Dry Run¶
Reviewing dry run results
A Dry Run evaluates one check against a sample of its data and shows what it would flag, right away and without side effects. It creates no anomalies and runs no Scan, so it is the safe way to test a new or edited check, or a Draft, before it runs for real. To start one, see Dry Run a Check.
What a Dry Run Evaluates¶
- A sample, not the whole table. The Dry Run reads up to 10K records by default, shown as the Sampling Limit in the results.
- The check as saved. It uses the check's current filter clause, properties, and coverage.
- No anomalies. The results stay in the dialog. No anomaly is created and no Scan runs.
The Results¶
The Dry Run Results dialog opens with a summary:
| Item | What it shows |
|---|---|
| Dry Run | Whether the Dry Run completed. |
| Timing | How long it took. |
| Sampling Limit | How many records it read, 10K by default. |
| Check | The check's number, rule type, and Description. |
| Table or File | The table or file the check runs on. |
| Field or Fields | The fields the check applies to. |
While it runs, the dialog shows its progress: preparing the data, asserting the check, and scanning for anomalous data.
Anomalies¶
The Anomalies section lists what the check would flag:
| Item | What it shows |
|---|---|
| Description | Why the records failed, in the same words an anomaly would use. |
| Asserted Records | How many records were evaluated. |
| Anomalous Records | How many of them failed the check. |
When nothing fails, the section reads No anomalies were identified, with the number of asserted records.
Source Records¶
The Source Records section shows the records that failed, laid out like an anomaly's Source Records:
- Computed fields and computed assets. The values of computed fields are calculated on the fly, as in Data Preview. For computed tables, files, and joins, every output column is shown.
- Masked fields. Values of masked fields are hidden by default. Users with the Editor team permission can turn on Show masked values, and every reveal is recorded in the masking audit log.
- Sort. The Sort By menu orders the columns by Original (the container's own order, the default for computed tables, files, and joins), Name, Importance (the default for tables, views, and files), or Quality Score. Your choice is remembered.
- Download. Download source records saves the records as a CSV file, with the columns in the order shown.
Source records cover only the anomalous rows found in the sample. To browse a table's data independently of a check, use Data Preview.
Logs¶
When the Dry Run cannot evaluate the check, the Logs section shows the message explaining why.
See Also¶
-
Check Views
The check lists, the check page, and the side panel: how to reach each one, what it shows, and what you can do there.
-
Check Anomalies
The anomalies a check has produced, and how to triage them from the check's page.
-
Check Timeline
The change log of a check, its past versions, and the comments your team leaves on it.
-
Check Examples
Scenarios that show ownership and AI Managed checks in practice.
-
Check Permissions
Which team permissions can view, create, edit, archive, and delete checks, change owners, and edit AI Managed checks.
-
How a Check Works
What a check is made of, who creates it, how AI Managed checks are generated, how a Scan evaluates it, and how its status changes.
-
Authored Checks
What an Authored check is, who governs it, and how it differs from an AI Managed check.
-
AI Managed Checks
What an AI Managed check is, which edits convert it to Authored, and how to go back.
-
Check Ownership
Who is responsible for a check, the four ways ownership changes, notifications, and filtering by owner.
-
Check Importance
How much a check's failures matter, and how its tags raise or lower it.
-
Impact Metrics
Numbers measured over a check's failed records that show how big a failure is.