Base sequencer bug caused back-to-back outages, post-mortem reveals
In brief
- Base sequencer bug halted block production for 116 minutes Thursday, then 20 minutes Friday.
- Stale journal state persisted after invalid transaction failures, preventing sequencer cache clearing.
- Race condition after system reset prevented sequencers from catching up to valid chain state.
- Base's single sequencer creates centralized point of failure for $11 billion layer-2 network.
- Engineering team applied patches and plans improved fuzz testing and validator recovery.
The Bug
The Base engineering team identified a bug in sequencer block-building logic that allowed stale journal state to persist after a transaction validation failure. An invalid transaction was received by the block builder and failed during execution, as expected, but erroneously did not clear the journal state that contained the accounts and storage slots that had been accessed.
The journal state is a cached record of data touched during block execution. When an invalid transaction failed, this cache should have been wiped clean. Instead, it remained in memory, contaminating subsequent block-building attempts and freezing the network.
Race Condition and Recovery
There was also a "race condition" after the system reset, which prevented the sequencers from catching up, causing the second outage. Mitigation took longer than expected due to infrastructure conditions unrelated to the original bug. The team fixed the outages by applying a patch to the sequencers to ensure the journal state was properly updated during execution.
Structural Risk
The Base layer-2 network runs a single sequencer, which means one bug can stop everything. It is a centralized blockchain component. Base is the second-largest layer-2 network by total value secured, with just under $11 billion. The single-sequencer architecture means outages carry outsized risk for the ecosystem.
This wasn't the first sequencer failure. Base experienced a sequencer-related outage that stopped block production for 17 minutes in September 2024, and for around half an hour in August 2025.
Next Steps
The Base engineering team plans to improve protocol fuzz testing, which involves bombarding the system with large volumes of random or malformed inputs to catch edge cases. They also plan to build graceful recovery mechanisms for validator nodes so future incidents degrade more gracefully instead of halting production entirely.
"There was also a "race condition" after the system reset, which prevented the sequencers from catching up, causing the second outage."
— Base engineering team
