All Core Devs - Consensus (ACDC) #143, October 3 2024

abcoathup · September 21, 2024, 1:57am

Agenda

Consensus-layer Call 143 · Issue #1158 · ethereum/pm · GitHub moderated by @ralexstokes

Summary

Recording

https://www.youtube.com/live/dplciLdQTM0?si=d_2r-mWR-AIfjF0e&t=9

Additional Info

Ethereum All Core Developers Consensus Call #143 Writeup | Galaxy by @Christine_dkim

ralexstokes · October 4, 2024, 2:50pm

ACDC #143 summary
Action Items

Resolve open devnet-4 spec issues: pectra-devnet-4 specs - HackMD
- Special call out for the execution requests refactor: engine: Make execution requests a sidecar, take 2 by mkalinin · Pull Request #591 · ethereum/execution-apis · GitHub
Assess implementation weight of Separate type for unaggregated network attestations by arnetheduck · Pull Request #3900 · ethereum/consensus-specs · GitHub
- Client teams lean towards inclusion in pectra-devnet-5 or later; another client team or two assessing the full scope of changes to weigh cost vs benefit would be very helpful
Prioritize implementation of engine_getBlobsV1 to more quickly roll out bandwidth savings for network nodes

Summary

Started with a check-in around the progress of pectra-devnet-3
- A few minor bugs (Besu, Lighthouse seem to have open issues) with work under way to resolve
Then turned to planning pectra-devnet-4
- Series of open spec issues remain
- Started with the design for the execution requests in the Engine API and how they are structured across the CL and EL client
- There has been active design work this week and we are very close to a final design
- Follow the conversation on this PR engine: Make execution requests a sidecar, take 2 by mkalinin · Pull Request #591 · ethereum/execution-apis · GitHub to track progress
- Touched on the other open devnet-4 CL PRs:
  - eip7251: Switch to compounding when consolidating with source==target by mkalinin · Pull Request #3918 · ethereum/consensus-specs · GitHub
    - Has been merged
  - eip6110: Queue deposit requests and apply them during epoch processing by mkalinin · Pull Request #3818 · ethereum/consensus-specs · GitHub
    - Ready for review, aim to merge by end of next week
And then look at some final Pectra items with two PRs to refactor the handling of attestations
- OnchainAttestation: Separate type for onchain attestation aggregates by mkalinin · Pull Request #3787 · ethereum/consensus-specs · GitHub
  - nice from a spec point of view, but we agreed to defer to a later HF to put into the specs to avoid client work that could potentially delay Pectra
- SingleAttestation: Separate type for unaggregated network attestations by arnetheduck · Pull Request #3900 · ethereum/consensus-specs · GitHub
  - this PR helps with DoS resistance in pathological cases and so there was much more support to include in Pectra relative to the previous link
  - going to target a “minimal” implementation that only addresses this type at the networking layer, and not pursue a full set of changes that could otherwise be imagined under this change
  - decided to target a later devnet (after pectra-devnet-4) for potential inclusion which gives a bit more time for further review, including assessment of implementation load
Next, turned to PeerDAS
- Work has continued on PeerDAS devnets, considering launching a new net after a variety of bug fixes
- Turned to engine_getBlobsV1 which was recently merged into the execution-apis repo
  - Clients have a diversity of implementation progress, but it is a top priority so that we can roll out potential bandwidth savings on the network ASAP
    - Note: needs support on both CL and EL!
  - Discussed this PR as a follow on to engine_getBlobsV1: P2P clarifications when introducing `engine_getBlobsV1` by tbenr · Pull Request #3864 · ethereum/consensus-specs · GitHub
    - Group agreed to treat mempool blobs retrieved via this method as blobs received over gossip, implying they should also be gossiped to peers
- Then had another check-in around various blob scaling options following up from last week’s call
  - Still a bit early as we are still in a data-gathering phase to best understand which solutions are preferred
  - Relevant analysis from the last week: On solo staking, local block building and blobs - Sharding - Ethereum Research
  - potuz raised an interesting point around syncing and ensuring that nodes can sync at higher blob counts, especially under situations of non-finality
    - experiment is tricky as it is hard to replicate mainnet network topology
  - Ansgar then asked about how to think about timing of blob changes so we aren’t in a position where we want to increase blob counts in Pectra but are waiting for a cycle of spec and implementation work
    - We agreed that EIP-7742 should be on client team’s radars sooner rather than later
Wrapped the call with an announcement around a named testnet (aka user facing) for Devcon, with the plan to target devnet-4