Frequently Asked Questions (FAQ)

Frequently Asked Questions (FAQ)

How does magnetico fetch metadata of torrents from DHT nodes?

Skim through BEP 5: DHT Protocol first.

magnetico basically works by launching a Sybil attack on the DHT network.

• It forges a (fake) node ID for every node it contacts to make the node consider us a neighbour.
  • "Proximity" and "neighbourhood" relies on the XOR difference (in Mainline DHT which is based on Kademlia), so we copy the n = 15 most significant bytes of the ID of the node we are contacting, and append 5 more random bytes to complete the node ID to 20 bytes in total.
• Then it floods the network with get_peers requests, whose responses will
  • ...either return a list of BitTorrent peers that we can connect and fetch metadata from.
  • ...or return a list of more DHT nodes that we can contact.
• Whenever an announce_peer request is received, the BitTorrent peer is contacted to fetch the metadata.

magnetico does "fairly egregious bittorrent spec violations" which is completely unsustainable in the long run. Therefore, I am working on BEP 51: DHT Infohash Indexing which directly addresses our use-case.