openvpn split tunnel
It seems that use of remote gateway is hardcoded into the internet openvpn connection GUI. Why? If i execute openvpn on the cli on the Jolla C sailfish 3 without using /usr/lib/connman/scripts/openvpn-script and --route-noexec I get the wanted result, which is a split tunnel to the internet and to my home network via vpn0. I always get a 0.0.0.0 default gw with dev vpn0 with the gui.
How can I remain using my lte connection directly for non related vpn networks using the VPN GUI?
Thanks.
Thats why I use openvpn only on the cli. Maybe they had fix this "bug" .
FYI: I use up/down scripts (mentioned by openvpn), but this does not work with the gui vpn (see here)
utkiek ( 2019-01-23 12:27:14 +0200 )editThe behaviour that you described is probably the desired one - i.e., route _all_ traffic through VPN, which is extremely useful when you want to protect yourself on an insecure WLAN. The use case that you are describing is for when you want to use services that are not public on the internet, but are available on your own network. I don't think this is a bug - perhaps a feature request to support this option? I am assuming that your ovpn server config doesn't push a route and yet SFOS decides to set it as a default gateway? if this is the case, why do you need to use --route-noexec in the init script?
gabriel ( 2019-01-23 14:35:30 +0200 )editYeah I'm aware of the common use case. It should have said without using /usr/lib/connman/scripts/openvpn-script and without using --route-noexec. Yes it is a feature request, but i think that the ones using the vpn as def remote gateway should just configure their vpn to send def route. I cannot see why this should be (hard?)coded this way..
jsm ( 2019-01-23 17:56:31 +0200 )edit