ChanServ changed the topic of #nixos-systemd to: NixOS <3 systemd | https://jitsi.nixcon.net/systemd | Next meeting 08.12.2020 14:00 UTC (every two weeks)
lukegb has quit [Quit: ~~lukegb out~~]
lukegb has joined #nixos-systemd
<Mic92>
flokli: the way I implemented it in the first version is to to resolve each endpoint for ever until it gets a response and than keep this ip address until networkd is restarted.
<Mic92>
Not sure if this is still the case
<flokli>
Mic92: what about switches in reachability?
<flokli>
Let's say that endpoint is dual stacked, you were at home, had an ipv6 default route, now you suspend your laptop and go to a ipv6 only cafe?
<flokli>
I think we need to restart that "endpoint discovery process" every once and then, and possibly trigger it on interface changes
<flokli>
It'd lead to a no-op most of the time
<Mic92>
flokli: this needs to be fixed in wireguard. It only resolves to reachable address families, however cannot change the remote afterwards if the network changes.
<Mic92>
I don't use wireguard because it cannot do meshing like tinc
<flokli>
I regarding the re resolving discussion, I think upstream says it belongs into userspace...
<flokli>
Same goes for synth of link-local addrs based on the pubkey
<flokli>
I don't know. I'd really like to use more wireguard, but it's somewhat useless without some additional userspace thing handling these things
<flokli>
But you only realize these subtle defects when already using it
<andi->
perhaps it is enough to call networkctl reconfig my-crappy-wireguard whenever the internet connection flaps?
<hexa->
but it's a netdev
<hexa->
you probably have to delete it :)
<Mic92>
might work
<flokli>
wg setconf can set stuff while it's still up
<Mic92>
the wireguard in networkd should allow this too but I am not sure how networkctl reconfigure is implemented