Workflow appears to trigger the server to restart leaving the webui waiting indefinitely #569
Description
I have tried ChatDev out for a few days and frequently get the issue where the webui will just be waiting indefinitely. Looking at the terminal output, it appears that the server is detecting the generated output (in this case, player.py) as a change to the server code and doing a reload of the server, naturally interrupting the workflow.
Should StatReload be ignoring these directories?
WARNING: StatReload detected changes in 'WareHouse/session_9d1c6544-8407-4067-b09d-fe817d609215/code_workspace/player.py'. Reloading...
INFO: Shutting down
INFO:server.services.workflow_run_service:Cancellation requested for session 9d1c6544-8407-4067-b09d-fe817d609215
INFO:session_db4036ba-a8ed-4258-ab46-a5fd8fa9c2ce:{"timestamp": "2026-03-09 03:22:50.199098+00:00", "log_type": "workflow", "level": "INFO", "logger": "session_db4036ba-a8ed-4258-ab46-a5fd8fa9c2ce", "message": "Cancellation requested for workflow session_9d1c6544-8407-4067-b09d-fe817d609215", "correlation_id": null, "workflow_id": "session_9d1c6544-8407-4067-b09d-fe817d609215", "node_id": null, "event_type": null, "execution_path": ["Programmer Coding"], "duration": null}
INFO:server.services.session_store:Updated session 9d1c6544-8407-4067-b09d-fe817d609215 status to cancelled
INFO:server.services.session_execution:Session 9d1c6544-8407-4067-b09d-fe817d609215 cleaned from execution controller
INFO:server.services.attachment_service:Attachment cleanup disabled; preserved files for session 9d1c6544-8407-4067-b09d-fe817d609215
03:22:50 [vite] ws proxy error:
Error: read ECONNRESET
at TCP.onStreamRead (node:internal/stream_base_commons:216:20) (x2)
INFO: Waiting for application shutdown.
INFO: Application shutdown complete.
INFO: Finished server process [296903]
INFO: Started server process [305891]
INFO: Waiting for application startup.
INFO: Application startup complete.