My setup is basically the following:
Windows Server Host
- Physical Ethernet to physical LAN
Fedora 26 Server VM
- LAN-br0 - Bridge for access to the physical ethernet on the Windows Server.
- LAN-eth0 - eth0 adapter which is part of LAN-br0 and is the server’s access point to the physical LAN.
- container1-veth0 - Container 1’s virtual ethernet paired adapter which is in LAN-br0 for access to the physical LAN.
- container2-veth0 - Same as Container 1’s but for 2.
Container 1
- LAN.veth0 - Virtual Ethernet adapter for physical LAN connectivity. Paired to container1-veth0 on Fedora 26 Server.
Container 2
- Same as Container 1, but for Container 2.
I have the exact same setup on a Fedora 26 Workstation for testing, and it works fine for my purposes. The difference is the second layer of virtualization in that my Fedora 26 Workstation is the host OS.
My desire is to give the containers direct access to the network without NAT/masquerading, so just using LXC’s ability to auto-handle networking is not what I need.
I’ve been using Network Manager’s CLI to set this up, and AFAICT every setting is the same between the Fedora Server and Workstation, but one gives me LAN access while the other does not.
I can ping the Fedora Server from the containers, but nothing else on the network. I think this is key. It might be a configuration difference between Fedora Server and Fedora Workstation’s SELinux or firewall.
I never made an exception for either container on my Workstation to give them access though, so I don’t think it is that.
Any help is appreciated.