P2P2026-07-15_004

P2P Networking #004

2026-07-15 1 decision 390 transcript lines

Transcript

At a glance

Sparse blob pool enabled optionally in Glamsterdam devnet 7 (not mandatory)
Sparse blob pool enabled on Glamsterdam devnet 7 — targeting next week
Lighthouse release — July 16, 2026 (tomorrow)verify date

Detailed notes

Client Updates

  • •Lighthouse: release tomorrow; mainnet traffic halved on nodes running partial-message support (~150/2000 nodes)verify date00:04:02
  • •Geth: first sparse blob pool PR merged to master; subsequent PRs (tests, protection) pending other clients' implementations00:06:03
  • •Erigon: switching to sparse blob pool testing; discv5/v4 improved with fork-hash filtering; targeting Glamsterdam devnet 7 enablement next week00:07:50

Spread Protocol

  • •Spread: GossipSub extension providing sender anonymity + lower latency simultaneously via geography-aware cluster routing (Vivaldi coordinates)00:11:13
  • •Performance-optimized config cuts mean stretch/latency ~20–30% vs GossipSub; tail behavior improved considerably00:24:07
  • •Privacy-equivalent config holds attacker accuracy ~30% vs GossipSub's 38% at 10% adversarial nodes, while still faster00:25:16
  • •Key open concern: reduced early fan-out may allow message propagation failures; repropagation fallback exists but censorship resistance not yet analyzed00:34:28
  • •Raúl: attestation on-chain confirmation provides a natural feedback loop to score peer routes; Spread could leverage this as a simple callback from CL clients00:37:05

Action items

  • •Review Spread spec PR and implementation PR; provide feedback to authorsCL client teams / P2P researchers
  • •Measure Spread propagation degradation under malicious (non-forwarding) node scenariosSpread authors (matheusfranco)

Agenda

From the call issue #2161Show
  • Client updates
  • SPREAD by Matheus

Key decisions

  • Sparse blob pool enabled optionally in Glamsterdam devnet 7 (not mandatory)

    Glamsterdam devnet 7

AI Disclaimer: Some content or metadata on EIPsInsight may be AI-inferred or automatically compiled. If you find any discrepancy, please contact us at dev@avarch.org.