Consensus-layer Call 152

Protocol Calls & happenings
1

Consensus-layer Call 152

prev: call 151

Meeting Date/Time: Thursday 2025/3/6 at 14:00 UTC
Meeting Duration: 1.5 hours
stream

  1. Electra
  2. PeerDAS / Blob scaling
  3. Research, spec, etc.
  4. Open discussion/Closing remarks

GitHub Issue

1 Like
2
3

ACDC #152 summary

Action Items

Summary

  • Pectra
    • The Holesky testnet succesfully upgraded to Pectra, but a series of bugs impacted network health and we have not been finalizing since 24th February.
    • Client teams are currently working to restore finality, and we began the call with discussions around the best way to achieve that.
    • Lighthouse started by highlighting some updates to their client to handle growing state size during times of non-finality.
    • Manu from Prysm built a model to estimate when we would expect finality given the current state of the network: Holesky wen finality - Google Sheets. This model estimates the restoration of finality on 28th March.
    • Next, we turned to a doc from Pari to explore some options on a path forward.
    • Everyone agreed we should restore Holesky, and given the potential for this process to take weeks, Pari raised the question of a temporary second network to allow stakers to test in parallel. A straightforward option would be a shadow-fork of Holesky that would replicate a similar setting to Holesky today, with the caveat that network topology would likely not match Holesky (which is more similar to mainnet than a shadow-fork would be). The biggest blocker to a shadow-fork useful for testing would be the tooling and infrastructure layer as we would need duplicates to run along with the current Holesky setup.
    • We had a Lido representative on the call who could chime in as one of the stakers who would like to test. In short, they could work with a second network, although they would still want to do some testing on Holesky once it finalizes.
    • Some client teams chimed in with some downsides to a second network:
      • Some bugs on Holesky need to be resolved, and we would not see them on a second healthy network.
      • Having a second network would dilute attention of client teams from recovering Holesky as their primary focus.
    • We briefly touched on pectra-devnet-7 and decided to spin the network down as we have verified everything on this network worth testing.
    • Another point was raised that even if we do finalize Holesky, many validators will be exited from slashing and so restoring Holesky to a high validator count (~1.8M) would take a long time due to the activation queue.
    • The PandaOps team highlighted the importance of non-finality testnets, and attendees agreed the best way to handle this is in parallel to other R&D efforts. This suggests we try to move ahead with finalizing Holesky as soon as possible.
    • Taken together, we ultimately agreed that a second network is worthwhile, and while we will evaluate the situation after Holesky finalizes again we expect to terminate the second network upon regaining finality.
    • Next, we turned to Sepolia which upgraded to Pectra this week and also had some small turbulence following the triggering of a bug after the upgrade. Sepolia has returned to normal operation.
    • One issue that came up from the Sepolia bug was the consideration of specifying how clients parse deposit logs and usage of the deposit contract’s ABI encoding.
      • There was quite a bit of back and forth between EL client teams around how to handle this as some clients do not want to enshrine the encoding in the spec.
      • Check the call for the full discussion, and expect more conversation on next week’s ACDE.
    • We rounded out the Pectra discussion with a number of bugs recently found:
      • 2 BLS issues that have been resolved, with enhanced test coverage.
      • A griefing vector in EIP-7002 (and to a lesser extent EIP-7251); we agreed no change is necessary at the moment as the extent of the griefing vector does not appear that severe.
  • PeerDAS / blob scaling
1 Like
4

Audio

Writeup