Outage Overview: Two Incidents, Impact and Duration
According to Base's official postmortem, the Base mainnet suffered two block production outages on June 25 and 26, 2026. The first outage lasted approximately 116 minutes, the second about 20 minutes. The team confirmed that user assets and on-chain data remained safe and accessible at all times, with no funds at risk.
Root Cause: Sequencer State Management Flaw
The core issue was a defect in the sequencer's block construction logic. After a transaction execution failure, the system failed to properly clean up historical journal state. This caused subsequent valid transactions to encounter gas computation anomalies, generating invalid state transition blocks and halting the entire L2 network. During the outage, users experienced complete block production cessation, inability to submit transactions, and mempool congestion, with eth_sendRawTransaction calls continuously returning errors.
Fix and Future Improvements
Base deployed a fix via PR #3806, cleaning up the journal state issue and resuming block production. However, a race condition during the sequencer cluster restart hindered synchronization, indirectly contributing to the second outage the following day. Going forward, Base plans to strengthen protocol-level fuzz testing and stress testing to detect abnormal transaction paths earlier. It will also upgrade monitoring and operations tooling and introduce a more graceful recovery mechanism to improve resilience against similar failures.