Glamsterdam fork on Hoodi cancelled: how to check whether your Ethereum staking is affected

The test fork that Ethereum developers had been working towards since early October is not happening. On October 22, the core developers cancelled the activation of the Glamsterdam upgrade on the Hoodi testnet planned for Monday, October 26. The reason is not a bug in the software but a security finding from the audit firm ChainSecurity concerning two improvement proposals that form part of the upgrade. No replacement date has been set, either for Hoodi or for the mainnet.

If you hold Ether, nothing changes in day-to-day terms for now. Anyone running a node, operating a validator or testing applications on Hoodi, by contrast, has to adjust the calendar and check their software versions. This article sets out what exactly was cancelled, why the auditors raised the alarm and which date is realistically on the table now.

Hoodi fork cancelled: October 26 is off

Glamsterdam was due to take effect on Hoodi on October 26 at 17:42:48 UTC, at Hoodi epoch 132,352. That activation has been removed from the revised configuration. Cancelled along with it are the activation of the Amsterdam execution layer and that of the Gloas consensus layer. The planned increase of the Hoodi gas limit to 200 million units has also been withdrawn.

The cancellation affects future configurations only. Whatever has already been recorded in blocks on Hoodi stays as it is; nothing is being rolled back. Equally untouched is the Glamsterdam fork on the Sepolia testnet, which took place on October 6 and has been running since.

What a testnet fork actually achieves

A testnet is a full copy of the Ethereum blockchain that works with worthless tokens. Hoodi has served since spring 2026 as the last stage before the mainnet: this is where client teams and staking providers check under realistic conditions whether their software survives a fork. If such a test falls through, everything that follows necessarily shifts. The mainnet date hangs on the testnet date, not the other way round.

EIP-8037 changes how new storage is charged for gas

Gas is the unit of account with which Ethereum measures and prices the effort of a transaction. EIP-8037 is titled «State Creation Gas Cost Increase» and makes creating new data on the blockchain more expensive. To that end it introduces a price per byte, which it puts at 1,530 gas, and accounts for the cost of creating storage separately from the rest of execution. Transaction and block limits therefore apply per dimension rather than to a shared pool.

According to the official EIP page, the proposal is at «Last Call» status with a deadline of November 1, 2026. It depends on EIP-8038, which redefines the cost values for accesses to existing storage. Those values are not yet final according to the same page. The two proposals together were to take effect with Glamsterdam.

Gas prices as the basis of security checks

Gas prices in Ethereum are not purely a question of fees. Many contracts read during execution how much gas is left and derive a decision from it. If an upgrade shifts those values, it shifts the basis of such decisions with them. That is exactly where the finding comes in.

Insufficient gas griefing: the attack route ChainSecurity describes

ChainSecurity published its analysis on October 13. Insufficient gas griefing refers to an attack in which someone deliberately sends too little gas so that an otherwise legitimate operation fails. Contracts traditionally defend themselves against this by checking the remaining gas: if there is still enough, the failure was genuine; if hardly any is left, too little was probably supplied, and the operation may be repeated.

Under the new rules, more gas can flow back to the calling contract after a failed external call than before. The contract then carries on where it would previously have stopped, and its check reaches the wrong conclusion. According to Cointelegraph, the auditors warned that the changes could disrupt existing security checks that rely on the remaining gas.

The audit firm describes the practical consequences as permanent: an application could book a failed operation as definitively failed, delete a user's claim or block a retry. Affected above all are bundled transactions and payouts, that is, sequences in which several steps hang together and the behaviour in the error case decides the outcome.

13 affected contracts and the confidential disclosure

ChainSecurity states it has found 13 examples of this pattern in contracts already live. Which applications those are is not public. The firm says it is working with the affected teams under a confidential disclosure, the usual procedure in which developers get time to fix things before details become known.

That approach explains why the developers stopped the test run instead of pushing it through. A fork on Hoodi would have armed the new gas rules in an environment where teams mirror their production logic. As long as it is unclear how many contracts in the mainnet carry the same pattern, every further step would be a bet on a number nobody knows.

Gloas, Amsterdam and the 200 million gas limit are on hold

Glamsterdam bundles several changes. On the consensus layer there is Gloas, on the execution layer Amsterdam. Its content includes the separation of block proposal and block building, fixed in the protocol itself, as well as block-level access lists. Both building blocks aim to make block building more predictable and to decouple validators' work from specialised builders.

Raising the gas limit to 200 million was the most visible part of the Hoodi test. On Sepolia, Glamsterdam has been running with an increased limit since October 6. A higher gas limit means more computation per block and therefore more throughput, but it also raises the demands on node operators' hardware. That is precisely why it is trialled in stages and on testnets before it goes to the mainnet.

On the mainnet, operations continue unchanged

The mainnet runs on unchanged. On the account of the reports cited, the Ethereum Foundation has not stated that this upgrade would have damaged the mainnet, and there is no indication of a problem in live operation. Anyone holding Ether in a wallet of their own, keeping it with an exchange or staking through a provider does not have to change anything because of this decision.

Ether trades at around 2,370 dollars on October 24, in the range of recent days. No price reaction to the cancellation can be read out of the data; the market moved in step with Bitcoin and the US equity indices during this period. If you have followed the roadmap to this point, the state of play before the cancellation is in our October 22 article on the planned Hoodi fork and the 200 million gas limit. October 26 still stood as the date there; that is the point which has now lapsed.

Staking through a provider: the role of the service

Staking Ether through an exchange or a specialist service means not carrying the client risk yourself. The provider updates its software and is contractually liable for operations. What stays relevant is how transparently it communicates such dates and how quickly it reacts to a postponed fork. An overview of the terms and conditions of the common providers is in our staking platform comparison, including the lock-up periods that decide, when it matters, how mobile your holdings are.

Solo stakers and node operators: the client versions by October 26

The case is different for everyone running software themselves. If a client or a configuration of your own keeps the old activation time, it heads on Hoodi towards a fork the rest of the network is not taking part in. The result would be a split from the current state, with all the consequences for syncing and validator operation.

For that reason the update by October 26 still applies even though the fork is off: the new versions carry the cleaned-up configuration. According to the revised overview, that means Lodestar in version 1.50.1 on the consensus layer and Besu 26.10.1 as well as Nethermind 2.2.1 on the execution layer; Ethrex is available as pre-release 30.0.1-rc.2. Instructions for the remaining clients were still outstanding at the time of reporting. Affected alongside the execution and consensus client are the validator client and external block-building infrastructure.

Application developers and the test calendar

Teams that wanted to test their contracts against Glamsterdam have lost their basis. The developers have expressly asked that tests no longer be planned around the cancelled date. Anyone running contracts that read the remaining gas to decide on a retry should review that logic regardless of the fork date. ChainSecurity's findings describe a pattern, not a one-off.

The December date for the mainnet is open

Before the cancellation, according to consistent reports, the developers had discussed December 1, 8 and 9 as possible mainnet dates. None of them was confirmed. The agenda of the October 22 meeting provided for choosing a date the following week, provided the Hoodi test succeeded. That condition has not been met.

A single report names a revised Hoodi date after Devcon in early November and still considers a mainnet launch in December possible. The other sources do not confirm that statement, so it should be read with caution. All that is solid so far is that no new date has been set.

Realistically, the timetable hangs on two questions: how quickly the affected projects adjust their contracts, and whether the developers refine the gas rules so that the old checks still hold. Neither can be pinned to a date. With Fusaka last December, several weeks lay between the final testnet fork and the mainnet launch; a similar span would be no surprise here either.

Enshrined PBS and block-level access lists: what Glamsterdam contains

Behind the two technical terms lie two strategic choices. Enshrined proposer-builder separation anchors in the protocol itself what today runs through voluntary software: the separation between the validator proposing a block and the service assembling it. Block-level access lists announce in advance which data a block will touch, so that clients can do parts of the work in parallel.

Both aim at robustness rather than at visible new features for end users. For investors, Glamsterdam is therefore less an event with price impact than a building block in the network's long-term development. A delay of a few weeks changes little about that reading, but it does show that the developers put security findings ahead of the timetable.

Glamsterdam cancellation: without a replacement date the December fork stays open

1. Check your software version if you run things yourself. Update the execution, consensus and validator client to the versions named and make sure no configuration of your own carries October 26 as the activation time. If you leave staking to a service, the providers' terms are in the staking comparison.
2. Separate custody and trading. The cancellation does not touch your holdings, but it is still a good occasion to take stock. Which devices are suitable for self-custody is shown in the hardware wallet comparison.
3. Watch the calendar, not the price. The next solid step is a new Hoodi date. If you want to buy more or rearrange in this phase, compare the fees and deposit routes of the crypto exchanges first, rather than betting on a fork rally nobody has scheduled.

(As of October 24, 2026. This article is not investment advice. Prices and fee structures change; check the terms with the provider before you buy.)