Great Expectations "Checkpoint ... failed" blocks CI

By Daniel Zoghalchali·Latchkey

A Great Expectations Checkpoint ran one or more ExpectationSuites plus its actions and returned success: false. The checkpoint is the unit you gate CI on, so a false result must fail the step.

What this error means

A great_expectations checkpoint run (or context.run_checkpoint) reports overall failure, but the workflow either continues or the message is buried in logs.

Checkpoint "nightly_orders" ran 2 validations.
  orders.warehouse: False
  orders.freshness: True
Checkpoint result: success=False

Common causes

A bundled suite failed validation

The checkpoint aggregates suites; if any one is unsuccessful the whole checkpoint is false. The per-suite lines show which one.

The CLI exit code is ignored

Older CLI invocations returned 0 regardless. Without checking the result explicitly, CI does not fail even when the checkpoint does.

How to fix it

Run the checkpoint and fail on false

1. Invoke the checkpoint from Python and inspect result["success"].
2. Raise SystemExit(1) when it is false.
3. On modern CLI, rely on the non-zero exit it now returns for a failed checkpoint.

- name: Data quality checkpoint
  run: |
    great_expectations checkpoint run nightly_orders
  # modern CLI exits non-zero on a failed checkpoint, failing the step

Read the per-suite breakdown

Open the printed per-validation lines or Data Docs to find which suite failed, then fix that data or expectation rather than the checkpoint wrapper.

How to prevent it

• Gate CI on the checkpoint result, not on the presence of logs.
• Keep one checkpoint per pipeline stage so failures are scoped.
• Upload Data Docs so the failing validation is one click away.

Frequently asked questions

Related guides

References

• GitHub Actions documentation