mDNS peer discovery stops working once a certain threshold is reached #2012
Labels
kind/bug
A bug in existing code (including security flaws)
need/triage
Needs initial labeling and prioritization
Comments
achingbrain
added
need/triage
This was referenced Aug 31, 2023
achingbrain
added a commit
that referenced
this issue
Sep 1, 2023
The data field of mDNS TXT records has a [hard limit of 255 characters](https://agnat.github.io/node_mdns/user_guide.html#txt_records) - some multiaddrs can be longer than this so filter them out before sending, otherwise remote peers can fail to parse our mDNS response. We should also only be sending [link-local addresses](https://github.com/libp2p/specs/blob/master/discovery/mdns.md#issues) in mDNS responses so filter any non-link-local addresses out too, though go-libp2p includes loopback addresses even though it probably shouldn't (according to the mDNS spec which conflicts with our mDNS Peer Discovery spec) so we continue to do so as well. Fixes #2012
This was referenced Jan 18, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
mDNS peer discovery works by making a broadcast query for
_p2p._udp.localservices. Peers respond with their multiaddrs.The number of multiaddrs a peer has affects the size of the UDP message being sent as the mDNS response.
Once this message hits a certain size the
multicast-dnsmodule fails to decode the message. The message size this occurs at iswell under the maximum size for a mDNS response message- turns out there's a hard length limit for mDNS TXT answers and certain addresses such as circuit relay over WebTransport can exceed this limit and cause parsing failures.Refs: mafintosh/multicast-dns#85
The text was updated successfully, but these errors were encountered: