ACDT#61: Call Minutes + Insights
The call discussing the final rounds of Fusaka testing
Good evening,
Last Monday, no client teams, except for one, were ready with their final Fusaka mainnet releases.
Today, all but one client team is ready.
Developers discussed the final rounds of testing they plan to run across all releases in the lead-up to December 3, the mainnet upgrade activation date.
While optimizations and minor fixes are needed in nearly all clients, developers’ primary focus is now on hardening their existing implementations rather than introducing new code that could require hotfixes.
Below is my full summary and insights on ACD Testing call #61.
Yours truly,
Christine D. Kim
(For background on the ACD governance process, refer to the Ethereum Governance 101 document in the ACD Toolkit.)
Fusaka Testing
The second and final blob parameter-only (BPO) hard fork is scheduled for activation on the Hoodi testnet this Wednesday, November 12.
Nethermind developer Marcin Sobczak said his team had to issue a second hotfix for their Fusaka mainnet release, v1.35.2. He said the prior release, v1.35.1, was undertested. The latest release removes undertested optimizations and non-critical fixes. These will be added to a future release after the Fusaka upgrade goes live on mainnet.
The Reth client team also issued a hotfix for their Fusaka mainnet release, v1.9.1.
There is a non-consensus critical bug in the latest Lighthouse client release, for which a fix is in the works.
Teku developer Enrico del Fante said he has been noticing unusual RPC requests on one of the Fusaka testnets. He said he would investigate and report back with more information on where these requests are originating from.
Ethereum Foundation (EF) Developer Operations Engineer Barnabas Busa reported an issue with Nimbus nodes on Fusaka Devnet-3. The nodes have stalled on the devnet. Investigations into the root cause of this are ongoing.
Fusaka Devnet-3 will be deprecated after the upgrade goes live on mainet.
“Kasey” from the Prysm team reported that their mainnet-ready release will be ready today.
Layer-1 Scaling
Sobczak is working on benchmarking the Geth and Nethermind clients for EIP 7904, general gas repricings.
Silva explained that there are no further scheduled calls for discussing gas repricing EIPs. There were two that happened over the past two weeks. More may be scheduled after Devconnect.
Glamsterdam
The first multi-client devnet for block-level access lists (BALs) is still delayed. EF Developer Operations Engineer Parithosh Jayanthi said developers expect to be ready for a launch end of November.
One of the reasons for the holdup has been a new test release by the EF STEEL team. Prior test releases only included tests specific to BALs. The latest release contains a broader set of tests that clients are failing and the testing team is working on fixes for. "Spencer-tb” on the STEEL team said that a new release will be out by this Wednesday at the latest.
EF STEEL Team Lead Mario Vega asked client teams to review the Hive test results for their BAL implementations and report back if any tests are buggy.
EF Researcher Justin Traglia said there have been a few changes in consensus specifications for enshrined proposer builder separation (ePBS). These changes will be included in the next consensus specs release. Client teams are expected to incorporate these changes for the ePBS Devnet-0 launch, which is likely to occur no earlier than mid-January.
The next ePBS breakout meeting on November 21 will be cancelled due to the developers meeting in person at Devconnect. The next online check-in for ePBS will be on December 5.
Fusaka Mainnet Readiness
Grandine client developer Saulius Grigaitis asked how developers were feeling about the number of bugs discovered in client releases recently.
Busa said only a few of the bugs were related to Fusaka code, and that he is not worried about the mainnet activation of Fusaka.
Jayanthi said hot fixes in client releases are expected.
Jayanthi asked whether client teams think any specific tests are necessary before the Fusaka mainnet activation on December 3.
Grigaitis suggested one more round of non-finality testing before the mainnet upgrade.
Nethermind developer Lukasz Rozmej asked for data on node performance when there are many pending blob transactions.
Kasey requested testing for pending blob-queue code paths and private blob transaction processing.
Busa noted that testing for private blob transactions can only be done on Fusaka Devnet-3, as his team does not run relays on public testnets like Hoodi.
Jayanthi noted that a new release of MEV-Boost will be available shortly, containing a bug fix.
Jayanthi said that his team is organizing a mainnet shadow fork before December 3. Busa confirmed that a final round of non-finality testing on Devnet-3 will also be scheduled.
Meeting Cancellation
Developers agreed to cancel the next two ACDT calls, as they will be meeting in person at Devconnect and also travelling for the holidays. They will resume their regular Monday calls on December 1.
Programming note: The ACD After Hours newsletter will also go on break during these dates, November 17 and 24, until the ACDT calls resume. There is no planned break for the ACD calls on Thursdays. So, you can still expect to receive new posts on these over the next two weeks!
🌕 That’s all for my summary of ACDT #61. Continue reading for pointed takeaways from the call, featuring direct quotes and additional context on key topics. To read the rest of the newsletter, make sure you are signed up for a premium subscription:
🌓 New to the ACD calls and want to learn more about Ethereum protocol development? Explore the ACD Toolkit, which is included with a premium subscription. It contains evergreen resources and materials that teach you the fundamentals of tracking the evolution of Ethereum like a pro:
🌑 I also offer professional consultations on Ethereum and Bitcoin protocol development. If you’d like to learn more, please visit my advisory services website:
(Quotes featured in this section may be edited slightly for grammar and clarity. For more information on the people quoted in this section, refer to the ACD Call Directory in the ACD Toolkit.)
Final client releases for the Fusaka upgrade on Ethereum mainnet are ready, but not optimized.
