Replies: 2 comments 2 replies
-
The PR only prevents dialing a peer that is already being dialed (which would prevent redundant dialing). Do you have any code or logs you could provide to be able to determine the actual cause? |
Beta Was this translation helpful? Give feedback.
-
You should be able to do this yourself by building a Identify could use that to share the self-reported addresses of a node with other behaviours and your custom behaviour could pick that up and attempt to dial such an address. In case the node has a public address, it would then be added to the routing table. We also have UPnP in the works which might help for deployments in people's home's to automatically perform a port-forwarding: #4156 |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
In a recent test we did with people running nodes from home, there were a few occurrences of Kademlia nodes that were seen by others as
UnroutablePeer
. In our codebase, we haven't done much that I would think causes this, so I am looking at some potential changes in libp2p.UnroutablePeer
seems to be emitted when the peer has dialed us, instead of us dialing them (here the address isNone
and that causes the event later on). The idea behind this is solid, as we do not want to add nodes that are not reachable/routable.I have a hunch this has to do with my PR #4224, causing us not to dial outwards for new nodes and thus never being able to tell if a peer is reachable/routable. (Idea: a new condition to dial only if we do not have an outgoing connection already.)But, perhaps there are more conditions that this can occur in. Perhaps there is something going on with AutoNAT or external addresses or with the client/server mode.
Beta Was this translation helpful? Give feedback.
All reactions