Skip to content

README.md

Latest commit

 

History

History
87 lines (68 loc) · 3.43 KB

README.md

File metadata and controls

87 lines (68 loc) · 3.43 KB
Raw

Performance Testing Infrastructure

This directory contains the performance testing suite for Scavenger.

Framework

We use k6 for load, stress, and endurance testing.

Directory Structure

  • scenarios/: k6 test scripts for different testing types (load, stress, endurance).
  • lib/: Shared functions and configurations.
  • baselines/: Performance baseline data for regression detection.
  • reports/: Generated test results and summary reports.
  • HOT_ENDPOINTS_LOAD_TESTING.md: Comprehensive guide for hot endpoints load testing (issue #950).
  • hot-endpoints-thresholds.json: Performance thresholds for identified hot endpoints.
  • k6-hot-endpoints-load-test.js: Load test script for hot endpoints (3 scenarios).
  • run-hot-endpoints-tests.sh: Runner script for hot endpoints load tests.

Test Types

  1. Load Test (load.js): Tests the system under expected normal load.
  2. Stress Test (stress.js): Tests the system's limits by gradually increasing load until it breaks or reaches a high threshold.
  3. Endurance Test (endurance.js): Tests system stability over an extended period.
  4. Hot Endpoints Load Test (k6-hot-endpoints-load-test.js): Comprehensive load testing for high-traffic endpoints (issue #950).

Running Tests

Requires a running environment (see the Developer Onboarding Guide) plus k6 installed locally.

General Test Suite

./performance/run-perf-tests.sh

Hot Endpoints Load Tests

# Run all hot endpoints load tests
./performance/run-hot-endpoints-tests.sh

# Run with custom base URL
BASE_URL=http://api.example.com ./performance/run-hot-endpoints-tests.sh

# Generate baselines
GENERATE_BASELINE=true ./performance/run-hot-endpoints-tests.sh

Specific Test

k6 run performance/scenarios/load.js
k6 run performance/k6-hot-endpoints-load-test.js

Hot Endpoints Load Testing (Issue #950)

As of v1.0, the hot endpoints load testing suite includes comprehensive tests for high-traffic endpoints:

Hot Endpoints Tested:

  • GET /api/v1/contracts/wastes - List wastes
  • GET /api/v1/contracts/participants - List participants
  • GET /api/v1/contracts/stats - Contract statistics
  • GET /api/v1/search - Full-text search
  • GET /api/v1/audit/logs - Audit logs
  • POST /api/v1/signing/sign - Transaction signing

Test Scenarios:

  1. Steady State: 100 concurrent users (9 minutes)
  2. Spike Test: Sudden increase to 1000 users (5 minutes)
  3. Stress Test: Gradual increase to 10000 users (13 minutes)

Performance Thresholds:

  • Steady State: p95 < 500ms, error rate < 5%
  • Spike: p95 < 1000ms, error rate < 10%
  • Stress: p95 < 2000ms, error rate < 20%

See HOT_ENDPOINTS_LOAD_TESTING.md for complete details.

Baselines and Alerts

The suite includes a baseline comparison tool.

  • To generate a new baseline: GENERATE_BASELINE=true ./performance/run-perf-tests.sh
  • The analyze-results.js script automatically compares current results with the baseline and reports regressions.
  • Hot endpoints baselines are defined in baselines.json under the hot_endpoints section.

Metrics Tracked

  • http_req_duration: End-to-end request time (p95, avg).
  • errors: Rate of non-200/409 responses.
  • api_duration: Custom trend for API specific timing.
  • Endpoint-specific trends: list_wastes_duration, search_duration, audit_logs_duration, etc.