Starknet announces the cause of the mainnet incident; the network has resumed stable operation.

The Starknet development team, the layer-2 (L2) scaling network on Ethereum, has published a post-mortem report clarifying the root cause of the mainnet outage on Monday.

According to the report, the issue originated from inconsistent network states between the blockifier (execution layer) and the proving layer (layer used to verify that the execution layer processes transactions correctly). The Starknet team explained that in a specific combination of cross-function calls, state write operations, reverts (revert), and error handling mechanisms, the blockifier had “remembered” a state change occurring inside a function that was reverted, leading to inconsistent transaction execution.

However, this error never reached a completed state on L1 thanks to Starknet’s proving layer. The proof layer detected the fault and prevented erroneous transactions from being recorded in the ledger, demonstrating that the verification mechanism still functions as designed.

Diagram illustrating how code errors affect the network | Source: StarknetThe incident forced the network to perform a block reorganization (block reorganization), resulting in approximately 18 minutes of network activity being rolled back. The development team stated that Starknet has now resumed normal operation.

Following the incident, the team committed to enhancing testing and code audits to prevent similar errors in the future. This disruption also highlights the increasing challenges in developing new generations of blockchains with multi-layered architecture and high complexity.

This is not the first time Starknet has experienced issues in 2025. Previously, the network faced several incidents, the most serious of which occurred in September after deploying a major protocol upgrade called Grinta.

Starknet operational time, with the red square indicating the power outage in September | Source: StarknetAccording to the post-mortem report, the incident lasted over five hours and was caused by a sequencer error — the component responsible for ordering transactions on the network. During the outage, block production was completely halted, and two chain reorganization events were carried out to restore operational status.

The reorganization process caused about an hour of network activity to be rolled back, forcing users to resend transactions. For casual users, resubmitting transactions may cause only minor inconvenience if not time-sensitive. However, for frequent traders or investors needing to exit positions or execute trades within short timeframes, this could lead to serious consequences.

Vương Tiễn