[TriLUG] Docker: network bridging with separate interfaces

[Ron Kelley via TriLUG]

Greetings all,

Started working on docker recently for a side project, and I have hit a road block.  Hoping someone from the TriLUG might be able to help.

I have a server (VM - Ubuntu 16.04) running docker 2.12 with two network interfaces (eth0 and eth1).  Per the docker details online, all traffic to/from the containers go across the docker0 bridge, which evidently, always goes over eth0 (or default GW from the best I can understand from the online readings).  My goal is to dedicate eth0 as the management interface and use eth1 for all docker container traffic.

I read about using the “brctl” command to create a new bridged interface specifying a particular interface.  However, this does not appear to work.  I created the new bridge interface, spun up a test container using the new bridged interface, shutdown the physical interface on the host, and ran a traffic tool (ping).  I expected traffic to stops since the bridge’s physical interface was down.  Unfortunately, the traffic still goes on via eth0.  Here are the commands I used:

----------------------
brctl addbr QA_Bridge eth1
docker network create -d bridge --subnet=192.168.1.0/24 --ip-range=192.168.1.0/24 --gateway=192.168.1.1 -o com.docker.network.bridge.name=QA_Bridge QA_Bridge 
docker run —network QA_Bridge -d -p80:80 nginx/nginx
----------------------


I have spent a ton of time looking over online examples, but still can’t figure out how to get this done.  Again, I want to use eth0 for server management only and use eth1 for data traffic.

Any help/pointers/clues?


Thanks.