Test severity levels in dbt - Step-by-Step Execution

Concept Flow - Test severity levels

Define test in dbt

↓

Assign severity level

↓

Run dbt tests

↓

Test result: Pass or Fail

↓

If Fail -> Check severity

↓

Fail build

↓

Report results

This flow shows how dbt tests are defined with severity levels, run, and how failures are handled differently based on severity.

Execution Sample

dbt

version: 2
models:
  - name: customers
    tests:
      - unique:
          column_name: id
        config:
          severity: error
      - not_null:
          column_name: id
        config:
          severity: warn

Defines two tests on the customers model with different severity levels: error and warn.

Execution Table

Step	Test Name	Severity	Test Result	Action Taken	Build Status
1	unique	error	pass	No action	continue
2	not_null	warn	fail	Log warning	continue
3	unique	error	fail	Fail build	stop
4	not_null	warn	pass	No action	stop
5	unique	error	pass	No action	stop
6	not_null	warn	fail	Log warning	stop

💡 Build stops immediately if any test with severity 'error' fails; warnings do not stop the build.

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	After Step 5	After Step 6
Build Status	continue	continue	continue	stop	stop	stop	stop
Warnings Logged	0	0	1	1	1	1	2

Key Moments - 2 Insights

Why does the build stop at step 3 but not at step 2 even though both tests fail?

Because step 3 fails a test with severity 'error' which stops the build immediately, while step 2 fails a 'warn' severity test that only logs a warning and continues. See execution_table rows 2 and 3.

What happens if multiple warn severity tests fail?

All warn failures log warnings but do not stop the build. The build continues running all tests. See execution_table rows 2 and 6 and variable_tracker for warnings count.

Visual Quiz - 3 Questions

Look at the execution table, what is the build status after step 2?

A stop

B continue

C paused

D unknown

💡 Hint

Check the 'Build Status' column at step 2 in the execution_table.

At which step does the build stop due to a test failure?

A Step 3

B Step 2

C Step 6

D Step 1

💡 Hint

Look for the first 'stop' in the 'Build Status' column in the execution_table.

If the severity of the unique test was changed from 'error' to 'warn', what would happen at step 3?

A Build would stop

B Test would pass automatically

C Build would continue with a warning

D Test would be skipped

💡 Hint

Refer to how 'warn' severity tests behave in the execution_table and key_moments.

Concept Snapshot

dbt test severity levels:
- severity: error stops the build on failure
- severity: warn logs a warning but continues
- define severity in test config
- errors are critical, warnings are advisory
- build status depends on test severity results

Full Transcript

In dbt, tests can have severity levels like 'error' or 'warn'. When tests run, if an 'error' severity test fails, the build stops immediately. If a 'warn' severity test fails, it logs a warning but the build continues. This allows teams to prioritize critical issues while still being aware of less critical ones. The execution table shows test results step-by-step, tracking build status and warnings logged. Understanding severity helps manage test failures effectively.