Getting started

Prepare a load plan for an owned app preview. Access to the service is estate-internal.
  1. 01

    Verify your hostname

    Sign in and add your owned hostname on Targets. Serve the issued token as plain text at /.well-known/loadtest-dev/<token> over HTTPS, then select Check now. The service rechecks ownership before a run and refuses private or reserved addresses.

  2. 02

    Add a plan

    Save a plan such as the example below in Config/load-plan.json. Set app to your repository slug and targets.preview to the hostname you verified. Stages use numeric durationS and rps fields. Only GET and HEAD requests are supported.

  3. 03

    Submit through the API

    An authorized operator submits POST /api/v1/runs with { app, environment: "preview", profile: "smoke", plan } through the authenticated API. This starts real traffic. Use only a target you own and a rate and duration you have approved. The console currently shows results and target checks; it does not provide a run-start form.

  4. 04

    Read the verdict

    Open Runs to read the result, or poll GET /api/v1/runs/:id. The API records pass, fail or aborted when a run ends. Your integration must handle those statuses explicitly; an HTTP 202 accepts a run and does not mean its thresholds passed.

Your first plan

A plan is declarative JSON: scenarios weighted by share of traffic, profiles made of arrival-rate stages, and the thresholds that decide pass or fail. No code runs from it.

Config/load-plan.json
{
  "version": 1,
  "app": "example-app",
  "targets": {
    "preview": "preview.example.com"
  },
  "scenarios": [
    {
      "id": "home",
      "weight": 3,
      "method": "GET",
      "path": "/"
    },
    {
      "id": "health",
      "weight": 1,
      "method": "GET",
      "path": "/api/health"
    }
  ],
  "profiles": {
    "smoke": {
      "stages": [
        {
          "durationS": 60,
          "rps": 5
        }
      ]
    },
    "baseline": {
      "stages": [
        {
          "durationS": 60,
          "rps": 50
        },
        {
          "durationS": 300,
          "rps": 50
        }
      ]
    }
  },
  "thresholds": {
    "p95Ms": 400,
    "errorRate": 0.01
  }
}

What to expect

While a run is live, every second of it is recorded:

  • Response time: p50, p95 and p99 latency in milliseconds, from merged histograms.
  • Throughput: requests per second actually sent, against the rate the plan asked for.
  • Errors by class: 5xx, 4xx, 429, timeout and transport failures, counted separately.
  • The build that answered: the x-build-version the target reported, so a result names the release it measured.

Production targets

Select environment: "production" in the API request only after approving traffic against that hostname. Production is capped at 300 requests per second; preview is capped at 1,000. Every profile is limited to 900 seconds, and ownership is rechecked before it starts.

loadtest.dev — Part of the Narduk Enterprises network