define a docker macvlan for br1 (eth1). i did this by puting in the cli:
   docker network create -d macvlan --subnet=10.1.1.0/24 --gateway=10.1.1.1 -o parent=br1 cam_network

4. then i added the following to the post arguments in the frigate docker container settings:
   && docker network connect cam_network frigate

5. In order to be able to ping my camera, i first had to install the ping utility in the frigate container, as it seemed to not have it by default, i did so by following these instructions i.e. putting following commands in the CLI:
   docker container exec -it frigate bash
apt update
apt install iputils-ping
exit

6. finally in frigate itself i used following code

mqtt:
  enabled: false

cameras:
  #Camera 2: Side
  Side:
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://admin:password@ip_address:554/cam/realmonitor?channel=1&subtype=0 # <----- The stream you want to use for detection
          roles:
            - detect
    detect:
      enabled: false # <---- disable detection until you have a working camera feed
      width: 1280
      height: 720

ffmpeg:
  hwaccel_args: preset-vaapi

i hope this will help someone else who also wants to achieve the same as me!