Reproduce missing, 0 and substituted telemetry after a refrigeration incident
Separate unobserved, measured and manually substituted values before reusing post-recovery telemetry.
Restored access is not restored observation
Restored access to equipment does not establish restored meaning of its telemetry. Measured values, collection gaps and manual substitutions need separate review. Truthound can check that data contract, not perform security recovery or determine equipment safety.
What the joint statement establishes
The November 4, 2026 joint statement hosted by ENISA describes EU commercial-refrigeration cyber incidents since October 11, 2026 and a further November 1–2, 2026 wave. It does not establish data corruption. The proposed fixture is a separate preparation exercise, not a claim about observed damaged data.
Create a separate fixture
Keep test inputs distinct from actual measurements. Within one sensor-and-period definition, include a measured value, an uncollected value, NaN and a manually entered value. Equipment owners define which states require a hold. Do not present fixture outcomes as results observed at an affected site.
Do not replace 0 with missingness
A measured 0 is not a missing reading. Nor should every variable be forced into a positive range: negative temperatures may be legitimate. Establish units and meaning, then retain observation state separately so collection convenience cannot redefine the measurement.
Two different gaps in Truthound
Truthound Core Python SDK 3.3.4, operating on a prepared Polars table, provides CompletenessRatioValidator for a supplied minimum non-null ratio and NotNaNValidator for floating-point NaN. Null, NaN and a physically meaningful value range are different questions. Operators establish expected sensors, periods and acceptable coverage first.
Compare the existing predicate
An existing Pandera 0.35.0 pandas predicate may remain. Compare whether both paths distinguish missingness from 0 and manual substitutes from original observations. General validation capability does not establish equivalent post-incident processing.
Preserve the interval and substitution reason
A populated value can still be a repeated last observation. Preserve time coverage. For a manual substitute, retain the affected interval, reason and reviewer rather than presenting it as remote measurement. Any interpolation procedure needs its own approved scope.
Route causes separately
Route an access problem to the security owner, a transmission gap to collection operations and measurement interpretation to equipment specialists. A failed data check is not automatically evidence of another attack. A passing check is not operating-safety approval either.
Conditions for using the readings again
Reuse readings when observed and substituted intervals are explainable. Keep unresolved gaps visible and identify affected analyses. Treat restored access and restored observation as different artifacts before either informs later decisions.
Conditions
Use approved post-response data or clearly identified illustrative inputs within authorised access.
Alternatives
Compare an existing Pandera 0.35.0 pandas predicate on the same observation-state examples.
Limitations
The source does not establish data corruption; checks do not prevent attacks, patch vulnerabilities or decide temperature safety.
Evidence
- Event checked
- Subject checked
- Modified
