GL-MT2500 DNS breaks with wireguard

Hello.

Openwrt Version
OpenWrt 21.02-SNAPSHOT r15812+879-46b6ee7ffc

Kernel Version
5.4.211

I just got the Brume 2 and I have to say that I’m pretty disappointed with the wireguard client so far. Whenever I connect it, DNS stops working for clients which breaks connectivity to all of my devices. And I do mean ALL my devices, not just the ones going though the VPN (so even with a policy to route a single device via the VPN, DNS resolution breaks for every device connected to the brume).

It doesn’t happen immediately but usually within a few minutes DNS just times out. I’ve re-done the configuration like 6 times at this point, every time resetting the firmware. Last test all I did was reset firmware, connect to WAN, and set up the client. No additional configuration done and it still won’t work with a basic setup. I’ve also tested different WAN setups (routed from my ISP’s ONT and internal IP via DHCP as well as bridged with PPPoE) just in case there was some weird bug with the WAN setup or the WAN DNS settings.

I have not done anything special and this same configuration file works just fine on my ASUS RT-AX92U (albeit with much slower performance than the Brume, which is the whole reason I bought it).

I’ve tested both my own Wireguard VPN set up on Linode and a commercial VPN and the same thing happens.

What’s more interesting is that if I use the commercial VPN’s OpenVPN profile (using the same endpoint IP), DNS does not break. This means that it’s a wireguard problem but using OpenVPN is not a solution because it’s ridiculously slow in 2023.

HELP!

I think there’s some sort of bug somewhere. I installed dnsmasq on my server and bound it to the wg0 interface so now that DNS resolutions can happen ‘locally’ over the tunnel the problem seems to have gone away. This makes no sense to me since DNS was set to 8.8.8.8 which the server on the other side of the tunnel can reach just fine to resolve…

Check you global proxy settings in VPN dashboard