Punchy vs lighthouse.interval vs handshakes

[Quacken8]

What is the difference between these settings? I understand that they all play a role in keeping the connection alive, but I fail to see the difference. According to the docs

• lighthouse.interval sends a packet from the host to the lighthouse every n seconds which I believe opens up a (or re-uses the) hole in NAT which everyone on the nebula network uses to talk to that host
• handshakes also sends packets to "all known addresses" which should also keep the connection to the lighthouse alive..?
• punchy also sends packets periodically to keep the connection alive...?

It makes sense to me that these would all try to achieve the same thing, but to me it feels like they all go about it the same way. They at most seem to differ in the fact that punchy and handshakes maybe go all the way from one host to another and not just to the lighthouse, but if all my hosts have one of the other settings on that should keep all connections open

[johnmaguire]

Hi @Quacken8 -

lighthouse.interval is actually just how frequently a node reports its IP addresses up to the Lighthouse. The Lighthouse is first and foremost a host-discovery service that maps Nebula overlay IPs to a node's underlay IPs.

handshakes.try_interval sets the backoff interval when creating a new handshake with a host you don't already have a tunnel with. "All known addresses" here refers to all known underlay addresses for the specific host.

pucnhy is NAT hole punching and its settings are fairly well described here: https://nebula.defined.net/docs/config/punchy/