Guide

Load tests

load BusyMorning {
  users 20
  ramp 10s
  duration 1m
  think 200ms..1s
  seed busy-morning
  threshold p95 < 800ms
  threshold errors < 1%
  threshold checks > 95%
  scenario {
    run listProducts
    run getProduct
  }
}

Thresholds decide the exit code, so a slow release fails a pipeline the way a broken test does. Metrics: p50, p90, p95, p99, avg, max, errors, rps, checks, requests.

stamp load .                      # every load test
stamp load . --json report.json   # machine readable

In the app: under Load Tests in the sidebar, with requests per second, p95 latency and active users charted live, plus the per request table and the thresholds.

A load test running: requests, throughput, p95 and errors, with three charts filling in live
A load test running: requests, throughput, p95 and errors, with three charts filling in live

Back: Test plans Next: Race conditions