NAT traversal

[emhane]

Optimised protocol based off ethereum/devp2p#207
Relating to issue: ethereum/devp2p#226

• NAT hole punching: moved into: Nat hole punching for discv5.2 sigp/discv5#176 from https://github.com/emhane/nat_hole_punch/tree/standardise-nat-hole-punch
• Type up wire protocol for devp2p: discv5: NAT hole punching wire protocol devp2p#225
• Type up theory for devp2p discv5: NAT hole punch theory devp2p#228
• Mechanism to keep hole punched for a peer, e.g. UTP stream.
• Remove sessions for peers that don't keep the hole punched for us after hole punching. already default behaviour in the rust impl of discv5 for req time outs
• Part 2 theory 'around the hole punching': discv5: additional text around NAT hole punching devp2p#231
• Make max total sessions with non-contactable enrs configurable (symmetrically NAT:ed, like nodes in many offices)
• Possibly also, look into predicting WAN reachable ports for peers behind symmetric NATs https://www.cs.cornell.edu/people/francis/fdna02-guha1.pdf https://netsec.ethz.ch/publications/papers/natblaster.pdf

[carver]

Work happening here: https://github.com/emhane/discv5/commits/nat-hole-punch

[emhane]

Work happening here: https://github.com/emhane/discv5/commits/nat-hole-punch

Thanks Jason. Also here https://github.com/emhane/nat_hole_punch/tree/nat-hole-punch, and here for testing https://github.com/emhane/discv5-cli/tree/nat .

[carver]

When following the docs, I get this for the first server attempt:

error: Found argument 'query' which wasn't expected, or isn't valid in this context

I think it started up fine by just removing query. I haven't finished running the test, though.

[carver]

Hm, the second cli instance doesn't recognize -u, -i, or -r, so I sense that I'm doing something really wrong. I'm not seeing those options in cargo run -p discv5-cli -- server -h. I'm running commit aa52bbd2879c83257

[emhane]

But the latest commit at origin/nat is 8feb22ede3203e567bfcae4071f117c6eae472c0 ?

[carver]

yup, I'm a goof, that's why we post commit hashes! :P

[carver]

sigp/discv5#134

[KolbyML]

After this is done we can look at closing this issue #22 since this would be the solution to it?

[emhane]

After this is done we can look at closing this issue #22 since this would be the solution to it?

That's right, both issues tackle the same general problem of how to include nodes behind NAT in the network.

[github-actions]

This issue is stale because it has been open for 42 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 14 days since being marked as stale.