EPF - Eighth Update

My last update was based on the implementation of the Req/Res Protocol module for the Consensus Layer peer-to-peer setup project. Some changes were still pending for this to be fully functional so that will be the main topic of this update

Req/Res Protocol

The work for this feature took place in this pull request:

• https://github.com/brech1/cl-p2p-setup/pull/1

I worked on implementing the required responses for each request. This is a code snippet for the match arm of a new RPC event:

BehaviourEvent::Rpc(rpc_message) => {
    match rpc_message.event {
        Ok(received) => match received {
            rpc::RPCReceived::Request(substream, inbound_req) => {
						...
						},
            rpc::RPCReceived::Response(_, _) => (),
        },
        Err(_) =>  (),
    }
}

I only match the event of the RPC message since the other data is not relevant for the message processing. The following is the RPCReceived::Request match arm:

rpc::RPCReceived::Request(substream, inbound_req) => {
    match inbound_req {
        InboundRequest::Status(status)=>{
            swarm.behaviour_mut().rpc.send_response(rpc_message.peer_id, (rpc_message.conn_id,substream), RPCCodedResponse::Success(RPCResponse::Status(status)));
        },
        InboundRequest::Ping(_) => {
            swarm.behaviour_mut().rpc.send_response(rpc_message.peer_id, (rpc_message.conn_id,substream), RPCCodedResponse::Success(RPCResponse::Pong(rpc::methods::Ping {
                data: 0,
            })));
        },
        InboundRequest::MetaData => {
            swarm.behaviour_mut().rpc.send_response(rpc_message.peer_id, (rpc_message.conn_id,substream), RPCCodedResponse::Success(RPCResponse::MetaData(MetaData{
                seq_number: 0,
                attnets: EnrAttestationBitfield::default(),
                syncnets: EnrSyncCommitteeBitfield::default(),
            })));
        },
        _ => {
            swarm.behaviour_mut().rpc.send_response(rpc_message.peer_id, (rpc_message.conn_id,substream), RPCCodedResponse::Error);
        },
    }
}

For each request I use the send_response method of the RPC and build the specific RPCResponse for it.

Debugging

I spent quite some time doing small fixes during the debugging stage. But after some time I got it working. This is the output of a local lighthouse node:

Feb 06 18:22:35.807 DEBG Connection established                  connection: Listener, peer_id: 16Uiu2HAmQAF6yFrbm7u1q9XJUsqupMYDgs6XZcuBavBrG6RBo7h5, service: libp2p
Feb 06 18:22:35.808 DEBG Obtained peer's metadata                new_seq_no: 0, peer_id: 16Uiu2HAmQAF6yFrbm7u1q9XJUsqupMYDgs6XZcuBavBrG6RBo7h5, service: libp2p

In the lighthouse client, the first step after connecting to a new peer is to request the metadata. In the above snippet, we can see that this metadata has been received successfully! 🎉🎊🎉🎊

Eventually you’ll be disconnected for the node having “Too many peers”, this clearly shows the need of a peer manager.

Feb 06 18:23:00.295 DEBG Peer Manager disconnecting peer         reason: Too many peers, peer_id: 16Uiu2HAmQAF6yFrbm7u1q9XJUsqupMYDgs6XZcuBavBrG6RBo7h5, service: libp2p

However, in the real world scenario I still get disconnected without the protocol being used, so I need to start looking at:

• Networking debugging tools
• Different client compatibilities

Helios CL P2P Possible Contributions

I laid out a document with possible contributions to the current state of this project.

• https://hackmd.io/@brech1/cl-p2p-contribution

This is for any contributor interested on contributing to this new feature.

Rust Ethereum Consensus Specs

While exploring the helios codebase, I noticed this issue:

• https://github.com/a16z/helios/issues/132

I’m aware that I need to refactor some parts of the code to be compatible with some of the current or future helios crates or standards. So I decided to check out the Alex Stokes’ Ethereum Consensus repository.

I noticed that the P2P Interface types were only implemented for the phase0 specs, so I reached out to Alex and asked about how could this be implemented. These are the links for the issue and pull request:

• https://github.com/ralexstokes/ethereum-consensus/issues/185
• https://github.com/ralexstokes/ethereum-consensus/pull/186

Next Steps

My first priority will be to test locally with other CL clients to see if any other errors come up. But also I need to implement the peer manager to keep a stable amount of peers.