Event Overview: Two Brief Outages
Base has released a postmortem report detailing two block production outages that occurred on its mainnet on June 25 and June 26, 2026. The first outage lasted approximately 116 minutes, while the second lasted about 20 minutes. The official team emphasized that the events did not impact on-chain asset security, and user funds remained accessible at all times.
Root Cause: Sequencer Journal State Not Properly Cleaned
The core issue was a bug in the sequencer's block-building logic. After a failed transaction execution, the system failed to clean up the historical journal state. This caused subsequent legitimate transactions to experience incorrect gas calculations during execution, resulting in invalid state transition blocks. Consequently, the entire L2 network ceased block production. In essence, dirty state residue corrupted the gas estimation for subsequent transactions, making it impossible for the sequencer to generate valid blocks.
Impact: Chain Stoppage, Transaction Failures, Mempool Congestion
During the outage, the network experienced multiple anomalies: complete block production halt, users unable to submit transactions, and persistent mempool congestion. Requests through eth_sendRawTransaction continuously returned errors. These issues were not due to user-side mistakes but stemmed from an underlying state machine fault causing chain-wide unresponsiveness.
Fix and Second Outage: Patch PR #3806 and Race Condition
The team quickly identified the problem and deployed a patch (PR #3806), which restored block production. However, during the sequencer cluster restart, a race condition in engine reset disrupted the recovery synchronization. This race condition is considered the indirect cause of the second brief outage on June 26, indicating that the initial fix did not fully address synchronization consistency during the restart phase.
Next Steps: Enhanced Testing and Graceful Recovery
Base stated it will strengthen protocol-level fuzz testing and stress testing capabilities to detect abnormal transaction paths earlier and prevent similar state residue issues. The team also plans to upgrade monitoring and operations systems and introduce a more graceful recovery mechanism, aiming to improve the network's ability to recover quickly from similar failures in the future, reducing downtime and scope of impact.