Skip to main content

Batch tests

The --batch-tests flag combines compatible data tests that target the same model into a single query per test type, rather than running one query per test. This reduces the number of queries dbt runs for tests, which can improve performance in projects with many data tests.

Batching tests​

Batching applies to dbt build and dbt test. dbt batches a group of tests when all of these are true:

  • The tests are unique or not_null
  • They're attached to the same model
  • They use their default configuration
  • At least two of them are the same type

dbt excludes tests from batching when any of the following is set:

  • severity
  • error_if
  • warn_if
  • fail_calc
  • limit
  • where
  • store_failures
  • store_failures_as

Excluded tests run on their own, as they do without batching.

Setting the flag​

dbt resolves batch_tests from three places, in order of precedence:

  1. The --batch-tests command-line flag
  2. The DBT_ENGINE_BATCH_TESTS environment variable
  3. The flags block in your root project's dbt_project.yml

For example, on the command line:

CLI
dbt build --batch-tests

With an environment variable:

Env var
export DBT_ENGINE_BATCH_TESTS=true
dbt build

In your dbt_project.yml:

dbt_project.yml
flags:
  batch_tests: true

Was this page helpful?

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

0
Loading