| ## Performance Optimization |
|
|
| ### API Performance Issues |
|
|
| If you experience slow response times or endless loops: |
|
|
| 1. **Check System Resources** |
| ```bash |
| # Monitor CPU and memory usage |
| htop |
| |
| # Check GPU memory (if using GPU) |
| nvidia-smi |
| ``` |
|
|
| 2. **Run Performance Tests** |
| ```bash |
| # Test API performance |
| python scripts/monitor_performance.py |
| ``` |
|
|
| 3. **Optimize Configuration** |
| ```bash |
| # Reduce batch sizes for memory-constrained systems |
| export EMBEDDING_BATCH_SIZE=32 |
| export MAX_CONTEXT_LENGTH=4096 |
| |
| # Use CPU-only mode if GPU memory is insufficient |
| export DEVICE=cpu |
| ``` |
|
|
| 4. **Common Performance Issues** |
| - **Endless loops**: Usually caused by LLM generation timeouts. Check your Ollama/Hugging Face connection. |
| - **Slow searches**: Vector database may need optimization. Consider rebuilding with smaller chunks. |
| - **Memory issues**: Reduce `MAX_SEARCH_RESULTS` and `MAX_CONTEXT_LENGTH` in config. |
|
|
| 5. **Timeout Settings** |
| - Embedding generation: 5 seconds |
| - Vector search: 10 seconds |
| - LLM generation: 30 seconds |
| - Query routing: 5 seconds |
|
|
| ### Performance Monitoring |
|
|
| The system includes built-in performance monitoring: |
|
|
| ```bash |
| # Monitor real-time performance |
| python scripts/monitor_performance.py |
| |
| # Check API health |
| curl http://localhost:8000/api/v1/health |
| |
| # Test specific endpoints |
| curl -X POST http://localhost:8000/api/v1/search \ |
| -H "Content-Type: application/json" \ |
| -d '{"query": "CVE-2021-44228", "top_k": 5}' |
| ``` |
|
|
