Docker Engine Not Using Correct Network

[kevin_h]

I am running Unraid 6.8.3 Stable. I recently upgraded from a Dual Xeon server board to a Ryzen 5 3600 (better performance, less power usage). I added in a new Dual 1GB NIC since the new motherboard only has 1 NIC on-board. I set the on-board NIC to be a single interface with a Static IP address with bonding and bridging disabled. I then set the new Dual NIC with Bonding and Bridging Enabled and added another Static IP to the new bond. In the Docker settings pane, I set the new bridge as the default interface to use.

 

I have set the Interface rules so that all the interfaces get the same identifier each time which is working.

• eth0 - Onboard NIC, Bond & Bridge disabled
• eth1 - Port 1 of Dual NIC, bond enabled (eth1 & eth2 are members), bridge enabled
• eth2 - Port 2 of Dual NIC, member of eth1 bond
• eth3 - Port 1 of 10GB SFP card (No issues, DAC to Plex server with static IP)
• eth4 - Port 2 of 10GB SFP card (Currently not in use)

 

The issue is Docker is still using the eth0 interface for all of it's networking when Bridge is selected for the containers, even if Bridge and Bond is disabled. Is Docker set to use eth0 all the time? When I go to the Docker page in the WebUI it is showing the IP address for the non-bonded on board NIC. Do I need to specify the new bridge for each container instead of using the Bridge selection in the drop-down?