Michael Freno
540ca5ebad
- Create load test directory structure (infra/load-tests/) - Implement k6 script for Darkwatch endpoints (darkwatch.js) - Tests watchlist, scan, exposure, and alert operations - Configured for 500 req/s sustained load with P99 < 200ms - Includes error rate metrics and threshold validation - Add documentation and usage guide (README.md) Related: [FRE-4807](/FRE/issues/FRE-4807) Co-Authored-By: Paperclip <noreply@paperclip.ing>
1.4 KiB
1.4 KiB
ShieldAI Load Tests
k6 load testing suite for ShieldAI services.
Prerequisites
- k6 v0.45+ installed
- Target services running on staging environment
- Authentication tokens for API access
Running Tests
Local Execution
# Run against local development environment
k6 run --env BASE_URL=http://localhost:3000 --env AUTH_TOKEN=dev-token src/darkwatch.js
# Run with results output
k6 run --out json=results.json src/darkwatch.js
CI/CD Execution
# Run on staging environment
k6 run --env BASE_URL=https://staging-api.freno.me --env AUTH_TOKEN=$STAGING_AUTH_TOKEN src/darkwatch.js
Test Configuration
Each test script includes:
- Stages: Ramp-up, sustained load, ramp-down
- Thresholds: P99 latency and error rate limits
- Metrics: Custom metrics for error tracking
Current Thresholds
| Service | P99 Latency | Error Rate |
|---|---|---|
| Darkwatch | < 200ms | < 1% |
Metrics Collection
Run with output options:
# JSON output for analysis
k6 run --out json=darkwatch-results.json src/darkwatch.js
# InfluxDB for visualization
k6 run --out influxdb=http://influxdb:8086/k6 src/darkwatch.js
Next Steps
- Create load test scripts for Spamshield and Voiceprint
- Integrate with GitHub Actions CI pipeline
- Set up metrics visualization dashboard
- Configure alerting on threshold breaches