Skip to main content

​
Troubleshooting

Find solutions to common issues when working with Meibel AI.

​
Authentication Issues

​
Invalid API Key Error

Problem: Getting “401 Unauthorized” or “Invalid API key” errors. Solutions:
  1. Verify your API key is correct: 
    print(os.getenv("MEIBELAI_API_KEY_HEADER"))
  2. Check the key hasn’t expired in the Console
  3. Ensure you’re using the correct header format: 
    client = Meibelai(api_key_header="your-key-here")

​
Permission Denied

Problem: “403 Forbidden” errors when accessing resources. Solutions:
  • Verify your API key has the necessary permissions
  • Check if you’re accessing resources in the correct organization
  • Ensure you’re not exceeding plan limits

​
Connection Issues

​
Timeout Errors

Problem: Requests timing out before completion. Solutions: 
# Increase timeout
client = Meibelai(
    api_key_header=os.getenv("MEIBELAI_API_KEY_HEADER"),
    timeout=60  # 60 seconds
)

# For specific operations
response = client.rag.chat(
    messages=[{"role": "user", "content": "Query"}],
    execution_control={"timeout": 45}
)

​
SSL Certificate Errors

Problem: SSL verification failures. Solutions:
  1. Update your certificates: 
    pip install --upgrade certifi
  2. For development only (not recommended for production): 
    client = Meibelai(
    api_key_header="your-key",
    verify_ssl=False  # Only for debugging
)

​
Rate Limiting

​
429 Too Many Requests

Problem: Hitting rate limits. Solutions:
  1. Implement exponential backoff: 
    import time
from meibelai.core import MeibelaiError

def retry_with_backoff(func, max_retries=3):
    for i in range(max_retries):
        try:
            return func()
        except MeibelaiError as e:
            if e.status_code == 429 and i < max_retries - 1:
                wait_time = 2 ** i
                time.sleep(wait_time)
            else:
                raise
  2. Use batch operations when possible
  3. Consider upgrading your plan for higher limits

​
Data Issues

​
Empty Response

Problem: Getting empty or null responses. Solutions:
  1. Check if datasource has content: 
    elements = client.dataelements.list(datasource_id=your_id)
print(f"Elements in datasource: {len(elements)}")
  2. Verify confidence threshold isn’t too high: 
    response = client.rag.chat(
    messages=[{"role": "user", "content": "Query"}],
    datasource_ids=[datasource_id],
    confidence_threshold=0.5  # Lower threshold
)

​
Incorrect Results

Problem: AI returning inaccurate or irrelevant responses. Solutions:
  1. Improve your prompts: 
    messages = [
    {"role": "system", "content": "You are a helpful assistant. Be precise and accurate."},
    {"role": "user", "content": "Specific question here"}
]
  2. Add more context to your data: 
    client.dataelements.create(
    datasource_id=datasource_id,
    name="Document",
    content="Your content",
    metadata={
        "category": "technical",
        "date": "2024-01",
        "relevance": "high"
    }
)

​
Performance Issues

​
Slow Response Times

Problem: API calls taking longer than expected. Solutions:
  1. Use streaming for long responses: 
    stream = client.rag.chat_stream(
    messages=[{"role": "user", "content": "Query"}],
    datasource_ids=[datasource_id]
)

for chunk in stream:
    print(chunk.choices[0].delta.content, end="")
  2. Optimize your queries:
    • Use specific datasources instead of searching all
    • Set appropriate token limits
    • Enable caching when possible

​
Memory Issues

Problem: Running out of memory with large datasets. Solutions:
  1. Process data in batches: 
    def process_in_batches(items, batch_size=10):
    for i in range(0, len(items), batch_size):
        batch = items[i:i + batch_size]
        yield process_batch(batch)
  2. Use async operations for better resource usage

​
SDK-Specific Issues

​
Import Errors

Problem: Can’t import the Meibel AI SDK. Solutions: 
# Reinstall the SDK
pip uninstall meibelai
pip install git+https://github.com/meibel-ai/meibelai-python.git

# Check installation
python -c "import meibelai; print(meibelai.__version__)"

​
Version Conflicts

Problem: Dependency conflicts with other packages. Solutions:
  1. Use a virtual environment: 
    python -m venv meibel_env
source meibel_env/bin/activate  # On Windows: meibel_env\Scripts\activate
pip install git+https://github.com/meibel-ai/meibelai-python.git
  2. Check Python version compatibility (requires 3.9+)

​
Debugging Tips

​
Enable Tracing

Get detailed execution information: 
response = client.rag.chat(
    messages=[{"role": "user", "content": "Query"}],
    execution_control={"enable_tracing": True}
)

# Access trace information
if hasattr(response, 'trace'):
    print(response.trace)

​
Log API Calls

Monitor what’s being sent: 
import logging

logging.basicConfig(level=logging.DEBUG)
logger = logging.getLogger('meibelai')

​
Getting Help

If you’re still experiencing issues:
  1. Check Status: Visit status.meibel.ai for service status
  2. Documentation: Review the API Reference
  3. Community: Ask questions in our community forum
  4. Support: Contact support@meibel.ai with:
    • Error messages and stack traces
    • Request IDs from error responses
    • Steps to reproduce the issue
    • SDK version and Python version

​
Next Steps