we have a simple fastapi application deployed on GCP cloud run, The cloud run has a timeout of 15s but sometimes the downstream API call done by fastapi exceed this time. We are experiencing that once the 15s of cloud run timeout is reached the API call is closed correctly from the perspective of the client but looking at the logs the fastapi request is not killed and still processing till normal finish of the process. We setup a timeout on the downstream API instead relying on the cloud run timeout but is a strange behavior. Any idea of what may be the issue? Continue reading...
