Improve health check logic in getClusterHealthStatus function to handle shutdown delay or missing readyz endpoint

Signed-off-by: huangyanfeng [email protected]
Signed-off-by: huangyanfeng <[email protected]>

Author: yanfeng1992

pkg/controllers/status/cluster_status_controller.go

@@ -435,8 +435,10 @@ func (c *ClusterStatusController) initLeaseController(cluster *clusterv1alpha1.C
 
 func getClusterHealthStatus(clusterClient *util.ClusterClient) (online, healthy bool) {
 	healthStatus, err := healthEndpointCheck(clusterClient.KubeClient, "/readyz")
-	if err != nil && healthStatus == http.StatusNotFound {
-		// do health check with healthz endpoint if the readyz endpoint is not installed in member cluster
+	if err != nil {
+		// do health check with healthz endpoint in two cases:
+		// 1. When the server is configured with --shutdown-delay-duration, /readyz returns failure but /healthz still serves success
+		// 2. When the readyz endpoint is not installed in member cluster
 		healthStatus, err = healthEndpointCheck(clusterClient.KubeClient, "/healthz")
 	}