ACD Testing Call #48 - Aug 11, 2025 - Quick Notes
Facilitator: Parithosh Jayanthi
Fusaka Update
Parithosh shared updates:
- Devnet-3 has been running for 10 days, with two non-finality experiments conducted last week.
- Experiment 1: Took out full nodes → ~1 hour of non-finality before recovery in the following epoch.
- Experiment 2: Subset of supernodes removed → ~half a day of non-finality, but finality regained within 20–30 minutes, showing rapid recovery.
- Results analysis showed no major issues; bandwidth limits appear fine and overall network health is good.
Next step:
Test with a subset of nodes (especially supernodes) capped at 1–2 Mbps bandwidth to simulate unreliable supernode participation. This is planned for next week.
Private Mempool Update
- Bharat shared that there’s a way to directly send transactions to builders, which is useful for private mempool testing.
Next step:
- Pari will set up spam testing for the private mempool.
- Once the force transition issue is resolved, Bharat will focus more on spamming tests.
More on Devnets
Devnet 3
Pari shared more updates:
- A fork of Prysm with a hook feature (allows changes) has been live on the network for 2 hours with no issues reported.
- Hooks can be extended for more scenarios — share your ideas here.
Devnet 4 Setup
- Setup coordinated with Sunnyside Labs, with additional testing planned here.
- Network is running fine; next BPO is scheduled in the next 2 hours to assess performance.
- Grafana dashboard available for monitoring.
- Nimbus encountered issues — Dustin is investigating, some details shared by Pari on the call.
- Pari clarified to Pawan that Devnet 4 has no non-finality tests planned but could add them if needed.
- Pawan suggested exploring client behavior under longer finality delays.
Next step:
- Parithosh will arrange a smaller devnet at the end of Devnet 4.
Holesky Release & Client Timeline
Parithosh shared updates
- Community feedback indicates no major issues with ad-hoc client releases for Holesky validators.
- James He relayed Preston’s note: “Releasing to holesky off develop isn’t difficult, but it messes with normal flow for releases”.
Next step:
- will be brought to the next ACD for further discussion.
Pari:
- One EIP-related PR pending: ethereum/execution-apis#678 — comments welcome before merging.
Gas Limit Testing Updates
- Progress can be tracked here: Syncoor Dashboard
- Mario shared the plan of another benchmark release this week.
- Further on updates, Mario shared that With the updated tool, the Nethermind team can pull the genesis, run benchmarks, and report results.
- Plans to extend benchmarking with another release expected soon.
Glamsterdam Updates
BALs
- No BALs devnet yet, though specs are ready and built on top of Fusaka.
Next Steps
- Toni & Pari will coordinate on setting up a devnet, possibly using Kurtosis.
- Further discussion planned for ACD on Thursday.
ePBS
- Local testing is available via a local branch.
- Ansgar: No strong value in combining ePBS, BALs, and repricing in the same devnet.
Next Steps
- Pari invited feedback on preferred Fusaka devnet setup.
- Justin Traglia: Plans to merge consensus-specs PR #4476 soon.
Sunnyside Labs Update
- No major updates today; another report expected tomorrow.
On Gas repricing
- Toni: Highlighted EIP-7904, EIP-7778, EIP-7981, and multidimensional gas EIPs for consideration.
- Pari: Asked if there’s a Meta EIP for gas pricing.
- Ansgar: Confirmed they can create one.
- Pari: Said a Meta EIP would be more useful than referencing individual EIPs.
Consensus Issues Found with Fusaka
- Marius raised consensus-related concerns for discussion.
- Mario: For EIP-7883, an update will be included in the next EEST release.
- Justin Traglia shared Related PR (in chat): ethereum/execution-spec-tests#1993
- Reth: Issues explained by Dragan are now resolved.
- Marius mentioned: Two issues found for ModExp gas cost, but they are unrelated.
- Pari asked if test coverage is sufficient.
- Mario: Extra coverage would be good; will prepare another devnet release for testing.
- Related PR: ethereum/execution-spec-tests#2005.
EIP-7825 Discussion
pk910: Asked if the transaction gas limit should also apply to eth_call and which clients have implemented it. Suggested directly adding the limit for checking.
- Marius(in chat): In Geth, the limit is configurable.
- Pari: Agreed to leave it as is for now and revisit the discussion later.
PS: This quick note is based on my following the call live and documenting it simultaneously. If you notice any missing updates or corrections, please share them here.