Skip to content
Blockchain Posts

protocol

Ethereum pauses Glamsterdam hardfork process over 8 broken invariants

Marius van der Wijden's Nov 6 Foundation post says AllCoreDevs will not fork Hoodi until EIP-8037/8038 breakages are resolved. Mainnet never had a target.

by 4 min read
protocolethereumglamsterdamsecurity

Ethereum's AllCoreDevs have paused the Glamsterdam hardfork process until the gas-repricing invariants broken by EIP-8037 and EIP-8038 are resolved. The decision was published on November 6, 2026 in a Foundation blog post by Marius van der Wijden, which names eight specific invariants the upgrade breaks and an open mailbox (glamsterdam@ethereum.org) for affected teams to flag exposure. Journal du Coin's Nov 7 write-up confirms the ACD decision.

What's new relative to the October pause

The Oct 23 Hoodi postponement dropped a single testnet activation (Hoodi, Oct 26, 17:42:48 UTC). The Nov 6 post is broader:

  • No Hoodi fork until resolution. Not just the Oct 26 slot — the Hoodi activation is deferred with no replacement date.
  • No mainnet target. Van der Wijden writes that mainnet "never had an official date to cancel"; the whole roadmap step is on hold.
  • Eight enumerated invariants. ChainSecurity's Oct 13 note described a pattern; the Nov 6 post links a GitHub repository that lists eight concrete invariants Glamsterdam will break, and asks contract authors and auditors to check each one.
  • Hegota continues. Development of the next upgrade (Hegota) is explicitly unaffected; the Glamsterdam fix will land before Hegota work merges in.

The post does not name any production protocol that is exploitable — only the invariant list and the request to check.

The two culprit EIPs

  • EIP-8037 — state-growth metering. A separate state-gas reservoir; once exhausted, state-touching operations are repriced to punitive levels rather than reverting.
  • EIP-8038 — cold-access repricing. Changes the cost of first-touch reads and writes to accounts and storage slots.

Both were in the Sepolia bundle that activated October 6, 2026. The Nov 6 post keeps them in-fork; it does not propose dropping them.

ChainSecurity's analysis — the trigger for the Oct review — flagged 13 real-world examples of contracts whose behavior changes after the fork. The same analysis showed that creating a new storage slot after the state-gas allowance is exhausted costs 110,020 gas against 22,100 gas under current rules, a ~5× jump that pushes well-tuned contracts past their gas budgets.

Numbers

- Nov 6 blog post author       : Marius van der Wijden
- Invariants listed as broken  : 8
- Affected-party mailbox       : glamsterdam@ethereum.org
- ChainSecurity real-world
  examples found               : 13
- Worked gas-cost delta
  (new slot after state-gas
   allowance spent)            : 110,020 gas vs 22,100 gas today (~5×)
- Sepolia activation (done)    : 2026-10-06
- Hoodi activation             : paused, no new date
- Mainnet target               : none set
- Status of Hegota work        : continuing

What to watch

  1. The GitHub invariant list. The Foundation post links a repository with the eight items. Contract developers should run through each for exposure; auditors integrating Glamsterdam coverage should use the list as their test corpus floor.
  2. Client-team EIP revisions. Geth, Nethermind, Besu, Reth, Erigon on execution; Lighthouse, Prysm, Teku, Lodestar, Nimbus, Grandine on consensus. Pull requests tagged glamsterdam on ethereum/execution-specs and the client repos will carry proposed fixes.
  3. Next ACD call. The agenda sits in ethereum/pm. The post does not name the next ACD meeting at which the fork could be re-scheduled.
  4. Contract-side mitigations. ChainSecurity's tool for spotting exposed contracts (referenced in the post) is the first line of triage for teams running on Ethereum mainnet. The Foundation is asking for direct reports to glamsterdam@ethereum.org.

Context

Glamsterdam's gas-model rewrite is the biggest reshape of Ethereum fee mechanics since EIP-1559. The Sepolia activation on Oct 6 held; the Hoodi slip in late October was the first material delay. The Nov 6 escalation — the whole fork process now pending a formal fix on invariants that ChainSecurity surfaced, not just the Hoodi testnet — is the first time in Glamsterdam's cycle that AllCoreDevs have taken the mainnet target off the roadmap entirely. The two prior Glamsterdam coverage milestones both landed on schedule; EIP-8037 is the item that has now cost two scheduled dates.

The Foundation's preferred resolution path is a change to the two EIPs themselves rather than dropping them, per the post — the gas-pricing redesign remains in scope, but the AllCoreDevs will not advance it until the eight-invariant list is cleared.

Sources:

Related stories