Skip to main content

Run your first checkpoint

Goal​

You'll connect a Postgres datasource, pick a table, author one rule, and run the rule against the table. End state: a green validation result on the dashboard.

Prereqs​

  • A running PLACEHOLDER Cloud (see Self-host).
  • A Postgres database with at least one table you can read from. The bundled seed_dev script does not create one — point at any other DB you have access to.
  • The Postgres user's connection string. The user needs SELECT on the target table and SELECT on pg_catalog (every Postgres user has this by default).

Steps​

  1. Open the UI at http://localhost:3000 and log in.

  2. Add a datasource. Navigate to Datasources → New. Fill in:

    • Name — local-postgres (anything readable).
    • Type — postgres.
    • Connection — host, port, database, user, password.

    Click Test connection. You should see a green check with the round-trip latency. Postgres, Snowflake, BigQuery, Redshift, Databricks, MSSQL, and Trino are all live (FEATURE-1…FEATURE-6).

  3. Discover assets. On the datasource detail page, click Discover assets. PLACEHOLDER Cloud lists every table and view it can see. Tick the one you want to validate and click Add selected.

  4. Author a rule suite. Click into the new asset → Rule suites → New. Pick a rule from the gallery — for a first run, "Row count is between" with min = 1, max = 10_000_000 is a safe choice that always passes. The full gallery shipped in FEATURE-40; you can also drop down to a raw SQL rule via "Custom SQL" (FEATURE-9).

  5. Create a checkpoint. From the rule-suite detail page, click Create checkpoint. Give it a name and accept the defaults (the asset and the suite are pre-filled).

  6. Run the checkpoint. Click Run now — from either the Checkpoints list or a checkpoint's detail page. A confirmation banner appears with a link to the Results page, and a row shows up there within seconds. Click into it and confirm the rule turned green.

Verify​

You see one passing validation result on the dashboard. The result detail page lists the expectation that ran, with a pass count.

Next steps​