You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Snodes only propagate data to other snodes when they receive a client store request.
The propagation is performed via the snode to snode api, which doesn't trigger any further propagation. So the snode that received the client's store request is responsible to share it with the rest of the swarm.
We might need to enforce a hash check across snodes, e.g. have a fast/light entrypoint that allows to check if a snode knows about a hash. That way, we could come up with a deterministic algorithm to propagate the data across the swarm, even if the original snodes skips some other snodes during propagation.
E.g. upon receiving data from the propagation api, after a set delay, the snode could check if a couple of random peers already know about the hashes. If they don't, then the snode can push the messages to them. (The algorithm should avoid cycles)
Here's an attack that a malicious snode could perform, @msgmaxim thought of:
The text was updated successfully, but these errors were encountered: