Startup-loop when using for machine learning deployment about uvicorn-gunicorn-fastapi-docker HOT 5 CLOSED

Did you find a solution for this? I am encountering the same problem.

from uvicorn-gunicorn-fastapi-docker.

Unfortunately no, we switched to a different container for now ..

from uvicorn-gunicorn-fastapi-docker.

Reporting same issue - for me it happens if I run this container dependent on a database connection when the database is not available. My code correctly decides that the other end is not available and issues a sys.exit() but the fastapi container keeps spinning.

from uvicorn-gunicorn-fastapi-docker.

Hey @JannikZed , that's probably because the model is taking a long time to load, then Gunicorn might be killing the workers before they finish.

You can now set timeouts to prevent that: https://github.com/tiangolo/uvicorn-gunicorn-fastapi-docker#graceful_timeout 🚀

from uvicorn-gunicorn-fastapi-docker.

Assuming the original issue was solved, it will be automatically closed now. But feel free to add more comments or create new issues.

from uvicorn-gunicorn-fastapi-docker.