Consensus

The consensus policy sends the same request to multiple upstreams and returns the result only when enough of them agree. This can be used to detect misbehaving nodes or inconsistencies between providers.

It can only be defined on network level since it requires a set of upstreams.

Config

projects:
  - id: main
    networks:
      - architecture: evm
        evm:
          chainId: 42161
        failsafe:
          consensus:
            requiredParticipants: 3       # how many upstreams participate
            agreementThreshold: 2         # how many identical responses are needed
            disputeBehavior: preferBlockHeadLeader
            lowParticipantsBehavior: acceptMostCommonValidResult
            punishMisbehavior:
              disputeThreshold: 3
              disputeWindow: 10s
              sitOutPenalty: 30s

Options

• requiredParticipants – number of upstreams selected for a round.
• agreementThreshold – minimum count of identical responses to reach consensus.
• failureBehavior – what to do when most participants return an error:
  • returnError
  • acceptMostCommonValidResult
  • preferBlockHeadLeader
  • onlyBlockHeadLeader
• disputeBehavior – how to pick a result when participants disagree:
  • returnError
  • acceptMostCommonValidResult
  • preferBlockHeadLeader
  • onlyBlockHeadLeader
• lowParticipantsBehavior – how to handle cases with fewer than requiredParticipants healthy upstreams:
  • returnError
  • acceptMostCommonValidResult
  • preferBlockHeadLeader
  • onlyBlockHeadLeader
• punishMisbehavior – optional settings to temporarily sit out upstreams that consistently disagree with the majority:
  • disputeThreshold – how many disputes trigger punishment
  • disputeWindow – time period used to measure disputes
  • sitOutPenalty – how long an upstream is ignored once punished