first validator has suspicious balance after running state sim about nimbus-eth2 HOT 4 CLOSED

It's not just the first validator.

There are five missing attesters at epoch processing time. The first attester (because state_sim.nim doesn't attest on first slot) and the last four of the epoch:

{1, 54, 22, 25, 27}

Where the overall validator sequence by shard is:

[25, 15, 30, 50, 20, 4, 35, 47, 49, 40, 32, 63, 31, 23, 60, 53, 3, 59, 21, 7, 33, 52, 48, 38, 36, 13, 2, 24, 39, 14, 6, 8, 34, 10, 5, 9, 19, 44, 17, 57, 62, 42, 26, 29, 37, 58, 28, 12, 41, 46, 45, 16, 43, 0, 56, 11, 61, 18, 55, 51, 22, 1, 54, 27]

This is related to the inclusion delay. So that's the direct cause, at least.

from nimbus-eth2.

ah right how silly.. start.sh doesn't attach all validators

from nimbus-eth2.

actually no - start.sh is a beacon_node sim thing - here, all validators were attesting.

what's special however is that we were not setting epoch_boundary_root - this effectively meant that the chain was not finalizing - this might have cause a whole lot of issues - would be good to understand/document this more in depth.

from nimbus-eth2.

Resolved by updating to a spec which doesn't use epoch_boundary_root, as far as I've been able to (not) reproduce.

from nimbus-eth2.