OpenVPN Client and Server on same machine - Server doesn't allow connections when client is connected (more detailed)

Question

I have exactly the same problem like the user khofm asked in https://unix.stackexchange.com/a/352968/394150

So in summary:

The openVPN client and the openVPN server instance work fine as long as I don't run them simultaneously. As soon as the openVPN client has a connection to the Paid VPN provider (in my case NordVPN), the WAN clients are unable to connect to the openVPN server.

My system is a Raspberry Pi 4 with dietpi (Debian Buster).

I'm very new to this community and this is my first question, therefore I can't add a comment on khofm question. My understanding of networks (especially about routes) is not very good, therefore I would like to know from khofm if he could explain his working setup in more detail with some explanations.

How does the OpenVPN server.conf looks, how about the client.ovpn of the Paid VPN Provider?

How do you separate the OpenVPN Server and OpenVPN Client in tun0 and tun1 with the two different subnets?

With the provided solution I could not solve my problem, maybe because I don't understand it that good. I wasted already serveral hours without any success. And I don't want to give up especially if I know that there is a solution. Please help me!

Thanks a lot in advance!

Answer 1

This is how a paid VPN service works (NordVPN for example):

1. You server's original public internet IP address is 1.1.1.1 at eth0 interface (for example)
2. When your server connects to a paid VPN, it becomes a client to NordVPN and all traffic is routed through the tunnel (tun0 interface) and breakout at provider's IP that is 2.2.2.2
3. At the same time, when WAN clients try to connect to your VPN server via 1.1.1.1 as per normal, their return traffic is being routed through NordVPN's tunnel and attempt to breakout at 2.2.2.2
4. As a result, tun1 will fail to establish because return packets are being lost and dropped.