All Core Devs - Consensus (ACDC) #166, October 2, 2025

ACDC #166 Summary
Fusaka

  • Devnet 3 had non-finality for ~2 days, then healed and resumed finalizing
  • Holesky fork activation went well overall despite some turbulence
    • Participation dropped initially due to missed client updates
    • Unexplained participation drop below 66% later in day, then recovered
    • Theories include late blocks, CPU spikes, spamming of blobs
    • Most validators run ~10,000 keys per beacon node on Holesky which could factor in
  • First Holesky BPO update scheduled at 2025-10-07 01:20:00 UTC

Glamsterdam

  • Shifting focus back to Glamsterdam development now that Fusaka is moving towards mainnet
  • ePBS updates:
    • No spec changes recently, adding tests
    • Only Teku has passing tests on main branch
    • Prysm has passing tests on separate branch
    • Other clients waiting for stable Fusaka before merging ePBS work
  • Process reminder: non-headliner EIP deadline will be one week after Fusaka mainnet date is set

ePBS Trustless Payments

  • linoscope presented case for separating payload separation from trustless payments:
    • Two different stakeholder groups: core devs (scaling experts) vs ecosystem (builders/validators)
    • Different complexity profiles and ecosystem impact
    • Risk of EOF-like outcome if bundled together
    • Bypassability concerns - rational actors might use off-protocol payments anyway
  • Supporting arguments from Greg (Lido) and others for separation
  • Decision deferred - needs written trade-off analysis before next discussion
  • ePBS breakout call scheduled for next Friday; check in there if you want to follow along

Administrative Items

1 Like