Hydra targets a large-head close failure under a short contestation window.

What happened.

Hydra’s plateau-1k benchmark, configured with a 1,000-UTxO head, 4,000 transactions, and a one-second contestation period, deterministically failed to close with OutsideValidityIntervalUTxO. The record attributes this to unevaluated snapshot-UTxO thunks being forced late, while the close transaction’s validity window was already running.

The cited commit moves that work earlier by forcing the UTxO when the snapshot is requested rather than when the close transaction is built. In the stated benchmark, that change is intended to let the large head close within the validity window again.

Why it matters.

For operators running large Hydra heads with short contestation periods, a close that cannot complete cleanly is a liveness and settlement concern. This change addresses a specific, reproducible failure mode in that path, rather than a general claim about all Hydra configurations.

The broader feed describes related Hydra work on serialization and close handling, but this record’s direct evidence is limited to the close-path commit and its benchmark scenario.

What to watch next.

Watch for the follow-up tracked as issue #2831, which is intended to make UTxO value-strict, and for evidence that the change has progressed beyond a commit on master into a released fix. Operators may also want to compare their own head sizes and contestation periods with the tested setup.