Use wireshark and listen on your ethernet interface. When you use tailscale, are the packets coming in/out from the tailscale server IP or the VPN ip? Check through the ip route routing table and figure out which pathway a packet will take in each use case. Might need to add a route exception specifically for the tailscale server IP to go out on the ethernet device.

otherwise empty bike lane

Over here in New York, everyone got an e-bike and now we get bike jams in the bike lane during commute hour. Dunno how I should feel about it. Aladeen? :(: Still faster than a car for sure.

PostUp = ip route add 100.64.0.0/10 dev tailscale0

Looks like you need to stick this line in the tailscale service file, since it’s the only time that the existence of the tailscale0 device is guaranteed. If you don’t want to modify the service file inside the package, could you write your own systemd service file and include the tailscale service as a prerequisite?

Also make sure that when you start the VPN first and then tailscale, you don’t get a double tunnel situation where tailscale goes out through the VPN (unless that’s what you wanted).

The exact script would depend on the use case; you’d use commands something like this:

mkdir -p /etc/netns/VPN
sh -c 'echo nameserver 1.1.1.1 > /etc/netns/VPN/resolv.conf'
ip netns add VPN
ip link add tun1 type wireguard
ip link set tun1 netns VPN

Because the wireguard device was created in the default namespace, it will “magically” remember its birthplace, even after you move its mouth (the tun1 device) to a separate namespace. The envelope VPN packets will keep going in/out in the default namespace.

ip netns exec VPN wg setconf tun1 /etc/wireguard/vpn.conf
ip netns exec VPN wg set tun1 private-key /etc/wireguard/vpn-key.private
ip -n VPN addr add 192.my.peer.ip/32 dev tun1

Get the wireguard config file from the VPN website, both mullvad and OVPN have a wizard to generate them. Your assigned private network ip is in the config file. Also get and save your device key.

ip -n VPN link set tun1 mtu 1420
ip -n VPN link set tun1 up
ip -n VPN route add default dev tun1
ip netns exec VPN su myuser -c 'firefox --no-remote'

Now all firefox (and only that firefox) traffic will go through the tunnel. Firefox has its own DNS, if you run another app it will use 1.1.1.1.

I actually do the reverse of this - I create a namespace ETH and move my eth0 device in there and attach dhcpcd to it. Then I create the wireguard tun1 device inside ETH namespace, and move tun1 to the default namespace. Then any software I run can only use the tunnel, because the ethernet device doesn’t even exist there. This keeps the routing table simple and avoids a whole class of issues and potential deanonymization exploits with the split routing table used in traditional single-namespace VPN configurations.

You can set up split tunneling yourself if you run the wireguard/OpenVPN daemon manually and move the “mouth” of the tunnel to a separate Linux network namespace.

OVPN is a 1-to-1 feature clone of mullvad (wireguard, multiple device keys, crypto payments/cash in the mail, no usernames/emails, etc.) AND has port forwarding. Switched to them when mullvad sadly closed their ports, no problems since. Can’t live without port forwarding.

That allows sending packets inside the VPN tunnel, but the outer envelope packets still need to be able to reach the VPN server.

sudo ufw default deny outgoing

I’m guessing this would block the VPN packets themselves as well.