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
When transaction redundancy in a node is too low (too few duplicates), a node is potentially open for attacks. When the redundancy controller of the protocol notices this, it triggers sending a ResetRoute message to a random peer of the node. This peer will then remove the node from any of its DisabledRoute lists and allow future traffic to flow unrestricted to the node. Routes are disabled when a node sends a HaveTx message to this peer.
Currently the choice of the peer to whom a ResetRoute message is sent is random. That means that a node can, in theory, send a ResetRoute message to a peer that has not previously received a HaveTx message, because the protocol does not keep information about to whom a HaveTx message has been sent.
At the moment this did not have any negative effects but it could mean a node takes more time to catch up on transactions.
We propose an optimization to the protocol, whereby a node remembers to whom HaveTx messages have been sent and only send a ResetRoute message to those peers.
DOG Protocol Change Proposal
When transaction redundancy in a node is too low (too few duplicates), a node is potentially open for attacks. When the redundancy controller of the protocol notices this, it triggers sending a
ResetRoute
message to a random peer of the node. This peer will then remove the node from any of itsDisabledRoute
lists and allow future traffic to flow unrestricted to the node. Routes are disabled when a node sends aHaveTx
message to this peer.Currently the choice of the peer to whom a
ResetRoute
message is sent is random. That means that a node can, in theory, send aResetRoute
message to a peer that has not previously received aHaveTx
message, because the protocol does not keep information about to whom aHaveTx
message has been sent.At the moment this did not have any negative effects but it could mean a node takes more time to catch up on transactions.
We propose an optimization to the protocol, whereby a node remembers to whom
HaveTx
messages have been sent and only send aResetRoute
message to those peers.cc @hvanz
The text was updated successfully, but these errors were encountered: