AllCoreDevs - Testing CL Breakout #098
Transcript
- Justin Traglia
Potuz PR: https://github.com/ethereum/consensus-specs/pull/5694 Nico PR: https://github.com/nflaig/consensus-specs/pull/9
- Enrico Del Fante
Seems potus PR is the simplest. So at this stage I’d prefer the simplest for sure
- Barnabas
Do we need to delay sepolia over this?
- Justin Traglia
Replying to "Do we need to delay ..." Nah
- Dustin
is there a specific plan of replacing Sepolia, or timeline of deprecation?
- cayman
I need to dig in deeper. My gut is either handle it properly or burn it
- Barnabas
Replying to "is there a specific ..." we haven’t put much thought into this yet
- Nico Flaig
~7 hours of no payloads seems unlikely to me
- Kevaundray Wedderburn
@Nico Flaig / @Jihoon Should we discuss the 14K epoch retention thing?
- Nico Flaig
Replying to "@Nico Flaig / @Jihoo..." sure
- Justin Traglia
- Dustin
it's the MEV way
- Dustin
it's always been
- Kevaundray Wedderburn
It will be deployed onto mainnet untested?
- Kevaundray Wedderburn
Replying to "It will be deployed ..." (I might have misheard)
- Justin Traglia
Replying to "It will be deployed ..." Depends on when flashbots updates this.
- Kevaundray Wedderburn
Reacted to "Depends on when flashbots updates this." with 👍
- Kevaundray Wedderburn
Replying to "It will be deployed ..." Is this the same for commit boost?
- Justin Traglia
Replying to "It will be deployed ..." Unsure, I think commit-boost is better prepared.
- Kevaundray Wedderburn
Reacted to "Unsure, I think commit-boost is better prepared." with 👍
- Justin Traglia
Replying to "It will be deployed ..." Not sure if we’ve tested commit-boost yet though.
- Nico Flaig
Replying to "@Nico Flaig / @Jihoo..." @Justin Traglia
- Justin Traglia
Replying to "@Nico Flaig / @Jihoo..." Oh oops. We can discuss this if we have time at the end.
- Nico Flaig
Reacted to "Oh oops. We can disc..." with 👍
- Ladislaus
Replying to "It will be deployed onto mainnet untested?" just fyi: CB folks recently published their plans for Glamsterdam here https://hackmd.io/@thUuBZa2ScW2i3kFbsrOFA/S1XTQgR_Ge#Commit-Boost-after-ePBS
- cayman
Replying to "@Nico Flaig / @Jihoo..." https://github.com/ethereum/consensus-specs/pull/5680 https://github.com/ethereum/consensus-specs/pull/5681
- Nico Flaig
we have a custom lodestar api for this
- Nico Flaig
aside the existing event
- Justin Traglia
Reacted to "just fyi: CB folks recently published their plans for Glamsterdam here https://hackmd.io/@thUuBZa2ScW2i3kFbsrOFA/S1XTQgR_Ge#Commit-Boost-after-ePBS" with 🙏
- Barnabas
anyone started work on shorter slots ?
- Barnabas
we have a WIP PR for ethereum package for it
- Kevaundray Wedderburn
It also needs to change the ptc and focil deadlines?
- Nico Flaig
im fine also testing on top, but then we are pretty far from testing it
- Kevaundray Wedderburn
Reacted to "im fine also testing on top, but then we are pretty far from testing it" with 👍
- Barnabas
Reacted to "im fine also testing on top, but then we are pretty far from testing it" with 👍
- Justin Traglia
https://github.com/ethereum/consensus-specs/tree/master/specs/_features/eip8198
- Kevaundray Wedderburn
How do we compute the deadlines wrt shorter slots?
- Jihoon
Replying to "How do we compute th..." Still BPS.
- Justin Traglia
Reacted to "Still BPS." with 👍
- Justin Traglia
Replying to "How do we compute th..." Basis points, a percentage of whatever the slot duration is
Call summary
Targets
- •Builder payment edge-case fix merged before Hoodi launch - 00:09:32
Decisions
- •No Sepolia delay for the builder payment bug; fix must land before Hoodi - 00:09:25
- •Merge epoch retention PR now; clients may implement pruning at their own pace after the fork - 00:30:37
Highlights
- Shorter Slots:
- ·Lodestar has a WIP branch for shorter slots but not on latest spec; FOCIL deadlines also need updating - 00:30:50
- ·Shorter slots should be tested on top of FOCIL due to timing interactions; EIP-8198 feature spec merged - 00:31:43
- Epbs Mevboost Commitboost:
- ·MevBoost plans to update for ePBS (GWAS); untested for Sepolia/Hoodi, targeting mainnet readiness - 00:17:19
- ·CommitBoost tested by Lodestar team on local Kurtosis only; ahead of MevBoost in readiness - 00:21:52
- Epoch Retention Reduction:
- ·Most CL clients support reducing epoch retention (EIP-8061, ~214K epochs); Paul from Teku objects, Grandine not yet heard - 00:24:51
- ·Spec change removes the constant and replaces it with a function; backporting to phase zero is problematic - 00:27:44
- ·ELs need the new minimum retention signal now so they know how long to keep blob data - 00:28:08
- Epbs Builder Payment Edge Case:
- ·ePBS bug: builder exit + unconditional proposer withdrawal can mint ETH if 2 epochs of empty payloads occur - 00:01:27
- ·Three candidate fixes: don't pay proposer (Potuz PR #5694), reorder state transition (Nico PR), or burn the ETH - 00:04:27
- ·Burning is most incentive-aligned but adds consensus-split risk; reordering needs edge-case analysis - 00:06:47
- ·Bug not triggerable without ~2 epochs of missed payloads; client circuit breakers should prevent this in practice - 00:10:08
Action Items
- •Nico Flaig - Nico to open PR against ethereum/consensus-specs repo for builder payment reordering fix - 00:15:58
- •Justin Traglia - Invite MevBoost/Flashbots maintainer (Fahil) to next ACDT to discuss ePBS update plans - 00:18:08
Key decisions
No Sepolia delay for the builder payment bug; fix must land before Hoodi
The builder payment bug must be fixed before the Hoodi release, with no postponement of Sepolia testing.Merge epoch retention PR now; clients may implement pruning at their own pace after the fork
The epoch retention change is approved for immediate merge; client implementations can adopt pruning independently post‑fork.
AI Disclaimer: Some content or metadata on EIPsInsight may be AI-inferred or automatically compiled. If you find any discrepancy, please contact us at dev@avarch.org.