Running VPN clients in multiple containers screws up networking

Hi,

We have a need to run VPN clients (SoftEther, OpenVPN, StrongSWAN) in multiple Docker containers. When I started out I was running with the default bridge network and when the VPNs connected it would screw up the whole container’s connectivity so it would keep disconnecting from the VPN.

After reading that using the default bridge network was a bad idea, I created a user bridge network and used that. Everything worked just fine then.

But then I needed to run multiple instances of that container all running VPN clients. With just one running , everything was fine, but when more than one was running, the networking would get screwed up again just like when I was using the default bridge network.

I have tried this both with all containers using the same bridge network, and also giving them each their own separate network. Neither way works.

Docker seemed like the perfect way to do what we wanted, but if we can’t run more than one instance without the network screwing up, then it’s no good :frowning:

Does anybody have any idea why this would be happening? We are using Docker version 18.09.6, build 481bc77156, on CentOS 7.6.1810.

Thanks.

1 Like