Why is split tunnelling on linux using wireguard so convoluted? I don't want to run 20 magic commands I will not end up understanding to set up a network namespace, why can't wg-quick or anything similar just do it? Is the "I just need to proxy this specific process" use case that niche? I find it hard to believe, honestly.
6
Replies
·
0
Boosts
@mid_kid
Can you explain the use case a little more? maybe youre making it harder than needed?
@dlakelan I just want one specific process to use a VPN. My use case for this varies (bypassing firewalls with firefox, torrenting, testing), but in my opinion that's such a basic primitive I end up needing a surprising amount.
I don't want to use docker.
@mid_kid
Yeah, I kind of think setting up a veth and a separate namespace are the ideal mechanism here. you can match on incoming veth and use that rule to route the packets to the VPN. basically you just need a bash script that takes a command and runs it in the namespace... and another bash script to set up or tear down the namespace. this isnt super hard but I can't write it off the top of my head.
@hipsterelectron Eh, for some reason I manage my way around unshare fine enough. Though sometimes I end up having to read the code of bubblewrap and co. because duh of couse you need to fork in order to mount a new /proc. (I'm sure this makes more sense to kernel devs, but not to me...)
