The team behind the Ethereum layer-2 solution Base has published a detailed report on two outages that occurred on June 25 and 26. Both incidents led to a complete halt in block production, and it turned out that the root cause was a bug in the sequencer logic.
The first outage lasted 116 minutes, the second — 20 minutes. During this time, the network stopped generating new blocks, transactions were not confirmed, and the queue of unconfirmed operations became overloaded, causing errors for users attempting to send new transactions. Importantly, user funds remained safe and were not affected.
What went wrong
According to the report, the issue began with a transaction that ended in an error. The sequencer did not clear the temporary state (journal state), causing the next valid transaction to receive an incorrect fee calculation. This led to the creation of a block with an invalid state transition, which other network nodes could not accept. As a result, block production completely stopped.
After applying a fix, developers encountered a second error: a race condition during the restart of the sequencer cluster prevented nodes from synchronizing quickly. This caused the second, shorter outage the following day.
What measures Base will take
Following the incident, the Base team pledged to enhance stress testing and fuzzing of the protocol to identify non-standard transaction processing scenarios in advance. Additionally, developers plan to improve network monitoring and implement a "graceful recovery" mechanism to restore network operation more quickly after such failures.
Recall that in February, the Base team announced a transition from the Optimism technology stack to its own unified network architecture.
Analyst comment: Both incidents highlight the critical importance of thorough edge-case testing in L2 solutions. Although user funds were not affected, such outages undermine trust in the network and demonstrate that even leading protocols are not immune to errors in sequencer logic. Strengthening fuzz testing is a correct step, but for full resilience, Base needs to implement backup mechanisms capable of instantly isolating problematic transactions without halting the entire network.