Hello, thanks for pointing that out. Actually the idea was to switch over my VPN endpoint from one VPS to another one, thus the any address route (0.0.0.0/0) was correct, but now it is clear to me it does not make sense really to have any address twice. I am trying to test the new endpoint with a dedicated machine now (which fails to install wireguard). But I am still wondering why the network stack kind of crashes. Even with a bullshit configuration like I did, imho this should never happen. Thanks, Joachim
-----Ursprüngliche Nachricht----- Von: WireGuard <[email protected]> Im Auftrag von Xand Meaden Gesendet: Tuesday, 14 July 2020 15:50 An: [email protected] Betreff: Re: AW: two client connections -> crash? "AllowedIPs" is the list of IP addresses/subnets that should be routed via that wireguard tunnel. In your case you've set both tunnels to be default route (0.0.0.0/0) for IPv4 traffic. So it depends on what is the other end of each tunnel - and what you want the tunnel to be used for. AllowedIPs might just be the private IP address of each peer if you just want to communicate with that. Xand On 14/07/2020 14:43, Joachim Lindenberg wrote: > Good observation. I never really understood what IPs I should put there and > also didn´t find a good documentation on that. And obviously with one > connection it wasn´t that important to get it right. What IP addresses or > network should AllowedIPs refer to? Client? Server? Tunnel? > Thanks, Joachim > > -----Ursprüngliche Nachricht----- > Von: M. Dietrich <[email protected]> > Gesendet: Tuesday, 14 July 2020 12:11 > An: [email protected]; 'WireGuard mailing list' > <[email protected]> > Betreff: Re: two client connections -> crash? > > > Quotation from [email protected] at Juli 13, 2020 20:53: >> I am trying to configure one client system (Ubuntu 18.04.4 LTS >> (GNU/Linux 5.3.0-62-generic x86_64)) against two servers. The >> configuration is very similar: >> >> root@Mailcow:/home/joachim# cat /etc/wireguard/wg0-client.conf >> [Interface] Address = 10.200.200.2/24 PrivateKey = *** DNS = 8.8.8.8 >> #10.200.200.1 >> >> [Peer] >> PublicKey = qn6CTz578gbrYpzYkvV2okoqkIFHKye+mRj4i/I8Sz8= >> Endpoint = fire.lindenberg.one:51820 >> AllowedIPs = 0.0.0.0/0 >> PersistentKeepalive = 21 >> >> root@Mailcow:/home/joachim# cat /etc/wireguard/wg1-client.conf >> [Interface] Address = 10.200.201.2/24 PrivateKey = *** DNS = 8.8.8.8 >> #10.200.200.1 >> >> [Peer] >> PublicKey = QAJANxtuAvdT+HR3fP1I2DXq0Azl0T3jF5s+cW7foSA= >> Endpoint = nc.lindenberg.one:51820 >> AllowedIPs = 0.0.0.0/0 >> PersistentKeepalive = 21 >> >> Wg-quick up wg0-client ist at system startup. Now unfortunately when >> I do wg-quick up wg1-client the network stack kind of crashes. The >> command does not terminate, and connectivity on all interfaces is >> broken. >> Is this a configuration issue? Should I change ports to be different? >> Is there some other issue? > The ports are fine because the IPs are different. You use the same AllowedIPs > for both. And they cover the whole network. > This cannot work. What is the intention of that config? > >> Do I have to define two interfaces or could I have just one with >> multiple peers? But how could I then specify which tunnel to use? > Depends on what you want to achieve. Sure you can use multiple peers for one > interface. >
