Questions tagged [consensus]

A fundamental problem in distributed computing and multi-agent systems is to achieve overall system reliability in the presence of a number of faulty processes. use this tag for consensus kind of problem.

67 questions
19
votes
4 answers

Why is the consensus number for test-and-set, 2?

According to Wikipedia, The test-and-set operation can solve the wait-free consensus problem for no more than two concurrent processes. Why can't it solve the problem for more than two processes?
ben
  • 295
  • 2
  • 9
9
votes
2 answers

Why is the commit phase in PBFT necessary?

I've read many papers and slides on Practical Byzantine Fault Tolerance (PBFT) but I'm still confused about why a COMMIT phase is required. Most material states that PREPARE phase ensures fault-tolerant consistent ordering of requests within…
qweruiop
  • 201
  • 2
  • 5
6
votes
1 answer

Why $e(C_i) = D_i$ is correct assumption? (FLP Impossibility 1985 - Lemma 3)

Please bear with my unhelpful typesetting. My question is regarding well known FLP paper Impossibility of Distributed Consensus with One Faulty Process by Fischer, Lynch and Patterson While discussing Lemma 3, in 4th paragraph, authors make an…
5
votes
1 answer

Consensus protocols in multiplayer video games

I'm not 100% sure this appropriate for CS. Sorry. I'm looking for prior art on an idea I had, but either my Google fu is lacking, or it genuinely hasn't happened. In distributed systems, there are a number of consensus protocols and implementations…
nfirvine
  • 151
  • 4
5
votes
1 answer

Paxos consistency

I'm looking at the paxos family of protocols for solving consensus in a network of unreliable processors. I'm working through scenarios where processors fail, and I know I'm wrong, but I don't know why. I have 3 machines with the state A. One…
dv1729
  • 151
  • 2
5
votes
1 answer

Proof of commonality of acceptors in majorities

In the Paxos consensus protocol, a fact that follows from its construction is "any two majority sets of acceptors will have at least one acceptor in common". This observation can be extended to any N majority sets of acceptors having either at least…
Grayskin
  • 51
  • 1
5
votes
1 answer

What distributed consensus algorithms don't rely on a time source?

I'm looking for information about distributed consensus algorithms that: maintain a consistent transaction log across all healthy nodes have similar high-availability properties to Paxos (i.e. not liveness, but practically very likely to terminate…
Max
  • 261
  • 1
  • 3
4
votes
2 answers

Can we have a strictly monotonically increasing/decreasing sequence generated by a distributed system?

Just what the title says. Can we have a strictly monotonically increasing/decreasing sequence generated by a distributed system (without a single point of failure)? My current thoughts are that this is a consensus problem, and is thus impossible to…
skittish
  • 65
  • 7
4
votes
3 answers

Where does the FLP impossibility proof depend on allowing a single process failure?

Question I'm reading the FLP impossibility paper. I think I understand the idea of the proof, and I don't have questions about it. However, it seems like the assumption of having at most, a single faulty process is not used in the proof. Put another…
statusfailed
  • 141
  • 3
4
votes
3 answers

What is the consensus algorithm that requires an odd number of nodes?

Last week I stood up and gave a talk on the Raft Consensus Algorithm. The first question I got was: Doesn't it need an odd number of nodes? I answered "During leader election, the first node to become a candidate, all the other nodes have to vote…
hawkeye
  • 1,199
  • 8
  • 20
4
votes
1 answer

Is there a difference between total-ordering consistency and linearizability?

By total-ordering I mean the following definition[0]: Total Event Ordering: Controller replicas should process events in the same order and subsequently all controller application instances should reach the same internal state. If my…
Erwan Aaron
  • 183
  • 1
  • 7
4
votes
1 answer

Symmetric protocols for consensus

All the consensus algorithms that I've seen always depend on having a leader elected with all queries directed at the leader (assuming that we're not OK with getting back stale data). Are there any efficient algorithms that are completely symmetric,…
eof
  • 121
  • 2
4
votes
1 answer

In the FLP Impossibility paper, why did the authors claim that e is applicable to every E in proof of lemma 3?

The paper is available here: https://groups.csail.mit.edu/tds/papers/Lynch/jacm85.pdf The 1st paragraph of lemma 3's proof says In other words,if event e is applicable to config C,and E is any config reachable from C without applying e,then e is…
user98670
  • 141
  • 3
3
votes
1 answer

Impossibility of Distributed Consensus with One Faulty Process

I'm reading Fischer et al.'s Impossibility of Distributed Consensus with One Faulty Process, and I don't quite follow 2 points: The first is in the proof of Lemma 3, pg. 378: Otherwise, $e$ was applied in reaching $E_i$, and so there exists $F_i ∈…
3
votes
1 answer

FLP Impossiblity Result assumption of $C_1 = e'(C_0)$

FLP86's famous proof regarding impossibility of async distributed processes with a single fault assumes in the proof of the third lemma the existence of an event $e'$ such that the neighbor configurations $C_0$ and $C_1$ can be related as $C_1 =…
1
2 3 4 5