Which kind of denial-of-service attack can a malignant replica accomplish?
Particularly, can a malignant replica prevent an update call to happen? Can it prevent several update calls in a row?
Hi qwertytrewq
Update calls are sent to a random replica of the subnet running the target canister. A malicious replica may tell the user it accepted the update call, but never send it to its peers and never put it in a block. In that case it will not be executed.
By sending another update call, it will most likely end up on a different replica and if that one is honest, it will addd it to its next block, if there is space.
If the message hasn’t been executed by its expiry time, the protocol guarantees that it will never be executed.
Does that help?
Yes, it helps. Thank you for this explanation.