Base has explained why its mainnet stopped producing blocks twice in two days.
Summary
- Base’s latest postmortem shows one sequencer bug caused two mainnet halts within two straight days.
- Funds stayed safe, but transaction queues overflowed as Base stopped producing new L2 blocks temporarily.
- The team plans stronger fuzz tests, load tests, monitoring, and recovery tools after the outage.
The Coinbase-backed Ethereum layer-2 network said both outages came from the same bug in its sequencer block-building logic.
The first outage began on June 25 and lasted about 116 minutes. The second began on June 26 and lasted about 20 minutes. Base said funds stayed safe during both incidents.
Sequencer bug stopped block production
In its official postmortem, Base said an invalid transaction failed during execution, as expected. The issue came after that failure, when stale journal state remained inside the block builder.
That stale state included accounts and storage slots touched by the failed transaction. When a valid transaction came next, the system used the wrong journal state and charged gas incorrectly.
This created a block with an invalid state transition. Other nodes could not accept the block, so the chain stopped producing new L2 blocks.
“The integrity of the chain was not compromised and all funds on Base were safe,” Base said.
The team added that block production resumed safely after mitigation.
Transactions queued during the halt
During the outages, users could not get new transactions included onchain. Base said transactions queued in the mempool while the chain waited for block production to recover.
The transaction pool later grew beyond what it could store. As a result, new eth_sendRawTransaction requests returned errors during the outage window.
The halt also affected sequencer and validator progress. Base said these nodes could not move beyond the invalid block until sequencing returned.
As previously reported, Base first flagged unhealthy block production on June 25 before engineers isolated a consensus problem tied to an invalid block.
Patch fixed stale state issue
Base said it fixed the main bug by applying a sequencer patch. The patch ensures journal state updates properly during execution after a failed transaction.
The team also found a second issue during recovery. Base said mitigation took longer because a race condition in the engine reset feature stopped sequencers from catching up after restart.
That second issue helped explain why the incident returned the next day. Base said the problem affected sequencers, not validator nodes, but it still slowed recovery.
The Base status page showed sequencing resumed on June 25. It also told ecosystem node operators to restart Base nodes if they were still stuck.
Testing and recovery changes planned
Base said it will strengthen protocol fuzz testing and load testing. These methods help teams find strange transaction patterns that may expose hidden bugs.
The team also plans better monitoring and operational checks. It said these changes should help engineers detect similar problems earlier and respond faster.
Base also wants to add graceful recovery to base-consensus. That change would make it easier for validator nodes to continue syncing after similar failures.
The outage came during a busy week for the network. Base also moved forward with its Beryl upgrade, which adds the B20 token standard and cuts the standard Base-to-Ethereum withdrawal period from seven days to five days.
The incident gives developers and users a clearer view of the weak point. Base has now named the bug, released a patch, and listed the tests it plans to improve.
