dc_003 · Deployment: empty batch (schedule fires, nothing new)
Status: ✓ Passing · Last run: 2026-06-09T06:20:40 · Pipeline: deployment
· History: 1/1 runs passed (100%)
What was tested
Scheduled fire activates but no new rows have arrived since the last run. Deployment should treat this as a clean no-op — write nothing, log "0 rows scored", and leave the prior predictions table alone. NOT raise an exception; NOT clobber the output table with empty rows.
What we planted in the data
- 1,000 entities
- Signal strength: 0.50 (sigmoid slope multiplier)
n_prior=500
What the system did
- Training rows built: 1,000
- Features used in model: 3
- Model selected:
DeploymentScoringRunner
What we checked — all assertions passed
| Status | Assertion | Detail |
|---|---|---|
| ✓ | deployment_completed_cleanly |
clean run |
| ✓ | scored_row_count_eq |
scored 0 rows, expected 0 |
| ✓ | prediction_distribution_in_band |
empty batch — no predictions to evaluate (clean no-op) |
| ✓ | prior_predictions_unchanged |
prior predictions row count unchanged (500) |
How this could have gone wrong
(no assertion descriptions provided)
Why this case matters
Quiet days happen: weekends, holidays, transient upstream pauses. An empty batch is operationally normal, not a failure mode. If the deployment halts on empty batches, on-call gets paged constantly.
Reproducing
# from auto_insight_api/
python -m validation.v2 run dc_003 --pipeline deployment -v
- Case config:
validation/v2/cases/dc_003_empty_batch.yaml - Data shape:
deployment_empty_batch - Analytics type:
deployment