Troubleshooting Exponential Backoff in OpenAI Node.js with Fikra API

When integrating Fikra API using the OpenAI Node.js SDK, you may encounter transient errors such as 429 (rate limit exceeded) or 503 (upstream/provider temporarily unavailable). Implementing exponential backoff helps manage these scenarios gracefully.

Error Codes

• 400: Malformed request or unsupported model.
• 401: Missing, invalid, or inactive API key.
• 402: Insufficient wallet balance.
• 429: Rate limit exceeded.
• 503: Upstream/provider temporarily unavailable.

Exponential Backoff Example

Below is a compact example of implementing exponential backoff for 429 and 503 errors using the OpenAI Node.js SDK with Fikra API.

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.FIKRA_API_KEY,
  baseURL: 'https://api.fikraapi.co.ke/v1',
});

async function chatWithBackoff(messages, model = 'fikra-flash') {
  let attempt = 0;
  const maxRetries = 5;

  while (attempt < maxRetries) {
    try {
      const response = await client.chat.completions.create({
        model,
        messages,
      });
      return response;
    } catch (error) {
      const status = error.status;
      if (status === 429 || status === 503) {
        attempt += 1;
        const delay = Math.min(1000 * Math.pow(2, attempt), 30000); // Cap at 30 seconds
        await new Promise((resolve) => setTimeout(resolve, delay));
      } else {
        throw error; // Non-retryable error
      }
    }
  }

  throw new Error('Max retries exceeded');
}

// Usage
chatWithBackoff([{ role: 'user', content: 'Hello!' }])
  .then((response) => console.log(response.choices[0].message.content))
  .catch((error) => console.error(error));

Notes

• Adjust maxRetries and delay caps based on your application's requirements.
• Ensure your API key is valid and your wallet balance is sufficient to avoid 401 and 402 errors.
• For persistent issues, check the Fikra dashboard for account status and rate limits.