--timeout 120, looking at logs when the worker first starts up reveals many worker processes being killed by Gunicorn when they don't respond in time (I think). Upping this parameter to be above the default (30s) should allow the first process spawned when Google Cloud Run scales up to successfully spin up and handle the http request
--preload passing this flag should help with initial startup time.
Update flags passed to gunicorn to include:
--timeout 120, looking at logs when the worker first starts up reveals many worker processes being killed by Gunicorn when they don't respond in time (I think). Upping this parameter to be above the default (30s) should allow the first process spawned when Google Cloud Run scales up to successfully spin up and handle the http request--preloadpassing this flag should help with initial startup time.