All Core Devs - Consensus (ACDC) #139, August 8 2024

Protocol Calls & happenings
,
1

Agenda

Consensus-layer Call 139 · Issue #1116 · ethereum/pm · GitHub moderated by @ralexstokes

Summary

Summary by @ralexstokes (from Eth R&D Discord)

  • Specs release v1.5.0-alpha.4 is out!

  • Pectra devnet-2 update

    • Minor known issues in clients but network is generally stable and going well
    • EF Devops ran a non-finality test that nodes were able to recover from
    • Discussed next steps to devnet-3
      • a few spec issues to resolve (i.e. EIP-7702), but devnet-3 should be ready soon
      • agreed on a testing breakout to streamline devnet coordination as we get closer to mainnet Pectra
        • see info/poll on R&D discord for timing, but be on the lookout for a testing call that ideally has representation from each client team every call!

  • Next, some open questions relating to Pectra

    • First, a set of changes around reducing the demand for blob propagation leveraging the work of the mempool
    • Then, revisited the topic of deprecating mplex , a multiplexing protocol used in the libp2p stack of the CL networking software
      • mplex has generally been deprecated in favor of other solutions like yamux
      • Need some coordination amongst clients to ensure nodes remain interoperable during bootstrapping on the network
      • At least one implementation (for the Teku client) is still not quite production-ready, and there were some performance concerns around the JS implementation in Lodestar
      • And so we decided to defer deprecation of mplex in Ethereum to a later point in time once all clients feel it is time
    • And last we discussed the work to include EIP-7688 for SSZ stable containers in Pectra
      • Progress has generally been good for the EIP in isolation
      • There’s some demand for this feature in Pectra from applications, although it is not clear the demand is urgent
      • Given this, and the fact that we prefer to iterate a few more devnets for the current Pectra feature set to get to a place of greater stability, we decided to defer the inclusion discussion until a later time in the hard fork delivery timeline

  • Then, we turned to PeerDAS

    • We started by checking in with the PeerDAS implementers on development progress
    • Landed on getting an explicit spec release to target for an upcoming PeerDAS devnet
    • Also touched on the blob Beacon APIs and how they should be updated to respect PeerDAS
    • Discussed this PR to drop peer sampling from PeerDAS: Spec without peer sampling by fradamt · Pull Request #3870 · ethereum/consensus-specs · GitHub and decided it is a worthy step on the data scaling roadmap; so expect to see it in future PeerDAS devnets

  • Remainder of the call covered a variety of topics

    • A question around the interaction of sync committees and consolidations under Max EB, as you can still have a validator assigned to a sync committee, even if their balance is in the process of consolidating to another validator
      • Agreed to leave as-is with an explicit call out around this behavior for implementers
      • Will also analyze further interactions in Pectra to ensure there are no security implications for this edge-case behavior
    • Touched on standardizing the quic entry to a node’s ENR
    • Closed the call with a presentation by ProbeLab to share their work on networking/discovery protocol analysis; check their website here: Week 2024-29 | ProbeLab

Recording

Additional info

Notes: Ethereum All Core Developers Consensus Call #139 Writeup | Galaxy by @Christine_dkim