What caused ETH client outage?

TLDR

An Ethereum (ETH) outage was caused by a bug in the Prysm consensus client (v7.0.0) that generated outdated states when handling old attestations, knocking many Prysm validators offline and briefly threatening finality per a Cointelegraph report.

1. Participation dropped about 2325%, matching Prysms share of validators, which brought the network close to the finality threshold per a media update.
2. Developers advised a temporary workaround flag, and client diversity kept the chain operating despite Prysm issues per a technical recap.
3. The incident occurred shortly after the Fusaka upgrade, underscoring risks from single-client concentration per a news summary.

Deep Dive

1. Prysm Bug Mechanics

The root cause was a Prysm consensus client flaw that unnecessarily created old states while processing outdated attestations, overloading compute and memory and degrading Prysm nodes. Prysm developers explained the pattern and suggested a temporary workaround using --disable-last-epoch-targets per a client-focused analysis.

• The bug created a denial-of-service-like condition via parallel historical state replays and stale attestations per a technical recap.
• Only Prysm users were affected; other consensus clients continued validating normally per a media summary.

2. Impact and Finality Risk

Validator voting participation fell to roughly 7475%, near the two-thirds threshold where Ethereum loses finality. About 2325% of validators went offline, largely mapping to Prysms footprint per a network impact report.

• Finality matters because, below two-thirds participation, blocks can be produced but are not finalized, raising reorg risk and prompting L2 bridges and exchanges to tighten operations per a risk overview.
• Client diversity kept the chain operating while Prysm validators implemented the workaround, with other clients like Lighthouse and Teku maintaining validation per a technical recap.

3. Timing and Client Concentration

The incident came shortly after the Fusaka upgrade, highlighting the importance of client diversity. Prysms share dropped after the event, but concentration remains a risk if any single client exceeds one-third of the network per a post-incident review.

• Reports noted ~23% of the network offline early during the incident, consistent with Prysms pre-bug share per a news summary.
• Consensus held due to multiple healthy clients, and participation rebounded after operators applied guidance per a technical recap.

Conclusion

In short, a Prysm consensus client bug caused a surge in stale attestations and heavy historical state generation, leading many Prysm validators to drop and pushing Ethereum close to a finality loss. The chain stayed operational thanks to client diversity, and participation recovered after the workaround. The key takeaway is operational resilience improves when no single client dominates, especially around major upgrades.