Consensus on info_sync duty is triggered every epoch (32 slots). That means that any cluster size that is a divisor of 32 will have a constant leader order, i.e., the same peers will be leader for round 1 and then for round 2 and for round 3 in each instance.
thomasheremans
commented
1 year ago
Problem to be solved
Consensus on info_sync duty is triggered every epoch (32 slots). That means that any cluster size that is a divisor of 32 will have a constant leader order, i.e., the same peers will be leader for round 1 and then for round 2 and for round 3 in each instance.
Proposed solution
Change the naive
leaderfunction to something like this: https://github.com/protolambda/eth2-shuffle