Combined MCP Server

# Troubleshooting Guide Common issues and solutions for the Combined MCP Server. ## Connection Issues ### PostgreSQL Connection Failed **Symptom:** `PostgresConnectionError: Failed to connect` **Solutions:** 1. Verify PostgreSQL is running on the configured host/port 2. Check credentials in environment variables or Secrets Manager 3. Ensure the `vectordb` database exists 4. Verify pgvector extension is installed ```sql -- Check if pgvector is installed SELECT * FROM pg_extension WHERE extname = 'vector'; -- Install if missing CREATE EXTENSION IF NOT EXISTS vector; ``` ### Redshift Authentication Failed **Symptom:** `RedshiftConnectionError: Failed to get credentials` **Solutions:** 1. Verify IAM permissions for `redshift:GetClusterCredentials` 2. Check cluster identifier is correct 3. Ensure the db_user has appropriate permissions --- ## Vector Store Issues ### Build Failed **Symptom:** Vector store build fails or times out **Solutions:** 1. Check S3 bucket permissions 2. Verify Bedrock access for embedding model 3. Reduce batch size for large document sets 4. Check PostgreSQL disk space ### Search Returns No Results **Symptom:** `query_vectorstore` returns empty results **Solutions:** 1. Verify vector store is built (`get_vectorstore_status`) 2. Check if documents were loaded: `document_count > 0` 3. Try different search types (semantic vs keyword) 4. Verify query embeddings are being generated --- ## Performance Issues ### Slow Query Response **Symptom:** Queries take several seconds **Solutions:** 1. Check if result is cached (should be faster on repeat) 2. Reduce `top_k` parameter 3. Ensure HNSW index exists on embedding column 4. Consider upgrading PostgreSQL resources ### High Memory Usage **Symptom:** Server consumes excessive memory **Solutions:** 1. Reduce `QUERY_CACHE_SIZE` environment variable 2. Use smaller embedding dimension if possible 3. Implement document chunking for large files --- ## Health Check Failures ### Readiness Probe Failing **Symptom:** `/ready` returns 503 **Causes:** 1. Vector store not yet initialized 2. PostgreSQL connection not established 3. Build process still running **Check status:** ```bash curl http://localhost:8080/status ``` --- ## Logging Enable debug logging for more information: ```bash export LOG_LEVEL=DEBUG export LOG_FORMAT=console ``` Logs include: - Connection attempts and results - Query execution times - Cache hits/misses - S3 operations