k6 Load Export & Insights

Prerequisites

• An account with k6 export enabled (Admin → user settings → k6_export_enabled). When enabled, sidebar shows Load Insights.
• A project with API-oriented NLP tests (HTTP / OpenAPI / Postman flows work best).
• Grafana k6 installed locally, or Docker with grafana/k6.
• Network reachability from your runner to the API under test.

1. Export a k6 load kit from the app (recommended)

Best for a whole suite or selected cases — mirrors your Test Manager grouping.

2. Export a single test case (optional)

From a test case editor, use Export k6 load script when you only need one scenario.

• Downloads a .js script plus run-with-summary.sh and collect-results.py.
• Copy the env snippet (BASE_URL, VUS, DURATION, K6_LOAD_PROFILE).
• Review any skipped steps / warnings — UI clicks and pure browser steps don’t compile to protocol k6.

3. Run k6 locally (end-to-end)

1. Unzip the kit and open the folder in your IDE (or commit it to your repo).
2. Configure env:

   cp .env.example .env
   # edit .env — set BASE_URL to the API under test
   # optional: API_TOKEN / auth headers used by generated scripts

3. Make runners executable (Unix):

   chmod +x run-smoke.sh run-load.sh scripts/run-with-summary.sh 2>/dev/null || true

4. Execute:

   ./run-smoke.sh          # quick sanity load
   # or
   ./run-load.sh           # heavier profile from export options

   Without helper scripts you can still run:

   k6 run --summary-export=k6-out/my-run-summary.json scripts/…/my_load_test.js

   Docker alternative if k6 isn’t installed:

   docker run --rm -i -v "$PWD":/work -w /work \
     -e BASE_URL -e VUS -e DURATION \
     grafana/k6:latest run scripts/…/my_load_test.js

5. Collect results — kit helpers write summaries under k6-out/ and aggregate to k6-out/testneo-import.json when Python 3 is available.

4. Tweak endpoints, VUs, and duration

Before a serious run, edit env (or shell exports) — you don’t need to re-export from TestNeo for small knobs:

export BASE_URL=https://staging-api.example.com
export K6_LOAD_PROFILE=load    # or smoke | ramp | spike | soak
export VUS=50
export DURATION=5m
./run-load.sh

• Point BASE_URL at staging/prod as your policy allows.
• Change profile or VUs to match the SLO you care about.
• Auth secrets stay in .env / CI secrets — never commit them.
• Re-export from TestNeo when NLP steps / suite membership change.

5. Import results into Load Insights

1. In the app sidebar open Load Insights (/web/load-insights).
2. Select the same project you exported from.
3. Upload one of:
   • k6-out/testneo-import.json (preferred — multi-run aggregate from the kit)
   • Any k6-out/*-summary.json from k6 run --summary-export=…
   • Optional: JMeter CSV / other supported formats listed in the upload UI
4. Review the verdict, p95 / p99, error rate, throughput, failed checks, and recommendations.
5. Use history / compare to spot regressions across builds (optional git branch / commit metadata when present).

Load profiles (export options)

Exact VUs/duration can still be overridden via VUS / DURATION / env after export.

Supported upload formats

• k6 summary JSON — from --summary-export or kit *-summary.json
• testneo-import.json — aggregated by the kit collector
• JMeter / other — see the formats listed in the Load Insights upload panel when enabled

CI tips

• Kit includes a CI scaffold (e.g. GitHub Actions) and CI_SETUP.md when selected at export.
• Store BASE_URL and auth tokens as pipeline secrets.
• Archive k6-out/ as a build artifact; upload testneo-import.json to Load Insights manually or via your automation.
• Keep k6 workers off your shared app server — run on a dedicated runner or ephemeral CI job.

What k6 can / cannot compile

Skipped UI-only tests are listed in manifest.json inside the kit — keep them as functional Playwright runs in TestNeo.

Troubleshooting

Quick links