typically, container crashes or something in error reporting where dependencies are not met, I would look in the log and isolate on a single container for the error, if a sidecar is involved as a backend and it is not available - add a health or liveliness probe to ensure service is up, likewise if you're using EFS or DB on the backend verify the services are up and ensure there isn't an error
Thanks @djs_75 There is no sidecar involved. Thanks for your suggestion on adding health checks to the CR service.