website-enchun-mgr/apps/backend/tests/k6

K6 Load Testing Framework

Overview

This directory contains load testing scripts for the Enchun CMS backend using k6. The tests are designed to validate the non-functional requirement NFR4:

• Target: p95 response time < 500ms
• Error rate: < 1%
• Concurrent users: 100

Prerequisites

1. Install k6:

   # macOS
   brew install k6
   
   # Linux
   sudo apt-get install k6
   
   # Windows (using Chocolatey)
   choco install k6
   

2. Set up environment:

   cp .env.example .env.k6
   # Edit .env.k6 with your test environment credentials
   

Test Scripts

1. Public Browsing Test (public-browsing.js)

Simulates 100 concurrent users browsing public pages:

• Homepage
• About page
• Solutions page
• Portfolio list
• Blog list
• Contact page

Run:

k6 run --env BASE_URL=http://localhost:3000 tests/k6/public-browsing.js

Expected Results:

• p95 response time < 500ms
• Error rate < 1%
• 100 concurrent users sustained for 2 minutes

2. Admin Operations Test (admin-operations.js)

Simulates 20 concurrent admin users performing:

• Login
• List pages/posts
• Create/edit content
• Delete content

Run:

k6 run --env BASE_URL=http://localhost:3000 \
  --env ADMIN_EMAIL=admin@example.com \
  --env ADMIN_PASSWORD=password123 \
  tests/k6/admin-operations.js

Expected Results:

• p95 response time < 500ms
• Error rate < 1%
• 20 concurrent users sustained for 3 minutes

3. API Performance Test (api-performance.js)

Tests specific API endpoints:

• GraphQL API queries
• REST API endpoints
• Global API endpoint

Run:

k6 run --env BASE_URL=http://localhost:3000 tests/k6/api-performance.js

Expected Results:

• p95 response time < 300ms (faster for API)
• Error rate < 0.5%
• Throughput > 100 requests/second

Configuration

Environment Variables

Variable	Description	Default	Required
BASE_URL	Target server URL	http://localhost:3000	Yes
ADMIN_EMAIL	Admin user email	-	For admin tests
ADMIN_PASSWORD	Admin user password	-	For admin tests
VUS	Number of virtual users	Varies per test	No
DURATION	Test duration	Varies per test	No

Load Profiles

Each test uses different load profiles:

Test	Virtual Users	Duration	Ramp Up
Public Browsing	100	2m	30s
Admin Operations	20	3m	30s
API Performance	50	5m	1m

Running Tests

Run Single Test

# Basic run
k6 run tests/k6/public-browsing.js

# With custom environment
k6 run --env BASE_URL=https://staging.enchun.tw tests/k6/public-browsing.js

# With custom stages
k6 run --env STAGED_USERS=200 --env STAGED_DURATION=10m tests/k6/public-browsing.js

Run All Tests

# Using the npm script
pnpm test:load

# Or manually
k6 run tests/k6/public-browsing.js
k6 run tests/k6/admin-operations.js
k6 run tests/k6/api-performance.js

Run with Output Options

# Generate JSON report
k6 run --out json=results.json tests/k6/public-browsing.js

# Generate HTML report (requires k6-reporter)
k6 run --out json=results.json tests/k6/public-browsing.js
k6-reporter results.json --output results.html

Interpreting Results

Key Metrics

1. Response Time (p95): 95th percentile response time

   • ✅ Pass: < 500ms
   • ❌ Fail: >= 500ms

2. Error Rate: Percentage of failed requests

   • ✅ Pass: < 1%
   • ❌ Fail: >= 1%

3. Throughput: Requests per second

   • ✅ Pass: > 100 req/s for API, > 50 req/s for pages
   • ❌ Fail: Below threshold

4. Virtual Users (VUs): Active concurrent users

   • Should sustain target VUs for the full duration

Example Output

✓ Status is 200
✓ Response time < 500ms
✓ Page content loaded

checks.........................: 100.0% ✓ 12000     ✗ 0
data_received..................: 15 MB  125 kB/s
data_sent......................: 2.1 MB 18 kB/s
http_req_blocked...............: avg=1.2ms    min=0.5µs   med=1µs     max=125ms
http_req_connecting............: avg=500µs    min=0s      med=0s      max=45ms
http_req_duration..............: avg=185.3ms  min=45ms    med=150ms   max=850ms  p(95)=420ms
http_req_failed................: 0.00%  ✓ 0       ✗ 12000
http_req_receiving.............: avg=15ms     min=10µs    med=50µs    max=250ms
http_req_sending...............: avg=50µs     min=5µs     med=20µs    max=5ms
http_req_tls_handshaking.......: avg=0s       min=0s      med=0s      max=0s
http_req_waiting...............: avg=170ms    min=40ms    med=140ms   max=800ms
http_reqs......................: 12000  100 req/s
iteration_duration.............: avg=5.2s     min=1.2s    med=5s      max=12s
iterations.....................: 2400   20  /s
vus............................: 100    min=100    max=100
vus_max........................: 100    min=100    max=100

CI/CD Integration

GitHub Actions

name: Load Tests
on:
  schedule:
    - cron: '0 2 * * *'  # Daily at 2 AM
  workflow_dispatch:

jobs:
  k6:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - uses: grafana/k6-action@v0.3.1
        with:
          filename: tests/k6/public-browsing.js
        env:
          BASE_URL: ${{ secrets.STAGING_URL }}

Troubleshooting

Connection Refused

Error: connect attempt failed

Solution:

• Ensure the backend server is running
• Check BASE_URL is correct
• Verify firewall settings

High Error Rate

Error: http_req_failed > 1%

Possible Causes:

1. Server overload (reduce VUs)
2. Database connection issues
3. Authentication failures
4. Network issues

Debug:

# Run with fewer users
k6 run --env STAGED_USERS=10 tests/k6/public-browsing.js

# Check server logs
pnpm dev
# In another terminal, watch logs while tests run

Slow Response Times

Error: p(95) >= 500ms

Possible Causes:

1. Unoptimized database queries
2. Missing indexes
3. Large payloads
4. Server resource constraints

Next Steps:

1. Review database queries
2. Check caching strategies
3. Optimize images/assets
4. Scale server resources

Performance Baseline

Initial baseline (to be established):

Test	p95 (ms)	Error Rate	Throughput
Public Browsing	TBD	TBD	TBD
Admin Operations	TBD	TBD	TBD
API Performance	TBD	TBD	TBD

Update this table after first test run to establish baseline.

Resources

• k6 Documentation
• k6 Metrics
• Payload CMS Performance
• Web Vitals

Maintenance

• Review and update test scenarios quarterly
• Adjust load profiles based on real traffic patterns
• Update thresholds based on business requirements
• Add new tests for new features

---

Last Updated: 2025-01-31 Owner: QA Team