DOCS / GUIDES

Candling your data

Add data-quality checks that catch cracks before they reach the dashboard.

Intermediate10 min read
Before you beginYou need the cluck CLI, a Pollosoft account, and a coop to work in. No prior experience required. Curiosity and a mild fondness for chickens are helpful.

Overview

Add data-quality checks that catch cracks before they reach the dashboard. The whole thing takes about 10 minutes, and the commands below are the ones you will use along the way.

$ cluck candle init
$ cluck candle run --suite orders
$ cluck candle report

Steps

  1. Define checks. Write checks in YolkML: not-null columns, unique keys, accepted values, and freshness rules.
  2. Run in CI. Add cluck candle run to your pipeline. A failing check blocks the deploy and shows the offending rows.
  3. Triage failures. The report groups failures by table and shows sample rows, so you know whether it is a crack or just a speckle.
  4. Alert on trends. Candle tracks failure rates over time and alerts when a table slowly gets worse.

Commands used

Keep going

Stuck? Check the error codes, the FAQ, or the glossary.