# PromQL query 1 - Sum total 4xx and 5xx
sum(litellm_proxy_total_requests_metric_total { cluster="AAA", litellm_identifier="atlas", status_code=~"[4-5].." })

#  Will not match total failed request
sum(litellm_proxy_failed_requests_metric_total { cluster="AAA", litellm_identifier="atlas" })

# But if we include total with status_code equal to none it will match litellm_proxy_failed_requests_metric_total 
sum(litellm_proxy_total_requests_metric_total {
  cluster="AAA",
  litellm_identifier="atlas",
  status_code=~"[4-5]..|None" 
})

---

import http from 'k6/http';
import { check, sleep } from 'k6';

// Test configuration
export const options = {
  scenarios: {
    constant_request_rate: {
      executor: 'constant-arrival-rate',
      rate: 2, // 2 requests per second
      timeUnit: '1s',
      duration: '10m', // Run for 10 minutes (adjust as needed)
      preAllocatedVUs: 2, // Pre-allocate 2 virtual users
      maxVUs: 10, // Maximum virtual users if needed
    },
  },
};

// Configuration
const BASE_URL = 'https://....';
const API_KEY = '.....';

// Models to rotate between
const MODELS = ['gpt-4.1-nano', 'gpt-4.1'];

export default function () {
  // Rotate between models
  const model = MODELS[__ITER % MODELS.length];

  // Alternate between malformed body (50%) and auth error (50%)
  const errorType = Math.random();
  
  let payload;
  let apiKey = API_KEY;
  
  if (errorType < 0.5) {
    // 50% malformed body - missing required messages field
    payload = JSON.stringify({
      model: model,
      // Missing messages field
    });
    console.log('Sending malformed request (missing messages)');
  } else {
    // 50% auth error - wrong API key
    apiKey = 'sk-invalid-key-12345';
    payload = JSON.stringify({
      model: model,
      messages: [
        {
          role: 'user',
          content: 'What is 1+1?',
        },
      ],
    });
    console.log('Sending auth error request (invalid API key)');
  }

  // Request parameters
  const params = {
    headers: {
      'Content-Type': 'application/json',
      'x-litellm-api-key': apiKey,
    },
    timeout: '60s',
  };

  // Make the request
  const response = http.post(`${BASE_URL}/chat/completions`, payload, params);

  // Check response
  check(response, {
    'status is 4xx or 5xx': (r) => r.status >= 400,
    'response has body': (r) => r.body.length > 0,
    'response is valid JSON': (r) => {
      try {
        JSON.parse(r.body);
        return true;
      } catch (e) {
        return false;
      }
    },
  });

  // Log the model used and response details
  console.log(`Request to ${model} - Status: ${response.status} - Duration: ${response.timings.duration}ms`);
  
  if (response.status !== 200) {
    console.log(`Error response body: ${response.body}`);
  }
}

---