Mapping between public keys and PeerIds

[laptou]

Since PeerIds seem to be derived from a user's public key in the recommended usage, what happens when the same user wants to be active across multiple devices? I figured that I would just find a way to synchronize the user's key pair between devices, but I am not sure that having two nodes on the same network with the same PeerId is allowed. What should I do?

The protocols I am using specifically at the moment are mDNS, Kademlia, and Gossipsub.

[dvc94ch]

You can probably use identify to pass an additional user public key in one of the fields. You'll have to do the rest yourself

[rkuhn]

My suggestion is to not couple PeerId to users at all — a peer is a p2p network node, and as you say a single user may well use multiple nodes. These nodes need one unique identity each, even though they serve the same human. The associated PublicKey is an implementation detail of libp2p, I wouldn’t use it to represent the user either. A user should have their own SecretKey etc. which is completely separate from libp2p keys.

So you’ll need to implement a way on your application level for transferring a SecretKey between nodes of the same end user and then use that for encryption and signatures on the application data — in addition to all the transport layer encryption employed by libp2p.

[mxinden]

but I am not sure that having two nodes on the same network with the same PeerId is allowed.

Two online nodes with the same PeerId in the same network is not allowed.

My suggestion is to not couple PeerId to users at all — a peer is a p2p network node, and as you say a single user may well use multiple nodes.

Agreed with @rkuhn. Similar strategy is e.g. followed by https://github.com/paritytech/substrate.