Ethereum 
$2,010 developers faced significant challenges during the Pectra upgrade on the Sepolia test network. A coding error in the deposit contract triggered the wrong type of transaction, leading to the production of empty blocks. The situation became more complex when an account executing zero token transfers, compliant with the ERC-20 standard, exploited the system. The team managed to stabilize the network with a rapidly implemented special filtering mechanism.
Source of the Technical Error and Its Impacts
Launched on the morning of March 5, the Pectra update unexpectedly revealed an error due to the deposit contract triggering a “deposit” event instead of a transfer. This inconsistency caused discrepancies in transaction pools and resulted in nodes producing empty blocks. Developer Marius van der Wijden noted that initial investigations suggested the error stemmed from reliable validators, but the actual source was a new account.
As the system continued processing faulty transactions, synchronization between nodes deteriorated. The increase in empty blocks negatively affected transaction confirmation times. The team recalled that similar issues had been experienced previously on the Holesky test network, announcing a review of testing processes before future updates.
Exploitation Triggered by Zero Token Transfers
The complexity of the error was exacerbated by the ERC-20 standard’s allowance for zero token transfers. One user exploited this vulnerability by sending numerous transactions that interacted with the deposit contract. Developers identified that the attacker likely used an account funded by faucets on the test network. Marius van der Wijden suggested that the attacker might have monitored communication channels.
To resolve the issue, the team released a fix that automatically filtered transactions interacting with the deposit contract. The majority of nodes were updated by 14:00, restoring normal network operations. The Ethereum team emphasized that they would tighten testing processes to prevent similar issues on the main network.
