3806

Base’s mainnet recently suffered two shutdown incidents in quick succession. The complete post-incident review report just released by the official is worth paying attention to. The first interruption on June 25 lasted 116 minutes, and the next day saw another 20-minute secondary interruption—the core issue was a flaw in the sequencer’s block construction logic.

After a transaction execution failure, the system did not properly clean up the historical journal state, causing subsequent valid transactions to have abnormal gas calculations. This ultimately led to the generation of blocks containing invalid state transitions, which directly caused the entire L2 to stop producing blocks. During the incident, user transactions could not be added on-chain, and the mempool became severely congested.

The official emphasized: **users’ funds are completely safe**, and on-chain assets have never been affected. Although a fix has already been deployed via PR #3806 , the indirect cause of the next day’s secondary interruption was an engine race condition when the sequencer was restarted.

Going forward, Base will focus on strengthening protocol-level fuzz testing and load/stress testing, while also upgrading its monitoring and operations framework and introducing a “graceful recovery mechanism” to improve incident response capability.

#Base #Layer2