temperary remove profileStatusPerNode until otel support reset metric

bytedance · Nov 13, 2024 · 6385f04 · 6385f04
1 parent 1f358f6
commit 6385f04
Show file tree

Hide file tree

Showing 5 changed files with 55 additions and 55 deletions.
diff --git a/cmd/varmor/main.go b/cmd/varmor/main.go
@@ -65,10 +65,10 @@ var (
 	webhookMatchLabel        string
 	bpfExclusiveMode         bool
 	enableMetrics            bool
-	syncMetricsSecond        int
-	statusUpdateCycle        time.Duration
-	auditLogPaths            string
-	setupLog                 = log.Log.WithName("SETUP")
+	//syncMetricsSecond        int
+	statusUpdateCycle time.Duration
+	auditLogPaths     string
+	setupLog          = log.Log.WithName("SETUP")
 )
 
 func main() {
@@ -92,7 +92,7 @@ func main() {
 	flag.DurationVar(&statusUpdateCycle, "statusUpdateCycle", time.Hour*2, "Configure the status update cycle for VarmorPolicy and ArmorProfile")
 	flag.StringVar(&auditLogPaths, "auditLogPaths", "/var/log/audit/audit.log|/var/log/kern.log", "Configure the file search list to select the audit log file and read the AppArmor and Seccomp audit events. Please use a vertical bar to separate the file paths, the first valid file will be used to track the audit events.")
 	flag.BoolVar(&enableMetrics, "enableMetrics", false, "Set this flag to enable metrics.")
-	flag.IntVar(&syncMetricsSecond, "syncMetricsSecond", 10, "Configure the profile metric update seconds")
+	//flag.IntVar(&syncMetricsSecond, "syncMetricsSecond", 10, "Configure the profile metric update seconds")
 	flag.Parse()
 
 	// Set the webhook matchLabels configuration.
@@ -150,7 +150,7 @@ func main() {
 	}
 
 	// init a metrics
-	metricsModule := metrics.NewMetricsModule(log.Log.WithName("METRICS"), enableMetrics, syncMetricsSecond)
+	metricsModule := metrics.NewMetricsModule(log.Log.WithName("METRICS"), enableMetrics, 10)
 
 	if agent {
 		setupLog.Info("vArmor agent startup")

diff --git a/internal/status/api/v1/manager.go b/internal/status/api/v1/manager.go
@@ -52,21 +52,21 @@ type StatusManager struct {
 	PolicyStatuses map[string]varmortypes.PolicyStatus
 	// Use "namespace/VarmorPolicyName" as key. One VarmorPolicy object corresponds to one ModelingStatus
 	// TODO: Rebuild ModelingStatuses from ArmorProfile object when leader change occurs.
-	ModelingStatuses     map[string]varmortypes.ModelingStatus
-	ResetCh              chan string
-	DeleteCh             chan string
-	UpdateStatusCh       chan string
-	UpdateModeCh         chan string
-	statusQueue          workqueue.RateLimitingInterface
-	dataQueue            workqueue.RateLimitingInterface
-	statusUpdateCycle    time.Duration
-	debug                bool
-	log                  logr.Logger
-	metricsModule        *varmormetrics.MetricsModule
-	profileSuccess       metric.Float64Counter
-	profileFailure       metric.Float64Counter
-	profileChangeCount   metric.Float64Counter
-	profileStatusPerNode metric.Float64Gauge
+	ModelingStatuses   map[string]varmortypes.ModelingStatus
+	ResetCh            chan string
+	DeleteCh           chan string
+	UpdateStatusCh     chan string
+	UpdateModeCh       chan string
+	statusQueue        workqueue.RateLimitingInterface
+	dataQueue          workqueue.RateLimitingInterface
+	statusUpdateCycle  time.Duration
+	debug              bool
+	log                logr.Logger
+	metricsModule      *varmormetrics.MetricsModule
+	profileSuccess     metric.Float64Counter
+	profileFailure     metric.Float64Counter
+	profileChangeCount metric.Float64Counter
+	//profileStatusPerNode metric.Float64Gauge
 }
 
 func NewStatusManager(coreInterface corev1.CoreV1Interface,
@@ -99,7 +99,7 @@ func NewStatusManager(coreInterface corev1.CoreV1Interface,
 		m.profileSuccess = metricsModule.RegisterFloat64Counter("profile_processing_success", "Number of successful profile processing")
 		m.profileFailure = metricsModule.RegisterFloat64Counter("profile_processing_failure", "Number of failed profile processing")
 		m.profileChangeCount = metricsModule.RegisterFloat64Counter("profile_change_count", "Number of profile change")
-		m.profileStatusPerNode = metricsModule.RegisterFloat64Gauge("profile_status_per_node", "Profile status per node (1=success, 0=failure)")
+		//m.profileStatusPerNode = metricsModule.RegisterFloat64Gauge("profile_status_per_node", "Profile status per node (1=success, 0=failure)")
 	}
 
 	return &m
@@ -653,7 +653,7 @@ func (m *StatusManager) Run(stopCh <-chan struct{}) {
 	go m.reconcileStatus(stopCh)
 	go wait.Until(m.statusWorker, time.Second, stopCh)
 	go wait.Until(m.dataWorker, time.Second, stopCh)
-	go m.syncStatusMetricsLoop()
+	//go m.syncStatusMetricsLoop()
 	<-stopCh
 }
 

diff --git a/internal/status/api/v1/status.go b/internal/status/api/v1/status.go
@@ -88,36 +88,37 @@ func (m *StatusManager) HandleProfileStatusUpdate(status varmortypes.ProfileStat
 	m.profileChangeCount.Add(ctx, 1, metric.WithAttributeSet(attrSet))
 }
 
-func (m *StatusManager) syncStatusMetricsLoop() {
-	ctx := context.Background()
-	for {
-		time.Sleep(time.Duration(m.metricsModule.Refresh) * time.Second)
-		logger := m.log.WithName("syncStatusMetricsLoop()")
-		logger.Info("start syncing status metrics")
-		m.profileStatusPerNode = m.metricsModule.RegisterFloat64Gauge("profile_status_per_node", "Profile status per node (1=success, 0=failure)")
-		for key, status := range m.PolicyStatuses {
-			namespace, name, err := PolicyStatusKeyGetInfo(key)
-			if err != nil {
-				logger.Error(err, "PolicyStatusKeyGetInfo()")
-				continue
-			}
-			for nodeName, nodeMessage := range status.NodeMessages {
-				labels := []attribute.KeyValue{
-					attribute.String("namespace", namespace),
-					attribute.String("profile_name", name),
-					attribute.String("node_name", nodeName),
-					attribute.Int64("timestamp", time.Now().Unix()),
-				}
-				attrSet := attribute.NewSet(labels...)
-				if nodeMessage == string(varmortypes.ArmorProfileReady) {
-					m.profileStatusPerNode.Record(ctx, 1, metric.WithAttributeSet(attrSet)) // 1 mean success
-				} else {
-					m.profileStatusPerNode.Record(ctx, 0, metric.WithAttributeSet(attrSet)) // 0 mean failure
-				}
-			}
-		}
-	}
-}
+// disable syncStatusMetricsLoop until otel support clear metrics
+//func (m *StatusManager) syncStatusMetricsLoop() {
+//	ctx := context.Background()
+//	for {
+//		time.Sleep(time.Duration(m.metricsModule.Refresh) * time.Second)
+//		logger := m.log.WithName("syncStatusMetricsLoop()")
+//		logger.Info("start syncing status metrics")
+//		m.profileStatusPerNode = m.metricsModule.RegisterFloat64Gauge("profile_status_per_node", "Profile status per node (1=success, 0=failure)")
+//		for key, status := range m.PolicyStatuses {
+//			namespace, name, err := PolicyStatusKeyGetInfo(key)
+//			if err != nil {
+//				logger.Error(err, "PolicyStatusKeyGetInfo()")
+//				continue
+//			}
+//			for nodeName, nodeMessage := range status.NodeMessages {
+//				labels := []attribute.KeyValue{
+//					attribute.String("namespace", namespace),
+//					attribute.String("profile_name", name),
+//					attribute.String("node_name", nodeName),
+//					attribute.Int64("timestamp", time.Now().Unix()),
+//				}
+//				attrSet := attribute.NewSet(labels...)
+//				if nodeMessage == string(varmortypes.ArmorProfileReady) {
+//					m.profileStatusPerNode.Record(ctx, 1, metric.WithAttributeSet(attrSet)) // 1 mean success
+//				} else {
+//					m.profileStatusPerNode.Record(ctx, 0, metric.WithAttributeSet(attrSet)) // 0 mean failure
+//				}
+//			}
+//		}
+//	}
+//}
 
 // updatePolicyStatus update StatusManager.PolicyStatuses[statusKey] with profileStatus which comes from agent.
 func (m *StatusManager) updatePolicyStatus(statusKey string, profileStatus *varmortypes.ProfileStatus) error {

diff --git a/manifests/varmor/templates/deployments/manager.yaml b/manifests/varmor/templates/deployments/manager.yaml
@@ -60,7 +60,6 @@ spec:
         {{- if .Values.metrics.enabled }}
         {{- with .Values.manager.metrics.args }}
           {{- toYaml . | nindent 8 }}
-        - --syncMetricsSecond={{ $.Values.metrics.syncMetricsSecond }}
         {{- end }}
         {{- end }}
         {{- end }}

diff --git a/manifests/varmor/values.yaml b/manifests/varmor/values.yaml
@@ -18,7 +18,7 @@ bpfLsmEnforcer:
 metrics:
   enabled: false
   serviceMonitorEnabled: false
-  syncMetricsSecond: 10
+#  syncMetricsSecond: 10
 
 restartExistWorkloads:
   enabled: true