Mentioned in This Article
1. Ethereum upgrade loses its Hoodi test date
Ethereum developers have canceled the October 26 activation of Glamsterdam on the Hoodi testnet, removing the last public rehearsal planned before the network's next hard fork reaches mainnet. Christine D. Kim, founder of the research and advisory firm Protocol Watch, announced the change on October 22, four days before the scheduled launch, and tied it to security issues that the audit firm ChainSecurity found in the upgrade's gas accounting.
The cancellation matters beyond the testnet itself. Ethereum upgrades move through a fixed ladder of test networks, and the result of each rung decides whether the next one proceeds. Hoodi was meant to be the final step before developers picked a mainnet date, so pulling it leaves the upgrade without both its test and the trigger for choosing when it would go live.
No replacement date for Hoodi has been named. Neither Kim's posts nor the rollback of the network's configuration files pointed to a new schedule, which means the question of when Glamsterdam arrives on mainnet is open again.
2. What did ChainSecurity find in the gas accounting changes?
ChainSecurity, a smart contract security firm, published an analysis on October 13 that concentrated on two proposals inside the upgrade, EIP-8037 and EIP-8038. Both alter how Ethereum charges gas, the unit that prices computation, for creating state and for accessing it. State is the data the network must store and look up, such as account balances and contract storage, so the cost of touching it shapes how applications behave.
Gas is central to this debate because it is both a price and a safety mechanism. Every operation on Ethereum consumes a set amount of gas, and a transaction fails if it runs out before finishing. Developers choose how much gas to send with a call, and contract authors often write checks that depend on how much will remain at a given point. When an upgrade changes the cost of creating or reading state, the amount of gas left at those checkpoints changes too, and logic that depended on the old numbers can take a different branch.
The firm's central warning was that contracts could act differently after the fork even though nobody had changed their code. In its words, a failure path that previously reverted could now complete. A revert is the mechanism by which a failed operation is undone and reported as an error, and many applications are written around the assumption that a particular failure will always end that way. If the same failure now completes instead, an operation that a program could once simply retry may turn into a permanent failure.
That kind of change is hard to spot because it does not require a bug in any single contract. It emerges from the interaction between the new pricing rules and code that was written under the old ones, and it can affect software that has been running safely for years.
3. A call for reassessment before the October 22 meeting
Ahead of the October 22 All Core Devs Execution call, the security researcher pcaversaccio asked developers on GitHub to discuss the ChainSecurity findings. He argued that the results call for a reassessment of this hard fork, while adding that the decision was not his to make.
The All Core Devs calls are the regular forums where client teams and researchers coordinate scheduling, and no single person controls the outcome. Within hours of that call, the Hoodi activation had been withdrawn.
Public scrutiny of this kind is a normal part of how Ethereum coordinates change. Proposals are written as Ethereum Improvement Proposals, discussed in open forums and implemented independently by several client teams, so concerns from outside reviewers can reach the people who set the schedule quickly and in the open.
4. What do node operators need to do?
Node operators running Hoodi nodes have two choices, according to a message Kim posted on X shortly before her announcement of the cause. They can revert to an older software release that does not have the Hoodi upgrade scheduled, or they can wait for the team behind their execution-layer client to publish an update that removes it.
The distinction exists because merging a code change does not alter software that is already deployed. The developers of the Besu client merged a pull request titled “Unschedule Amsterdam on Hoodi” on October 22. Yet the pull request itself cautions that Hoodi nodes running Besu 26.10.0 will still activate Amsterdam at the scheduled timestamp until they move to a release containing the change.
Barnabas Busa, a DevOps engineer at the Ethereum Foundation, handled the network-wide side of the rollback by unscheduling the fork in Hoodi's shared configuration files. The edit removed the Amsterdam start time of October 26, the Gloas fork at epoch 132,352, and a planned increase of the gas limit to 200 million at that epoch. The files were returned to the state they were in before the upgrade was scheduled.
Any node that is not updated or reverted would try to follow a fork that the rest of the test network is no longer heading toward, so clear guidance to operators was a practical necessity rather than a courtesy.
Ethereum runs on several independently built client programs, each maintained by a different team, and a coordinated change like a hard fork requires all of them to ship matching updates. That diversity protects the network from a single point of failure, but it also means a schedule reversal has to be carried out client by client, as the Besu pull request illustrates.
5. How the December mainnet dates depended on Hoodi
The agenda for the October 22 call listed three candidate mainnet dates: December 1, December 8 and December 9. The plan was to choose one on the following week's consensus-layer call, provided the Hoodi activation went smoothly.
With Hoodi withdrawn, that sequence cannot run as written. The December candidates were conditional on the test, and the test no longer has a date. Developers could still reach a December mainnet launch only if a new Hoodi activation is scheduled soon and passes without trouble, which would also require the gas accounting concerns to be resolved or ruled out first.
The story does not say whether the proposals will be revised, delayed or defended as designed. What it does establish is that the schedule published only days earlier is no longer in force.
Choosing a mainnet date is itself a significant commitment. Once a date is set, exchanges, wallet providers, infrastructure companies and application teams plan their own upgrades around it, so developers prefer to announce it only after the final test network has run cleanly. That is why the Hoodi activation carried so much weight in the original plan.
6. What is Glamsterdam, and where does the upgrade stand?
Glamsterdam is the name given to the combined next hard fork, pairing an execution-layer upgrade called Amsterdam with a consensus-layer upgrade called Gloas. Ethereum names its upgrades this way by joining the two layers' labels into a single title.
Among its contents are EIP-7732, which moves proposer-builder separation into the protocol itself, and EIP-7928, which introduces block-level access lists. Proposer-builder separation divides the job of proposing a block from the job of assembling its contents, and building it into the protocol replaces arrangements that now rely on outside software. The gas accounting proposals at the center of the ChainSecurity review sit alongside these changes in the same fork.
The upgrade went live on the Sepolia testnet on October 6, and the Ethereum Foundation announced the Hoodi schedule on October 16. The rollback therefore reverses a public timetable that had been in place for only about a week.
7. Why test networks matter before an Ethereum upgrade
Hard forks change the rules every node on the network must follow, and a mistake on mainnet can be expensive because real value depends on the contracts that run there. Testnets such as Sepolia and Hoodi let developers, client teams and application builders watch the new rules operate under realistic conditions before anything of value is at stake.
The episode shows that process doing what it is designed to do. An independent security analysis, published while the schedule was still moving, prompted a public request for reassessment, and developers responded by pulling the activation rather than proceeding on the calendar. The cost is delay; the alternative would have been to discover any behavioral change after the rules were live on the main network.
For contract developers, the practical lesson in ChainSecurity's warning is that a fork can alter how existing code behaves without anyone touching that code, which is why the firm's review focused on failure paths and not only on the headline features.
Gas repricing is a recurring theme in Ethereum's history, because the cost of an operation is meant to reflect the burden it places on every node. Changes that make state creation or access cost more or less can improve the network's capacity and long-term health, but they also reshape the economics and behavior of existing applications. That tension explains why even a well-motivated pricing proposal draws careful review before it is allowed near mainnet.
8. What comes next for Glamsterdam
The immediate steps are procedural. Client teams need to publish releases that drop the Hoodi activation, and operators need to either update or roll back, with Besu users running 26.10.0 among those who must act. Developers must then decide how to treat EIP-8037 and EIP-8038 in light of the audit.
Until a new Hoodi date is set, the December 1, December 8 and December 9 options remain candidates only in name. The next consensus-layer call, where the mainnet date was due to be chosen, now has a different question in front of it: whether the upgrade is ready to be tested publicly at all.