All news is rigorously fact-checked and reviewed by leading blockchain experts and seasoned industry insiders.
- Ethereum developers pulled Glamsterdam from the Hoodi schedule over EIP-8037 compatibility risks.
- The concern centers on deployed contracts that nobody can change.
- Some failure paths could start succeeding instead of simply running out of gas.
- The planned gas limit increase was frozen together with the fork.
Ethereum core developers removed the Glamsterdam upgrade from the Hoodi testnet schedule during All Core Developers Execution call #248 on October 22. The decision came four days before activation, which was due at 17:42:48 UTC on October 26. The Ethereum Foundation confirmed that the Amsterdam and Gloas activations are now unscheduled, along with the planned jump in Hoodi’s gas limit to 200 million, and it has not set a new date. The trigger is EIP-8037, the proposal that reprices the gas needed to create new state on Ethereum. According to the Foundation, developers paused the fork to investigate how that change interacts with smart contracts already running on mainnet, code that no one can edit and that was written under the old gas rules.
A price list from 2021
Ethereum last adjusted gas prices for state operations in the Berlin fork in 2021, according to a repricing impact post from the Ethereum Foundation published in August. The state is the permanent record of every balance, storage slot and piece of contract code that each node has to keep, and it has grown considerably since then. The gas limit increases of the past year sped that growth up.
EIP-8037 makes the creation of new state pay for its long-term burden on node operators by metering it as a separate “state gas” dimension, while its companion EIP-8038 changes what it costs to access state that already exists. Individual operations show the scale of the change most clearly.
Per-operation charges, not average fees. Source: ChainSecurity
The 110,020 figure applies when the new state-gas reservoir cannot cover the charge, so real-world costs will vary. Even so, a nearly fivefold increase for writing a fresh storage slot puts pressure on every gas assumption that older code carries.
Why the gas limit froze with it
The 200 million gas limit and EIP-8037 are linked by design. At the fork epoch, Hoodi validators would have started moving the gas limit from 60 million toward a 200 million target, more than three times today’s capacity. If capacity rises without the repricing, the state database grows faster, and that growth is exactly what the Foundation is trying to contain. Its August post argued that gas reflecting real resource costs is what allows block limits to keep rising without degrading the network for node operators. Activating one change without the other would mean testing a configuration nobody plans to ship on mainnet, so developers held back both.
The failure developers did not plan for
The Foundation expected some breakage, and in August its protocol teams split affected contracts into two groups. The first could be fixed by supplying a higher gas limit. The second was potentially broken even with a much larger one, and these contracts typically depend on fixed stipends such as the 2,300 gas Solidity passes with transfer and send, hardcoded gas values inside calls, or logic that branches on the gas remaining.
ChainSecurity’s October 13 review, titled “Glamsterdam Hard Fork: When Spent Gas Comes Back”, described a different kind of risk. Under the new accounting, gas reserved for state creation can stay available, or come back immediately, when an external call fails or when an intermediate contract in a call chain reverts. A contract that measures leftover gas to work out what went wrong may draw the wrong conclusion. A failure-handling routine that used to run out of gas could also run all the way through. The checks at stake exist to stop what ChainSecurity calls insufficient gas griefing, where an attacker deliberately supplies too little gas so that a valid operation fails and the contract records it as a permanent failure instead of leaving it open to retry.
That difference shapes how the problem can be tested, because a transaction that suddenly runs out of gas shows up as soon as historical traffic is replayed under the new rules. A transaction that now succeeds where it used to fail is much harder to catch, since nothing visibly breaks. ChainSecurity’s work describes plausible mechanisms, and it has not shown an exploit against any live application.
Hoodi was the final rehearsal
Glamsterdam has run on Sepolia since October 6, which made Sepolia the first public network with EIP-8037 active. Hoodi carries more weight as the validator-focused testnet that mirrors mainnet staking conditions, and Ethereum Foundation researcher Toni Wahrstätter described it as the last testnet before mainnet. The pause also holds back the upgrade’s headline features, which are not part of the current concerns: enshrined proposer-builder separation under EIP-7732 and block-level access lists under EIP-7928. Both are among the 18 upgrade EIPs the Foundation listed on October 16.
Who has to act, and who does not
- ETH holders and mainnet users: no action required. No mainnet date had been set, so none has moved. The mainnet step cannot be scheduled until Hoodi completes.
- Hoodi node operators: they need builds that include the revised schedule, such as Lodestar v1.50.1, Besu 26.10.1, Nethermind 2.2.1 and the Ethrex v30.0.1-rc.2 prerelease. Earlier Glamsterdam-ready builds, including Nethermind 2.2.0, still contain the October 26 activation.
- Wallets, bundlers and contract teams: teams that recalibrated gas estimation against Sepolia over the past three weeks may have to redo that work if the specification changes.
The next ACDE calls have to decide whether the risks can be handled through implementation fixes and wider contract scanning, or whether the text of EIP-8037 itself must change. The question is familiar, since earlier in the cycle developers were split over whether a new state-growth metering mechanism belonged in Glamsterdam at all, and alternatives such as EIP-8075 were discussed before the current design won out. If the specification is rewritten, part of that debate reopens, including whether the repricing could move to Hegotá, the next upgrade on Ethereum’s roadmap.
Be the first to comment