3

I am trying to configure wireguard to work as a VPN server. The main problem is, that the gateway only forwards the VPN server ip to other server, not my client IP.

My setup is the following:

                                                        - server A (10.10.0.4)
                                                      /
CLIENT (10.10.1.3) -> wireguard server (10.10.1.2) -- 
                                       (10.10.0.2)    \
                                                        - server B (10.10.0.3)

The wireguard server runs on a machine with two interfaces:

  • eth0 (10.10.0.2)
  • wg0 (10.10.1.2)

When the VPN connection is established, I am able to connect to server A and server B (via ssh). The problem is, that the IP address of the wireguard server is forwarded (nat) to server A and B. Logged in via ssh shows me every time, that the last connection came from 10.10.0.2 (on server A and B). But on the wireguard server, the last logged in IP is my real client IP (10.10.1.3).

What I am trying to do is to configure wireguard so that my IP (10.10.1.3) is forwarded correctly to server A and B.

This is my client wireguard config file:

[Interface]
PrivateKey = xxx
Address = 10.10.1.3/24
DNS = 10.10.0.2, 8.8.8.8

[Peer]
PublicKey = XXX
AllowedIPs = 10.10.0.0/24
Endpoint = xxx.xxx.xxx.xxx:41194
PersistentKeepalive = 15

My wireguard server (wg0.conf) config:

[Interface]
Address = 10.10.1.2/24

## My VPN server port ##
ListenPort = 41194

PrivateKey = xxx

# Internet Gateway config: nat wg1 out to the internet on ens10
PostUp = iptables -A FORWARD -i wg0 -j ACCEPT; iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE
PostDown = iptables -D FORWARD -i wg0 -j ACCEPT; iptables -t nat -D POSTROUTING -o eth0 -j MASQUERADE

[Peer]
## Desktop/client VPN public key ##
PublicKey = xxx

AllowedIPs = 10.10.1.3/32

I guess that the iptables configuration is wrong, because of the nat / MASQUERADE, but I wasn't able to configure the gateway correctly.

I appreciate your help.

Update

Executing on server A (same on B)

ip -br link; ip -br address; ip route

Returns (public IP is masked):

lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
eth0             UP             96:00:01:29:d6:9b <BROADCAST,MULTICAST,UP,LOWER_UP>
ens10            UP             86:00:00:08:9c:c5 <BROADCAST,MULTICAST,UP,LOWER_UP>
lo               UNKNOWN        127.0.0.1/8 ::1/128
eth0             UP             10.10.0.3/32  fe80::9400:1ff:fe29:d69b/64
ens10            UP             49.xxx.xxx.xxx/32  2a01:xxx:xxx:xxx::1/64  fe80::8400:ff:fe08:9cc5/64
default via 172.31.1.1 dev ens10 proto dhcp src 49.xxx.xxx.xxx metric 100
10.10.0.0/16 via 10.10.0.1 dev eth0
10.10.0.1 dev eth0 scope link
172.31.1.1 dev ens10 proto dhcp scope link src 49.xxx.xxx.xxx metric 100
0

1 Answer 1

2

NAT is done by the configuration so you get NAT as asked. To avoid using NAT you have to:

  • make sure end servers A and B have a real route back to the client

    If that's not the case, add at least this on A and B (if running Linux):

    ip route add 10.10.1.3/32 via 10.10.0.2

    UPDATE: OP's routing setup (in a cloud) makes A and B's traffic to 10.10.0.2 (or even between each others) go through an additional router 10.10.0.1 (part of the cloud network). So the route had to be added on this part, as confirmed by OP.

  • remove NAT on wireguard server

    Just remove the second iptables commands in the two WireGuard's PostUp and PostUp configuration, and make sure there's not a previously added entry left, by running this time only:

    iptables -t nat -D POSTROUTING -o eth0 -j MASQUERADE
    
  • optional: update AllowedIPs on client

    If client wants to access wireguard server using server's address on its tunnel side rather than its eth0 side, or to be sure ICMP sent back by the wireguard server are received (eg: to get traceroute to server A working without * * *), 10.10.1.2 should also be in AllowedIPs to satisfy WireGuard's cryptokey routing.

    Replace on client:

    AllowedIPs = 10.10.0.0/24
    

    with:

    AllowedIPs = 10.10.1.2,10.10.0.0/24
    
4
  • Thanks for your response. I tried to add the IP route on both server (ip route add 10.10.1.3/32 via 10.10.0.2), but I only got an error: "Nexthop has invalid gateway". It seems, that there might be a network misconfiguration?!
    – Daniel
    Mar 25, 2022 at 18:46
  • I'll updated my question to provide the result of ip -br link; ip -br address; ip route
    – Daniel
    Mar 25, 2022 at 19:05
  • With the added information: all traffic for 10.10.0.0/16 which would usually be LAN traffic is routed through 10.10.0.1 leaving no LAN traffic on server A or B that could have exceptions. The route has then to be added on 10.10.0.1 which is not appearing anywhere in the question.
    – A.B
    Mar 25, 2022 at 19:28
  • 1
    Thanks for pointing me into the right direction. 10.10.0.1 is not completely out of control. It's a cloud network, but I added a route from 10.10.1.0/24 to 10.10.0.2. Now it's working perfectly. Thanks!
    – Daniel
    Mar 25, 2022 at 20:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .