Ethereum transactions may sometimes appear successful but could conceal underlying issues.
According to the draft pull request #36084 on GitHub, a Geth developer has proposed a fix for the client’s gas estimation mechanism. The patch addresses an issue in the current estimator: gas limit estimates may be inaccurate because failures from internal calls between smart contracts are being ignored.
Although the transaction itself may still be completed, there is no evidence that the action the user intended to perform actually occurred.
What is a gas estimation issue?
The issue lies in a mechanism where one smart contract calls another, catches an error, and then continues to process refunds, log failures, or simply skip an item. Technically, the transaction succeeds, but the outcome for users differs.
This could lead wallets to suggest a gas limit sufficient to complete the transaction but insufficient to fulfill the actual work required internally. The developer also noted that EIP-8037 may make this situation more common, as gas refunded from failed internal executions could help fund fallback procedures.
What did Geth pull request #36084 change?
Based on this pull request, this change provides a stricter definition of "success" in gas estimation. This version of Geth will improve its handling of internal failures: instead of only checking whether a transaction completed, it will also account for failed internal calls and contract creations.
Even if the transaction completes, this estimate will be rejected if the number of internal failures it generates exceeds those from a reference execution under a higher gas limit.
However, the developer emphasized that an internal call that consistently fails should not cause the estimate to rise indefinitely. The purpose of the reference execution is to allow the estimator to preserve pre-existing failures while preventing new failures from being introduced due to insufficient gas. The proposal also introduces a separate Amsterdam reference to prevent unnecessary upward adjustments solely for accessing state gas reserves.
Developers stated that, following the Amsterdam update's new rules, they replayed 432 Ethereum mainnet transactions, and the proposed changes resolved 64 out of 67 cases.
The gas limit has been increased by 4%. Additionally, 254 other estimated results were unaffected, as these cases did not exhibit the aforementioned issue; three cases were not detected.
For retail investors, this proposed change should improve the reliability of wallet estimates and reduce instances where transactions are completed without achieving the expected outcome, even though some cases may involve higher gas limits. Fortunately, there is currently no evidence of a proven link between higher gas limits and higher fees.
