Coinbase Base L2 Experiences Two-Hour Outage Following Consensus Failure

Coinbase’s Ethereum layer-2 network, Base, was taken offline for approximately two hours on Thursday after an invalid block disrupted its sequencing process and triggered a consensus failure. The outage halted all block production on the mainnet, temporarily freezing transaction processing across the network.

The disruption began at 16:03 UTC on June 25, when the network’s official status page began reporting that mainnet block production had degraded to an unhealthy state. The Base engineering team publicly acknowledged the halt roughly 40 minutes later, confirming on X that “Base Mainnet is currently halted while the team works on an issue with block production” and stating that all user funds remained secure.

The engineering team identified the root cause shortly after the initial announcement. According to updates published on the incident log, a consensus fault allowed an invalid block to enter Base’s sequencing pipeline immediately after block 47806542. The malformed block prevented the sequencer from constructing valid subsequent blocks, effectively stalling chain progression until the team intervened.

Base operates with a single centralized sequencer managed by Coinbase. While this architecture prioritizes transaction throughput, it does not include an automatic failover mechanism for consensus errors. When the sequencer encountered the fault, network activity stalled completely until engineers isolated the invalid block and cleared the sequencing pipeline. Internal recovery was achieved around 17:21 UTC, but the team advised ecosystem node operators to restart and resync their infrastructure to properly propagate blocks across the network.

Two hours after the initial disruption, Base confirmed widespread recovery across its decentralized application and node ecosystem. The available incident report did not specify the exact technical trigger behind the invalid block, and the precise scope of the consensus fault remains under review by the network engineers. Block production and transaction processing have since normalized, with dependent services completing synchronization following the outage window.