Smoke testing

Definition

A smoke test answers an operationally simple question: is this build or deployment usable enough to justify deeper testing? The suite is intentionally narrow. It may check that the application starts, a health endpoint responds, a user can authenticate, and one essential operation completes.

When to run smoke tests

Smoke tests are most useful immediately after a build, deployment, or environment change. They can stop an obviously unusable artifact from moving to a longer test stage or from receiving broader manual attention. Because the suite runs early, its checks should be fast, deterministic, and easy to diagnose.

The exact scope depends on the product. A background worker may need to consume a representative message, while a web application may need a basic page load and authenticated request. Document the boundary so the team understands what a passing smoke result means.

Example

After deploying an order service, a smoke suite can verify that the service starts, its health endpoint reports ready, a test account can authenticate, and a small order can be created in an isolated environment. If the service cannot connect to its database, the suite stops before slower scenario tests begin.

These checks do not need to exercise every discount rule, error response, or permission combination. Those belong in more complete test layers.

Limitations

Smoke testing is a build or deployment gate, not a complete quality assessment. A green smoke suite can miss subtle defects, edge cases, performance degradation, and broken behavior outside the selected paths. Keep it small enough to provide early feedback, then rely on regression, integration, system, and exploratory tests for broader confidence.