metrics: improve the node updates gauge

Rename the metric, better describe what we're measuring and better
comply with prometheus naming conventions. Also change it to represent
actual updates of the node object on the Kubernetes apiserver.

pkg/nfd-master/metrics.go

@@ -29,7 +29,7 @@ import (
 // When adding metric names, see https://prometheus.io/docs/practices/naming/#metric-names
 const (
 	buildInfoQuery         = "nfd_master_build_info"
-	updatedNodesQuery      = "nfd_updated_nodes"
+	nodeUpdatesQuery       = "nfd_node_updates_total"
 	nfrProcessingTimeQuery = "nfd_nodefeaturerule_processing_duration_seconds"
 )
 
@@ -43,8 +43,8 @@ var (
 			"version": version.Get(),
 		},
 	})
-	updatedNodes = prometheus.NewCounter(prometheus.CounterOpts{
-		Name: updatedNodesQuery,
+	nodeUpdates = prometheus.NewCounter(prometheus.CounterOpts{
+		Name: nodeUpdatesQuery,
 		Help: "Number of nodes updated by the master.",
 	})
 	nfrProcessingTime = prometheus.NewHistogramVec(
@@ -69,7 +69,7 @@ func registerVersion(version string) {
 func runMetricsServer(port int) {
 	r := prometheus.NewRegistry()
 	r.MustRegister(buildInfo,
-		updatedNodes,
+		nodeUpdates,
 		nfrProcessingTime)
 
 	mux := http.NewServeMux()

pkg/nfd-master/nfd-master.go

@@ -854,7 +854,6 @@ func (m *nfdMaster) refreshNodeFeatures(cli *kubernetes.Clientset, nodeName stri
 		return err
 	}
 
-	updatedNodes.Inc()
 	return nil
 }
 
@@ -1073,6 +1072,7 @@ func (m *nfdMaster) updateNodeObject(cli *kubernetes.Clientset, nodeName string,
 	}
 
 	if len(patches) > 0 || len(statusPatches) > 0 {
+		nodeUpdates.Inc()
 		klog.InfoS("node updated", "nodeName", nodeName)
 	} else {
 		klog.V(1).InfoS("no updates to node", "nodeName", nodeName)