I got a problem with rancher, suddenly all service can not access, return the 503 error code from nginx

sounds like a problem with your ingress-controller, are the backing pods of ingress-nginx in a healthy/working state?

In healthe state. But the 'cattle-cluster-agent' is not healthy