Faster recovery on a restart of the daemon

[RubenKelevra]

Currently the bootstrapping takes ages to reach the connection "quality" the daemon had before it was restarted (in my case the machine was restarted).

Wouldn't it make sense to just write out the currently connected nodes (and their working IP/Port combinations on outgoing connections) and reestablish these connections on a restart of the daemon within say one hour of the shutdown?

This would remove the need to bootstrap for this case completely if there are enough connections stored on shutdown and would avoid having to establish connections to newly learned nodes, which often fails.

My node has not established have of the number of connections it used to have before after 12 minutes, while it's obviously also not the same nodes.

So the old nodes might be more useful to reconnect to since the connections have grown organically instead of forced by a bootstrapping process.

[Stebalien]

Duplicate of #3926.

However, I think you're also making some assumptions here. We don't actually care about having a lot of connections, we just don't bother closing connections until we hit the limit. The DHT will bootstrap immediately on start and will immediately connect to enough nodes to be findable in the DHT. That's at most ~160.