Smoke Test

A smoke test answers the cheapest, most important question in software: is this build alive at all? A handful of critical-path checks run first, before anything elaborate — if the system can't log in and load its main screen, everything else is postponed. Named from electronics: power on, see if anything literally smokes.

What It Is

A smoke test (or build verification test) is a minimal, fast suite confirming that a build's most fundamental functions work — it starts, authenticates, serves its core pages or endpoints, and connects to its database. It runs at the earliest opportunity (on every build, on every deployment, including to production after release as a deployment verification) and its verdict is binary: proceed or stop. Its economics: thirty seconds of smoke testing prevents hours of elaborate testing performed on a build that was dead on arrival — and, in the production case, catches a catastrophically bad release before users do.

Why It Matters to You

Smoke tests are the tripwire layer of the quality system, and their absence is visible in one signature pattern: deployments that fail and stay failed for an hour before anyone notices. Modern operations run post-deployment smoke continuously or on a schedule — the system verifying its own pulse — which converts many 'outage discovered by users' incidents into 'outage discovered in 90 seconds and rolled back'. In vendor reviews, ask what runs automatically after each of their deployments; a confident answer with specifics is the marker of an operation that knows its own systems.

When You'll Meet It

In deployment pipeline reviews and monitoring design. The question 'what runs after a release goes out?' distinguishes operations that find their own failures from those that wait for the helpdesk to.