Autonat changes the status to private incorrectly

[master255]

Libp2p and DHT gets a lot of connections with an error at startup time. Always. Because of this Autonat almost always switches to private mode and only after a few minutes to public.
I think it is necessary to delay before starting autonat. Now there is no delay at all. A few seconds after the start an error arrives and the host with a white ip address gets private status.
Because of this, after the start, for several minutes, no one can connect to this host through the Internet.

In continuation of #2046

[marten-seemann]

1. This doesn’t have anything to do with autonat (?): node frequently changes its mind about its reachability status #2046.
2. I haven’t observed this on my public nodes.

[master255]

1. You don't have any errors when passing the bootstrap period in the DHT?
2. I only use the TCP protocol. Maybe nodes that use Quic will cause errors.
3. What if someone were to purposely cause errors when connecting to the DHT? Attackers?

[marten-seemann]

This might happen because at the beginning AutoNAT is operating on a limited amount of data. It's usually resolved very quickly:

I don't see any need to fix this. Also, we're working on AutoNAT v2, which will have a much more sophisticated logic.

[master255]

Yes. That's the problem I'm talking about. It can be solved with a timeout.