perf-ecosystem.yml

perf-ecosystem.yml is a single configuration file that all ecosystem tools discover automatically. Place it in your project root and each tool connects itself to the configured services.

Discovery

Tools walk up the directory tree from the current working directory until they find perf-ecosystem.yml or reach the .git boundary. No environment variable or flag is needed — discovery is automatic.

Full Reference

# Schema version: 1.0

version: "1.0"

workspace:
  name: "my-application"       # Project name, used in reports
  team: "platform-team"
  environment: staging         # dev | staging | prod

services:
  dummydatagenpro:
    url: "https://dummydatagenpro.co.uk"
    api_key: "${DUMMYDATA_API_KEY}"

  perf_lint_api:
    url: "https://perflint.martkos-it.co.uk"
    api_key: "${PERF_LINT_API_KEY}"

licenses:
  perf_reporting: "${PERF_REPORTING_LICENSE}"
  perf_lab: "${PERF_LAB_LICENSE}"
  perf_script_studio: "${PERF_SCRIPT_STUDIO_LICENSE}"
  # perf_migrator is still read, so existing configuration keeps working. There has
  # never been a perf_autocorrelator_pro key: the binary does not look for one, and a
  # licence configured under that name is silently ignored.
  perf_migrator: "${PERF_MIGRATOR_LICENSE}"

tools:
  default: jmeter              # jmeter | k6 | gatling
  jmeter:
    home: "/opt/apache-jmeter-5.6.3"
    image: "ghcr.io/markslilley/perf-jmeter:latest"
  k6:
    image: "ghcr.io/markslilley/perf-k6:latest"
  gatling:
    image: "ghcr.io/markslilley/perf-gatling:latest"

thresholds:
  response_time:
    avg_max_ms: 500
    p95_max_ms: 1500
    p99_max_ms: 3000
  error_rate:
    max_percent: 1.0
  apdex:
    min_score: 0.85

notifications:
  slack:
    webhook_url: "${SLACK_WEBHOOK_URL}"
    channel: "#perf-alerts"
    on_failure: true
    on_sla_breach: true
    on_regression: true

integrations:
  auto_lint_on_run: false      # Run perf-lint before executing scripts
  migrator_validate: false     # Auto-run perf-lint on perf-script-studio output

Environment Variable Interpolation

Values matching ${VAR_NAME} are substituted before parsing. An unresolved variable resolves to null — the integration is silently disabled rather than throwing an error. This means a partial config — only perf_lint_api, say — works correctly without needing to remove unconfigured sections.

services:
  perf_lint_api:
    api_key: "${PERF_LINT_API_KEY}"   # → null if var not set → remote rule sync disabled

Store secrets in .env (excluded from git) or your CI/CD secrets store. Never commit API keys directly.

Minimal Config

You don’t need all sections. A minimal setup that licenses the tools and sets one SLA:

version: "1.0"

licenses:
  perf_script_studio: "${PERF_SCRIPT_STUDIO_LICENSE}"
  perf_reporting: "${PERF_REPORTING_LICENSE}"

thresholds:
  response_time:
    p95_max_ms: 1500
  error_rate:
    max_percent: 1.0

Which Tools Consume Which Sections

Section	Consumed By
`services.perf_lint_api`	perf-lint (remote rule sync), perf-script-studio (post-convert lint)
`services.dummydatagenpro`	test-data generation steps
`licenses.*`	perf-reporting, perf-lab, perf-script-studio (`perf_migrator` still read)
`tools.*`	executor image selection
`thresholds.*`	perf-reporting (SLA checks and regression gating)
`integrations.*`	All tools — control auto-integration behaviour

Getting the Example File

curl -O https://raw.githubusercontent.com/markslilley/perf-ecosystem/main/perf-ecosystem.example.yml
cp perf-ecosystem.example.yml perf-ecosystem.yml