Smoke test

A smoke test is the smallest, fastest performance test you can run. It only confirms that the system is up, the target URL is reachable, and basic request flows return successful responses. If the smoke test fails, nothing else is worth running yet.

Before you start

• You have a MaxoPerf account and a project. See Run your first test.
• You know the URL or endpoint you want to verify.
• The target environment is already deployed (staging or dev is fine).

What is a smoke test?

A smoke test uses a handful of virtual users (typically 1 to 5) for 30 to 60 seconds. It does not measure performance. It checks that the test setup is correct and the target responds, so you do not fire a high-VU test at a broken system.

Run a smoke test every time you:

• Set up a new test for the first time.
• Change the test file or target URL.
• Deploy a new version of the system before a load test.

Not for: measuring throughput, finding limits, or stress-testing.

How to run a smoke test in MaxoPerf

Load profile

Parameter	Value
Virtual users (VUs)	1–5
Duration	30–60 s
Ramp-up	None (start at full VU count instantly)
Stop mode	Duration
Locations	1 (nearest to your target)

Console walk-through

1. Open your project and click New test (or select an existing test to edit).

2. Give the test a clear name, for example checkout-smoke, so it is easy to spot in the runs list.

3. Upload your Taurus YAML, k6 script, or JMeter JMX. For a simple smoke test, a minimal Taurus file is enough:

   execution:
     - executor: jmeter
       concurrency: 2
       hold-for: 30s
       scenario: main
   
   scenarios:
     main:
       requests:
         - url: https://your-staging.example.com/health
           label: health-check

4. In the Load profile section, set Virtual users to 2 and Duration to 30s. Leave ramp-up at 0s.

5. Select one managed cloud location close to your target (for example, us-east-1).

6. Click Run. The run moves from queued → allocating → running in seconds.

How to read the result

A passing smoke test shows:

• Status: finished (not failed or cancelled).
• Error rate: 0 % or very close to zero.
• Throughput: any positive value. You only need proof that requests landed.
• p95 latency: not the focus yet, but note it for later load-test comparison.

If the error rate is above 0 %, stop. Fix the target or the test file before you move on to a load test.

Do / don’t

Do	Don’t
Run a smoke test before every load or stress test	Use smoke-test results to make performance claims
Use it to validate changed test files	Add ramp-up; this adds complexity you don’t need
Keep it short (≤ 60 s) so it fits in CI pre-check	Run from multiple locations when one is enough
Run against staging, not production	Run more than 5 VUs. It’s a sanity check, not a load test

Tip

Add a smoke test to your CI pipeline as a pre-gate before the full load test job. If the smoke test fails, the pipeline skips the load test step and wastes no capacity.

Where to go next

• Load test: once the smoke test passes, move on to a real load test.
• Foundations: Core metrics explained: what latency and error rate mean, before you measure them.
• Cookbook: CI-gated performance test: add a smoke gate to your pipeline.