As of now (from what I can gather), the backend registers servers (adds the server to the public list) using the outgoing IP of the server.
This causes some issues:
- In IPv6, the outgoing IP address may not be the same as the IP allowed for incoming requests; this can cause confusion to users who may open their port on one address, but the outgoing address is something different (i.e., they open a port for their EUI-64 address, but privacy address is used for outgoing)
- If the server admin wishes to use a domain to support dual-stack (having both A and AAAA records), there is currently no way doing this
My suggestion is adding a new API request to the backend that allows specifying a "host" to be used instead making the assumption that the IP used to connect with the backend is the desired IP for the server.
I'm willing to submit another PR to add a "Host" setting to the server. When "Host" is unspecified, the server and backend behave just like they do now, but when specified, the server will send the backend an API request hinting it to use the set domain or IP for public server registration.
But before I can make such PR, the backend needs to be updated as described above.
As of now (from what I can gather), the backend registers servers (adds the server to the public list) using the outgoing IP of the server.
This causes some issues:
My suggestion is adding a new API request to the backend that allows specifying a "host" to be used instead making the assumption that the IP used to connect with the backend is the desired IP for the server.
I'm willing to submit another PR to add a "Host" setting to the server. When "Host" is unspecified, the server and backend behave just like they do now, but when specified, the server will send the backend an API request hinting it to use the set domain or IP for public server registration.
But before I can make such PR, the backend needs to be updated as described above.