Issue with multiple wireguard configs (ProtonVPN) at the same time

Started by Unsocial_Firewall, November 15, 2026, 07:35:00 PM

Previous topic - Next topic
Hey guys,

I set up ProtonVPN following this guide: https://forum.opnsense.org/index.php?topic=38550.0

Everything worked pretty well until I recently tried adding a second instance of ProtonVPN (with a new config).

ProtonVPN's documentation says you can run multiple WireGuard instances at the same time by incrementing the IP at a certain point: https://protonvpn.com/support/wireguard-configurations

So for example, instead of 10.2.0.1, you can use 10.3.0.1 for the gateway and DNS.

I spent quite a while debugging why the second instance wasn't working, until I noticed that traffic to 10.3.0.1 was going out over my default gateway instead of the WireGuard tunnel, because the route was missing.

Once I added the route manually, everything worked for a few minutes, including my gateway monitoring. But after a while it stopped working again, and I can't figure out why.

Any help would be really appreciated, or even just hearing about other people's experiences with this.

November 15, 2026, 07:44:22 PM #1 Last Edit: November 15, 2026, 07:46:36 PM by Unsocial_Firewall
Just to clarify: the first instance has never had any issues, while for the second one I'm using the gateway/DNS address (10.3.0.1) as the monitoring target.

I did everything again from the guide, just for the second instance, so the settings should not overlap.

The Wireguard handshake also does not appear to be the source of the problem


November 15, 2026, 10:23:01 PM #3 Last Edit: November 15, 2026, 10:35:24 PM by Unsocial_Firewall
The tunnel IP for the first instance, 10.2.0.1/32, did set the route properly until I disabled the static routes.

I changed the tunnel IP from 10.3.0.1/32 to /30 .

I'm a bit confused: the routes now look correct, and everything works when I set both the first and second instances to a /30 subnet.

For some reason, only a single WireGuard instance at a time sets an additional route to the gateway/DNS address. Right now it's 10.3.0.1; before, it was 10.2.0.1. I am not using any manual routes right now.

My routes now look like this:


10.2.0.0/30        link#14            U               wg0
10.2.0.2           link#3             UHS             lo0
10.3.0.0/30        link#15            U               wg1
10.3.0.1           link#15            UHS             wg1
10.3.0.2           link#3             UHS             lo0


Can I leave them like this?  also I have issues with an additional third ipv6 wg so I should also just change this 2a07:b944::2:2/128 to /126?

Quote from: Unsocial_Firewall on November 15, 2026, 10:23:01 PMat a time sets an additional route to the gateway/DNS address.
Monitor IPs get a route added automatically.

Quote from: Unsocial_Firewall on November 15, 2026, 10:23:01 PMCan I leave them like this?
Sure, why not if it is working.

Quote from: Unsocial_Firewall on November 15, 2026, 10:23:01 PMso I should also just change this 2a07:b944::2:2/128 to /126?
Never done it myself (IPv6 VPN) but it is looking good to me.

November 15, 2026, 10:48:39 PM #5 Last Edit: November 15, 2026, 10:55:44 PM by Unsocial_Firewall
Thanks a lot! Yup, everything seems to work for now, and I'll do some more testing.

QuoteSure, why not, if it is working.

Well, I barely know anything about WireGuard, so I don't know all the ramifications of changing the provided subnet from ProtonVPN from /32 to /30.


The ipv6 works now aswell but there still is no static route for 10.2.0.1... (Even though it's the monitor ip)

Quote from: Unsocial_Firewall on November 15, 2026, 10:48:39 PMstill is no static route for 10.2.0.1... (Even though it's the monitor ip)
If it is local itself, then it doesn't need one. Try external IPs.

Did you create additional source nat rules for the second wireguard tunnel? I tried once I think to do it as a 'group' but connections were flaky until I separated it all out.

I run several ProtonVPN and Mullvad WireGuard tunnels at the same time in a failover group, so here's what has worked for me:
/30 vs /32: with a /30, Proton's 10.x.0.1 sits on the tunnel's own subnet, so the firewall treats it as directly connected. That's why it doesn't get a separate route, as Bob.Dig said.

Monitor IP: for the same reason, monitoring 10.x.0.1 only proves the tunnel is up, not that traffic gets past Proton to the internet. I use a different public IP for each tunnel. Each one gets a host route through its own tunnel, so it shouldn't also be one of the DNS servers your clients use.

Handshake vs real traffic: a recent handshake only means the encrypted session is up. dpinger tests the whole path. After any change I test both ICMP and a real TCP connection, because I've seen one work while the other failed.

NAT: +1 to EOC_Jason. One outbound NAT rule per tunnel.

Debugging: Firewall → Log Files → Live View has been far more reliable than tcpdump on the wg interfaces, which can be misleading on FreeBSD.

One Proton-specific thing: some Proton servers reply to the shared 10.2.0.2 address instead of your tunnel address, so dpinger shows 100% loss even though the handshake is fine. Those servers need BINAT; others don't. I wrote this up here: https://forum.opnsense.org/index.php?topic=52712.0 (see reply #3 for the per-server behavior and the Live View check). Since then I've also seen a server's behavior change over time with no change on my side, so if a tunnel suddenly shows down with a fresh handshake, check Live View again before changing anything else.