load-testing

Create and manage K6 load tests for the REST and GraphQL APIs of a PHP backend service. Use when creating load tests, writing K6 scripts, testing API performance, debugging load test failures, or sett

SKILL.md

GitHub

Load Testing Skill

Profile keys consumed

  • capabilities.load_testing
  • make.load_tests
  • make.start
  • framework.graphql
  • framework.api_platform
  • persistence.engine
  • persistence.mapper
  • architecture.bounded_contexts

Gating

This skill is gated by capabilities.load_testing. When it is false, or when make.load_tests is null, record a capability-absent note ("load testing not configured for this project", NFR-4) and skip — never improvise raw k6 run commands against a repository that does not declare the capability.

Overview

This skill provides guidance for creating and managing K6 load tests for both REST and GraphQL APIs. GraphQL scenarios apply only when framework.graphql is true. IRI-handling guidance applies when framework.api_platform is set (non-false).

Core Principles

1. Individual Endpoint Testing

  • Create separate test scripts for each endpoint (REST) or operation (GraphQL)
  • Follow the pattern: createResource.js, getResource.js, updateResource.js, deleteResource.js
  • For GraphQL (when framework.graphql is true): graphQLCreateResource.js, graphQLGetResource.js, etc.
  • Avoid composite/random operation scripts for better debugging and clarity

2. Deterministic Testing

  • NEVER use random operations in load tests
  • Use predictable, iteration-based patterns (__ITER % N)
  • Ensure reproducible results for reliable performance analysis

3. Proper Resource Management

  • Implement setup() function to create test dependencies
  • Implement teardown() function to clean up test data
  • Use proper IRI handling for REST APIs (API Platform resources)
  • Use proper ID handling for GraphQL queries/mutations

4. Automatic Integration

  • All test scripts are automatically discovered from the harness scripts directory (conventionally tests/Load/scripts/)
  • No separate commands needed — GraphQL and REST tests run together
  • Invoke load tests only through the profile's make target map

Available Commands

The full suite runs through the target mapped by make.load_tests. Repositories that adopt the canonical K6 harness also expose level-specific variants (smoke/average/stress/spike), a single-scenario runner, and a scenario discovery script — find them in the repository Makefile next to the make.load_tests target. Canonical upstream layout:

make load-tests            # make.load_tests → all load tests (REST + GraphQL)
make smoke-load-tests      # minimal load (2-5 VUs, 10s)
make average-load-tests    # normal load (10-20 VUs, 2-3 min)
make stress-load-tests     # high load (30-80 VUs, 5-15 min)
make spike-load-tests      # extreme load (100-200 VUs, 1-3 min)
make execute-load-tests-script scenario=createCustomer    # one scenario
./tests/Load/get-load-test-scenarios.sh                   # list scenarios

Prerequisites: service containers running (target mapped by make.start); harness .env pointing BASE_URL at the running service (see reference/configuration.md).

Quick Start Guide

1. Choose Test Type

  • REST API: Use for HTTP endpoint testing
  • GraphQL: Use for GraphQL query/mutation testing — only when framework.graphql is true

2. Create Test Script

# Create in the harness scripts directory
touch tests/Load/scripts/yourOperation.js         # REST
touch tests/Load/scripts/graphQLYourOperation.js  # GraphQL

3. Follow Script Structure

Use the script-structure template and operation patterns for your test type:

The Quick Reference below summarizes the step order; Supporting Files cover configuration, Utils extension, and troubleshooting.

4. Add Configuration

Update the scenario configuration file (tests/Load/config.json.dist) with the script's VU/duration parameters — see reference/configuration.md.

5. Test and Verify

  • Run the smoke level first (lowest-load variant of make.load_tests)
  • Verify cleanup: check no test data remains in the database

Load Test Levels

Level VUs Duration Success Rate Purpose
Smoke 2-5 10 seconds 100% Basic functionality verification
Average 10-20 2-3 minutes >99% Normal traffic simulation
Stress 30-80 5-15 minutes >95% Find breaking points
Spike 100-200 1-3 minutes >90% Test resilience under sudden load

Common Pitfalls

Don't Do This

// Random operations - unpredictable results
const operation = Math.random();

// Hardcoded test data
const email = 'test@example.com'; // Will cause conflicts

// Missing cleanup in teardown()

Do This Instead

// Deterministic operations
const operationIndex = __ITER % 3;

// Dynamic test data
const email = `test_${Date.now()}_${randomString(6)}@example.com`;

// Proper cleanup
export function teardown(data) {
  // Clean up all created resources
}

Checklist for New Load Tests

Before Creating

  • Identify the specific endpoint/operation to test
  • Determine if REST or GraphQL (GraphQL only when framework.graphql is true)
  • Identify required dependencies (related resources the endpoint needs)
  • Plan realistic test data generation
  • Choose appropriate load test parameters

During Creation

  • Follow the appropriate script structure template (reference/rest-api-patterns.md for REST, reference/graphql-patterns.md for GraphQL)
  • Implement proper setup/teardown functions
  • Use deterministic operations (no random)
  • Handle IRI/ID paths correctly
  • Add configuration to the scenario config file
  • Use proper naming: graphQL prefix for GraphQL tests

After Creation

  • Verify automatic discovery via the harness scenario-listing script
  • Test with smoke load first
  • Verify 100% success rate in controlled environment
  • Check that cleanup works properly (no leftover data)
  • Document any special requirements

Performance Monitoring

Success Criteria

  • Smoke Tests: 100% success rate
  • Average Tests: >99% success rate
  • Stress Tests: >95% success rate
  • Response Times: <threshold configured per endpoint

Key Metrics

  • HTTP status codes (201, 200, 204 for success)
  • Response times (avg, p95, p99)
  • Error rates and types
  • Throughput (requests per second)

Supporting Files

For detailed patterns, examples, and reference documentation:

For functional (unit/integration/E2E/mutation) testing, use the testing-workflow skill instead.

Quick Reference

REST API Test Structure

  1. Import required modules
  2. Create Utils and ScenarioUtils instances
  3. Export options from scenarioUtils
  4. Implement setup() for dependencies
  5. Implement default function for main test logic
  6. Implement teardown() for cleanup
  7. Use IRI format for resource references (when framework.api_platform is set)

GraphQL Test Structure

  1. Import required modules
  2. Create Utils and ScenarioUtils instances
  3. Export options from scenarioUtils
  4. Use REST API in setup() for faster dependency creation
  5. Use GraphQL in default function for actual testing
  6. Use REST API in teardown() for faster cleanup
  7. Handle full IRI format in queries/mutations
  8. Validate response.data and check for errors

This skill ensures consistent, professional, and effective load testing for both REST and GraphQL APIs of any profiled PHP backend service.