A critical software bug in the Prysm consensus client led to significant network disruptions and financial losses for Ethereum validators shortly after the Fusaka mainnet upgrade. The incident, detailed in a post-mortem report titled "Fusaka Mainnet Prysm incident," occurred on December 4, 2025, and resulted from a resource exhaustion event that affected nearly all Prysm nodes.
The bug, which was introduced and deployed to testnets approximately a month earlier, was triggered on the mainnet following the Fusaka upgrade. It caused delays in validator requests, leading to missed blocks and attestations across the network. Specifically, Prysm beacon nodes received attestations from potentially out-of-sync nodes that referenced a block root from a previous epoch. To validate these attestations, nodes were forced to replay old beacon states, triggering thousands of costly recomputations that exhausted system resources.
The disruption spanned 41 epochs (from epoch 411439 through 411480), during which the network missed 248 blocks out of 1,344 available slots. This represented an 18.5% missed slot rate and pushed overall network participation down to 75% at the peak of the incident. The direct financial impact was substantial: validators missed out on approximately 382 ETH in attestation rewards, equivalent to over $1 million at the time.
The Prysm team, developed by Offchain Labs, implemented immediate and permanent fixes. A temporary mitigation was achieved by advising users to enable the --disable-last-epoch-target flag in version 7.0.0. Subsequently, permanent changes were introduced in versions 7.0.1 and 7.1.0, which updated the attestation validation logic to rely on the head state and avoid historical state replays, preventing a recurrence.
The outage has intensified scrutiny around Ethereum's client concentration and the systemic risks of software monocultures. Data from Miga Labs shows Lighthouse as the dominant consensus client with 51.39% of validators, followed by Prysm at 19.06%, Teku at 13.71%, and Nimbus at 9.25%. Offchain Labs warned that a bug in a client with more than one-third of the network could cause a temporary loss of finality, while a client with over two-thirds could finalize an invalid chain. This incident has renewed calls from developers and ecosystem participants for validators to switch to alternative clients to improve network resilience and reduce single-point-of-failure risks.