Troubleshooting#
This section contains common issues and how to troubleshoot them.
Common issues#
If you ingest documents and then run retrieval, but see the following result
Response: "I don't know"
Make sure that the request id in both config/init requests is the same. If they are the same, then try reinitializing the context manager.
Health Checks#
Both services, ingestion and retrieval, provide health check endpoints that can be used to verify service status:
import requests
# Check Data Ingestion Service health
response = requests.get("http://localhost:8001/health")
print(response.text)
# Check Retrieval Service health
response = requests.get("http://localhost:8000/health")
print(response.text)
Common Error Messages and Solutions#
“Context manager not initialized”
Cause: Service not initialized or wrong UUID
Solution: Initialize both services with the same UUID
“No context was provided”
Cause: Documents not properly ingested or wrong UUID
Solution: Verify documents were added successfully and UUIDs match
“Invalid request format”
Cause: Incorrect JSON structure in request
Solution: Check request format against examples in the documentation
Monitoring and Debugging#
Important Log Messages#
Data Ingestion Service:
“Adding doc {index}” - Document being processed
“Batch {index} is full. Processing …” - Batch processing
“start_pts or end_pts not found” - Missing timestamp metadata
Retrieval Service:
“Using {model} as the {type} llm” - Model initialization
“Setting up QnA, rag type: {type}” - RAG configuration
Metrics#
Both services expose metrics at /metrics
endpoint:
Request latency
Document processing time
Error rates
System resource usage