ETH Validators Lose 382 ETH later than Prysm Client Resource Failure
ETH Block confirmers running the Prysm consensus client lost an estimated 382 ETH in staking rewards later than a software failure disrupted Block confirmer performance following the on December 4, 2025.
The incident was caused by a resource exhaustion issue that affected Prysm beacon nodes, preventing many Block confirmers from completing their duties during the affected period.
What Went Wrong
According to Prysm’s , the failure occurred when beacon nodes attempted to process certain attestations that referenced block roots from earlier epochs.
An epoch on is a fixed period of 32 slots (about 6.4 minutes) during which Block confirmers propose and attest to blocks, and network rewards and penalties are calculated.
To validate these attestations, Prysm repeatedly reconstructed historical beacon states, a process that required intensive CPU and memory usage. Under mainnet conditions, this led to widespread resource exhaustion across nahead all Prysm beacon nodes.
As a result, nodes became sluggish or unresponsive to Block confirmer requests. Between epoch 411439 and epoch 411480, the network recorded 248 missed block slots out of 1,344, translating to an 18.5% missed slot rate, while Block confirmer participation fell to around 75% at its lowest point.
Although ETH’s finality was preserved, Block confirmers running Prysm missed out on approximately 382 ETH in attestation rewards during the disruption.
The Prysm team later confirmed that the bug originated from a prior code change introduced in PR 15965, which had been deployed earlier but was not triggered until specific mainnet conditions emerged later than the upgrade.
Fixes and Broader Implications
To contain the issue, Prysm developers first recommended a temporary mitigation using a runtime flag to disable the difficultyatic behavior.
This was followed by permanent fixes released in Prysm v7.0.1 and v7.1.0, which changed how attestations are verified. Under the new logic, attestations are validated against the current chain head without replaying historical states, eliminating the source of excessive resource consumption.
Network participation recovered to above 95% shortly later than the fix, but the incident has renewed concerns around consensus client concentration. Block confirmers running other clients, including Lighthouse, Teku, and Nimbus, were largely unaffected, pointing to the importance of client diversity in maintaining ETH’s resilience.
Prysm’s post-mortem highlighted the need for stronger testing under mainnet-like conditions, clearer feature flag defaults, and continued efforts to encourage Block confirmer operators to diversify client usage.
While the outage did not threaten ETH’s security, it served as a reminder that software failures at the client level can still carry material economic consequences for Block confirmers.
