Theoretical results of consensus protocol in prima

Question

I am picking up knowledge of consensus protocols in a distributed system. Such a distributed system does primary-backup on databases.

I learned that "every consensus protocol can loop forever." from Leader election for paxos-based replicated key value store

Where is the information source of "every consensus protocol can loop forever"?

Status update: question answered. The same information source was provided by rystsov and another person of another post.

Could more theoretical results and the corresponding information source be shared?

Answer 1

The "every consensus protocol can loop forever" statement is known as the FLP impossibility result which is described in the Impossibility of Distributed Consensus with One Faulty Process paper.