REFACTOR external service and improve websocket training

services/training/Dockerfile

@@ -41,5 +41,7 @@ EXPOSE 8000
 HEALTHCHECK --interval=30s --timeout=10s --start-period=5s --retries=3 \
     CMD curl -f http://localhost:8000/health || exit 1
 
-# Run application
-CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]
+# Run application with increased WebSocket ping timeout to handle long training operations
+# Default uvicorn ws-ping-timeout is 20s, increasing to 300s (5 minutes) to prevent
+# premature disconnections during CPU-intensive ML training (typically 2-3 minutes)
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000", "--ws-ping-timeout", "300"]