Ethereum developers have paused the Glamsterdam upgrade's activation on the Hoodi testnet over concerns about how already-deployed smart contracts behave under the upgrade's new gas pricing rules. The decision was taken on All Core Devs Execution call #248, the Ethereum Foundation said in its October 22 announcement.

Key points from the decision:

• The state cost of a new storage slot rises from 20,000 to 97,920 gas.
• Hoodi's planned 200M gas limit increase has also been removed.
• No Glamsterdam mainnet activation date has been set.
• EIP-8037's Last Call review ends November 1.

The last testnet step before mainnet is on hold

Glamsterdam was due to activate on Hoodi on October 26 at 17:42:48 UTC. The postponement removes both the Amsterdam execution-layer and Gloas consensus-layer activations — the two halves of the upgrade — along with the schedule that would have raised Hoodi's gas limit to 200 million at the same time. It changes the testnet's future configuration without reverting its history, and no replacement date has been set.

During Fusaka, the previous upgrade, Hoodi was the third and final testnet to fork before mainnet, and the Foundation said mainnet would follow at least 30 days later, according to The Block. In that rollout, the final testnet fork was the last public checkpoint before mainnet. Glamsterdam has already passed a 200-million-gas rehearsal on Sepolia, where the fix was a Prysm update to validator defaults — a client configuration issue. This time the question is how contracts that are already deployed behave. The Foundation has not set a Glamsterdam mainnet date and says mainnet users and ETH holders do not need to act.

A storage price rise that cuts execution gas on some writes

EIP-8037 exists to slow the growth of Ethereum's state, the data every node must store. Its authors put that state at about 390 GiB in January 2026. They estimate that under today's pricing, a 200-million gas limit could push it past 650 GiB — the level at which nodes start to slow down — in under a year. Their answer is to charge 1,530 gas per byte of new state, raising the state cost of a new storage slot from 20,000 gas to 97,920 and of a new account from 25,000 to 183,600.

To keep large contract deployments possible, the proposal splits gas into two pools. Execution gas covers computation, and state gas covers new storage. Any gas a transaction supplies above 16,777,216 goes into a separate state-gas reservoir, which state charges draw on first. That reservoir is passed in full to every sub-call, and contracts cannot see it through the standard gasleft() reading.

Execution gas required for a cold write to a new storage slot, per ChainSecurity, using EIP-8037 and EIP-8038 figures:

• Today: 22,100
• Glamsterdam, reservoir pays the 97,920 state charge: 12,100
• Glamsterdam, reservoir empty: 110,020

In other words, the identical cold write can cost roughly nine times more execution gas with the reservoir empty than when the reservoir covers the state charge.

Gas can also flow back during execution. Clearing a slot created earlier in the same transaction immediately refills up to 97,920 gas, and a sub-call that reverts can hand back the state charges it paid. A contract's visible gas balance can therefore rise mid-execution. The EIP warns that contracts measuring work as the difference between two gas readings can get a negative result, and that contracts already deployed with that design cannot be fixed.

How an attacker could use the new rules

Security firm ChainSecurity set out the risk in an October 13 analysis. Contracts often catch a failed call and record the result, which lets a malicious caller deliberately supply too little gas so that a valid operation fails while the contract still has enough gas to save that failure. The technique is known as insufficient gas griefing, and it can permanently block a retry or remove a user's claim.

For contracts that rely on the failure handler itself running out of gas:

• Step 1: Attacker sends a call with too little gas.
• Step 2: The target operation runs out of gas.
• Today: The handler runs out too, everything reverts and the user can retry.
• Glamsterdam: Reservoir or refilled gas lets the handler finish and save the failure permanently.

Contracts that check remaining gas after a failed call can be fooled when refills or a reverting intermediary push the balance back above their threshold. Fixed gas budgets can also fall short, because EIP-8038 raises the cost of touching existing state: the first change to an existing warm slot rises from 2,900 gas to 10,100.

Not every defence breaks. ChainSecurity says a check made before the call still guarantees delivery of the requested gas, as long as the budget covers every execution path under the new prices without counting on the reservoir. It advises against relying on post-call gas checks at all. The Foundation's notice names only EIP-8037 and does not cite the firm's work, which also covers EIP-8038.

What operators and developers need to do

Hoodi operators must update execution and consensus clients before October 26, using releases whose notes explicitly confirm that the old activation has been removed. Version tables from the original announcement do not count as confirmation. As of October 23, the Foundation had listed updated releases for two clients, and said others would be added as they are confirmed.

Validators must update both the beacon node and the validator client, and anyone running external block-building infrastructure should follow their provider's instructions. Application teams should stop planning Hoodi tests around the original date.

For contract developers, ChainSecurity's checklist starts with every call that lets execution continue after failure. It covers tracing the full path, including proxies; finding slots set and cleared in the same transaction, and intermediaries that revert; re-checking fixed budgets; and forcing sub-calls to fail and comparing the outcome under both rule sets. Where possible, a failed call should revert the whole operation. Wallets and node software must also update gas estimation for the two-pool model, according to the EIP.

The next decisions sit with the core developers. EIP-8037's Last Call review — the final review window before a proposal is finalized in Ethereum's process — closes on November 1, the investigation may lead to specification changes, and a revised Hoodi date will be set on a future call.

This article is provided for informational purposes only and does not constitute financial or investment advice. Ethereum upgrade schedules and technical specifications may change during testing.

Source: Coindoo