I have had trouble twice now with modifying a working vpn configuration, and then being unable to get it to start. Both times I never actually solved it, so much as eliminating the problem by switching to a different nordvpn config file.
There was a penetration at nordvpn in which some passwords and userinfo were leaked. I wanted to change my password, and did, and had to get into the vpn router and change it there. And after I did the vpn just would not start. Eventually, I switched to another vpn endpoint, put in a new .conf file in /etc/openvpn/client and it came right up.
I don’t know what this is about.
March 26, 2021: I spent all morning on this again. I was changing scripts on Rosemary and also, double and triple checking that I did not allow IPv6 on obelisk/rosemary (if routable IPv6 addresses are available they will be used in preference to the IPv4 vpn, which is the whole point of obelisk). After a reboot of obelisk, it lost dns. I spent several hours trying to solve this, most of the time spent on obelisk. The vpn would come up, and I could ping raw addresses, but the dns wouldn’t work. I don’t even see what this has to do with the vpn tunnel.
Yet in the end, out of desperation, I tried bringing in a new nordvpn conf file (actually it comes in as an ovpn file, and once I put a password link in it it is saved as a conf file). A link in /etc/openvpn points to whichever of these is active. So I installed a new one, rebooted the router (again), and like magic it began to work.