skip to content

Ethereum All Core Developers Consensus Call #116 Writeup

044

On August 24, 2023, Ethereum core developers gathered over Zoom for their 116th All Core Developers Consensus (ACDC) call. The ACDC calls are a bi-weekly meeting series where developers discuss and coordinate changes to the consensus layer (CL) of Ethereum. Chaired by Ethereum Foundation Researcher Danny Ryan, ACDC #116 was a short call, 30 minutes in duration. Developers primarily discussed progress on testing for the upcoming Cancun/Deneb (Dencun) hard fork upgrade.

Barnabas Busa, a DevOps Engineer at the Ethereum Foundation, shared updates on Devnet #8, which is a developer focused Dencun test network that launched last Wednesday on August 16. Parithosh Jayanthi, also a DevOps Engineer at the Ethereum Foundation, presented a comprehensive testing overview for the various parts of the Dencun upgrade to be worked on over the next few weeks. Finally, developers discussed plans for moving onto Devnet #9 and minor outstanding issues for the launch of the Holesky testnet.

Devnet #8 Updates

All EL and CL clients have been onboarded to Devnet #8. However, the two EL clients that have not yet been spun-up to run validators on the testnet are Besu and Erigon. Busa noted that validators running Besu software will be created shortly, while he is still working on testing out block production capabilities with the Erigon client. Every other client is being tested for block production on Devnet #8 and all appear to be doing well, except the Prysm (CL) client. Terence Tsao, a developer for the Prysm team, shared color on the issue with the client. “Essentially, the processing of the block gets stuck,” said Tsao, adding that a fix for the issue would be released by his team today. A live view of Devnet #8 and its network activity can be found here. Given the stability of the network, Busa recommended starting fuzz testing on Devnet #8.

Dencun Testing Overview

Jayanthi presented a Dencun testing overview document detailing specific areas of testing for the Dencun upgrade that will be developed over the coming weeks. Categories of testing include:

  • EVM-relating tests for the Dencun EIPs that impact the EVM,

  • Sync testing for blocks and blobs between both EL and CL clients,

  • Chaos tests to assess varying chaotic situations such as EL/CL client failures, network resource stress, and time faults,

  • Public tests including more Devnets and shadow forks,

  • MEV-related tests,

  • Regression tests,

  • Bad block and blob tests,

  • Blob specific tests.

Jayanthi encouraged developers on the call to reach out to his team if there are any areas of testing or testing tools that should be added to the overview document.

Devnet #9 Updates

Then, developers briefly discussed plans for the launch of Devnet #9. Ryan confirmed with other client teams that the only major change in code specifications between Devnet #8 and Devnet #9 would be the deployment strategy of EIP 4788. For more information on EIP 4788’s deployment strategy, refer to prior call notes. EthereumJS (EL) and Lodestar (CL) client developer Gajinder Singh suggested that Devnet #9 test the block production flow with MEV relays and third-party block builders. Geth (EL) developer “Lightclient” recommended waiting until all Dencun-related specification changes are finalized and merged into the “master” repository before launching Devnet #9.

Developers also shared thoughts on how to proceed with testnets after the launch of Devnet #9. Ryan suggested that Devnet #9 be a short-lived testnet before moving on to testing more mainnet-like environments, presumably public testnets such as Holesky and Sepolia. Besu (EL) developer Justin Florentine emphasized the importance of alerting Layer-2 protocols to a testnet where they can try out operations and workflows with the finalized version of EIP 4844 code. Ryan also emphasized the importance of a testnet launch geared towards validator node operators and supporting the testing of their set-ups before mainnet launch. Jayanthi mentioned that one of the forthcoming testnets should also assess the performance of EIP 4844 with a target of 3 blobs, and a maximum of 6 blobs, per block, instead of the 2/4 limit. Lightclient questioned whether finalized Dencun specifications should be tested on Devnet #9 first or on the Holesky testnet once it launches in three weeks’ time.

Ryan said that developers should try to come to an agreement on a rough testnet schedule for Devnet #9 and other forthcoming testnet upgrades during the next ACD call. Before moving on to discussions about Holesky, a developer by the screenname “Sean” noted a minor proposed change to block builder specifications needed review. Ethereum Foundation Researcher Alex Stokes said that he would review it after the call.

Holesky Discussion

As a final point of discussion, Busa ran through three outstanding issues for the launch of the Holesky testnet.

  1. Developers agreed to adjust the epoch time for the Shanghai/Capella upgrade on Holesky from 10 to 256 in efforts to mirror the environment of Ethereum mainnet more accurately.

  2. They agreed to use a validator ejection balance of 28 instead of 16.

  3. To complete a checkpoint sync of the Holesky testnet, developers agreed to support a compressed version of genesis state.