Mandatory Think Time Between Requests

This rule applies to files matching the patterns above.

# Mandatory Think Time Between Requests

## Rule
k6 scripts MUST include `sleep()` between HTTP requests to simulate realistic user think time. Scripts without think time create artificial burst traffic that does not represent real-world load.

## Format
```javascript
import { sleep } from 'k6';

export default function () {
  http.get(BASE_URL + '/page1');
  sleep(1);                        // 1 second think time

  http.get(BASE_URL + '/page2');
  sleep(Math.random() * 3 + 1);   // 1-4 seconds random think time

  http.post(BASE_URL + '/submit', payload);
  sleep(2);                        // User reviews confirmation
}
```

## Think Time Guidelines
| Action | Think Time |
|--------|-----------|
| Between page navigations | 2-5 seconds |
| After form submission | 1-3 seconds |
| Between API calls in a flow | 0.5-2 seconds |
| After reading content | 3-10 seconds |
| After login | 1-2 seconds |

## Realistic vs Burst Traffic
```javascript
// Bad — burst traffic (no think time)
export default function () {
  http.get(BASE_URL + '/api/users');
  http.get(BASE_URL + '/api/products');
  http.get(BASE_URL + '/api/orders');
  // 3 requests in ~0ms per VU — unrealistic
}

// Good — realistic user behavior
export default function () {
  http.get(BASE_URL + '/api/users');
  sleep(2);
  http.get(BASE_URL + '/api/products');
  sleep(3);
  http.get(BASE_URL + '/api/orders');
  sleep(1);
  // 3 requests in ~6s per VU — realistic
}
```

## Randomized Think Time
```javascript
// More realistic — users don't all pause for exactly the same duration
function thinkTime(min, max) {
  sleep(Math.random() * (max - min) + min);
}

export default function () {
  http.get(BASE_URL + '/page');
  thinkTime(1, 5); // 1 to 5 seconds
}
```

## Exceptions
- API-to-API load tests (no human user) may omit think time
- Spike tests may use minimal think time to simulate sudden bursts
- Document the reason when omitting think time

## Anti-Patterns
- No sleep() between any requests (artificial burst)
- sleep(0.01) as a token think time (still effectively burst)
- Uniform think time across all actions (not realistic)
- Removing think time to "increase load" (increase VUs instead)