EL7.AI
CryptoMedium•17 November 2026•
4 min read

Glamsterdam Simulation Uncovers Two Cases of Inter-Contract Transfers Being Skipped on Sepolia

Key Facts

1The gas range in which the transfer to the other project was skipped ran from 717k to 1,105k on Sepolia.
2Creating a new cold storage slot could cost 110,020 execution gas, compared with 22,100 in Osaka.
3EIP-8037 aims to limit state database growth to 120 GiB per year.
4The repricing schedule is based on a target supporting an increase of roughly 3 times in baseline throughput.

A simulation published on GitHub showed that some transactions on Sepolia, which uses Glamsterdam upgrade rules, can succeed even as a payment from a contract to another component or project falters. The findings concern the distribution of funds between contracts, not users bypassing gas fees paid by Ethereum transactions. In one case, a payment call wrapped in try/catch handling could be skipped when a certain gas limit was selected. The simulation’s author said they found no comparable limit that would allow the transaction to proceed without paying the component what it was owed on mainnet in the same test. The findings come as Glamsterdam remains under test on devnets, before any mainnet activation.

In another simulation of payment splitting on Sepolia, the repository showed that certain gas limits led to a successful payment while skipping the transfer to the other project and returning the funds to the paying project. The range in which this result appeared ran from 717k to 1,105k gas. The repository’s comparison with Ethereum mainnet showed no similar window. This case shows that a successful outer transaction does not guarantee, in this simulation, that every intended sub-transfer within it is completed. The result could be relevant to decentralized finance protocols that distribute payments among multiple contracts, particularly contracts that are difficult to modify after deployment.

The published cases show behavior linked to changes in gas costs and how contracts handle reverted calls, not evidence of a flaw that allows network fees to be avoided. ChainSecurity said that refunding state gas after a reverted call may leave the outer handler with enough gas to continue along the failure path. If the contract records the failure and removes the retry option, an operation that could have been repeated may become a permanent failure. This means the design of error-handling paths may determine whether a failed payment can be recovered. That matters more for applications that rely on sequential calls between contracts to complete the distribution of funds.

The tests coincide with a proposed repricing of state storage and access under Glamsterdam. ChainSecurity said creating a new cold storage slot could cost 110,020 execution gas when the cost of creating state is charged to gas, compared with 22,100 in Osaka. Ethereum’s roadmap also says EIP-8037 aims to limit state database growth to 120 GiB per year. The Ethereum Foundation says repricing state operations paved the way for a higher gas limit, and that the new fee schedule is based on a performance target supporting an increase of roughly 3 times in baseline throughput. These changes put contract behavior under a new cost schedule, making the testing of gas assumptions part of upgrade preparation.

The Ethereum Foundation said replaying historical transactions under the proposed gas schedule showed a small group of contracts relying on assumptions that could change, while the vast majority of contracts were unaffected. It added that most observed cases can be fixed by raising the gas limit. This provides broader context for compatibility risks that may emerge with repricing, without changing the scope of what the published Sepolia simulation showed about the outer transaction succeeding while an internal transfer failed. Glamsterdam also includes the ePBS proposal, under which the data publication window will expand from 2 seconds to roughly 9 seconds to support processing a larger number of transactions and layer 2 data, according to the official roadmap. That change remains part of a broader upgrade package under test, not a direct result of the published cases.

The Ethereum Foundation said the repricing schedule is running on devnets, and that public network testing will precede mainnet activation, advising contract teams to test their applications before then. The roadmap page describes Glamsterdam as the next upgrade after Fusaka, with testing continuing on devnets. It lists the fourth quarter of 2026 as the expected mainnet timing, but says the date has not yet been finalized. The published payment simulation remains a test on a test network and does not provide evidence of bypassed gas fees or an impact on network stability. Assessing its practical impact will depend on how projects handle failed payment paths before the upgrade moves to public networks and then mainnet.

Open on chart