f1e6639374
* Switch stream type of ActiveSession to EthStream * Start `StatusBuilder` for initializing the `Status` message required for the handshake * Add `Hardfork` for `Status` default forkid * Add `MAINNET_GENESIS` constant * finish `StatusBuilder` * initialize eth streams in session * add status, hello, and fork filter to session manager * fix status builder example * add status and hello to network config * will probably remove * removing status and hello from networkconfig * move forkid to primitives * change imports for forkid * add hardfork to primitives * remove hardfork and forkid from eth-wire * fix remaining eth-wire forkid references * put mainnet genesis in constants, remove NodeId * replace NodeId with PeerId * the only NodeId remaining is inherited from enr * PeerId still needs to be documented * also run cargo fmt * replace loop with iter().any() * ignore missing docs for hardforks * use correct PeerId for Discv4::bind example test * document PeerId as secp256k1 public key * cargo fmt * temporarily allow too_many_arguments * the authenticate and start_pending_incoming_session methods have many arguments, we can reconsider the lint or fix the methods at a later point
discv4
This is a rust implementation of the Discovery v4 peer discovery protocol.
For comparison to Discovery v5, see discv5#comparison-with-node-discovery-v4
This is inspired by the discv5 crate and reuses its kademlia implementation.
Finding peers
The discovery service continuously attempts to connect to other nodes on the network until it has found enough peers. If UPnP (Universal Plug and Play) is supported by the router the service is running on, it will also accept connections from external nodes. In the discovery protocol, nodes exchange information about where the node can be reached to eventually establish RLPx sessions.
Trouble Shooting
The discv4 protocol depends on the local system clock. If the clock is not accurate it can cause connectivity issues because the expiration timestamps might be wrong.