Merge pull request #72470 from danielqsj/kl

pkg/kubelet/cm/cgroup_manager_linux.go

@@ -284,7 +284,8 @@ func (m *cgroupManagerImpl) Exists(name CgroupName) bool {
 func (m *cgroupManagerImpl) Destroy(cgroupConfig *CgroupConfig) error {
 	start := time.Now()
 	defer func() {
-		metrics.CgroupManagerLatency.WithLabelValues("destroy").Observe(metrics.SinceInMicroseconds(start))
+		metrics.CgroupManagerDuration.WithLabelValues("destroy").Observe(metrics.SinceInSeconds(start))
+		metrics.DeprecatedCgroupManagerLatency.WithLabelValues("destroy").Observe(metrics.SinceInMicroseconds(start))
 	}()
 
 	cgroupPaths := m.buildCgroupPaths(cgroupConfig.Name)
@@ -411,7 +412,8 @@ func (m *cgroupManagerImpl) toResources(resourceConfig *ResourceConfig) *libcont
 func (m *cgroupManagerImpl) Update(cgroupConfig *CgroupConfig) error {
 	start := time.Now()
 	defer func() {
-		metrics.CgroupManagerLatency.WithLabelValues("update").Observe(metrics.SinceInMicroseconds(start))
+		metrics.CgroupManagerDuration.WithLabelValues("update").Observe(metrics.SinceInSeconds(start))
+		metrics.DeprecatedCgroupManagerLatency.WithLabelValues("update").Observe(metrics.SinceInMicroseconds(start))
 	}()
 
 	// Extract the cgroup resource parameters
@@ -446,7 +448,8 @@ func (m *cgroupManagerImpl) Update(cgroupConfig *CgroupConfig) error {
 func (m *cgroupManagerImpl) Create(cgroupConfig *CgroupConfig) error {
 	start := time.Now()
 	defer func() {
-		metrics.CgroupManagerLatency.WithLabelValues("create").Observe(metrics.SinceInMicroseconds(start))
+		metrics.CgroupManagerDuration.WithLabelValues("create").Observe(metrics.SinceInSeconds(start))
+		metrics.DeprecatedCgroupManagerLatency.WithLabelValues("create").Observe(metrics.SinceInMicroseconds(start))
 	}()
 
 	resources := m.toResources(cgroupConfig.ResourceParameters)

pkg/kubelet/cm/devicemanager/manager.go

@@ -354,6 +354,7 @@ func (m *ManagerImpl) Allocate(node *schedulernodeinfo.NodeInfo, attrs *lifecycl
 func (m *ManagerImpl) Register(ctx context.Context, r *pluginapi.RegisterRequest) (*pluginapi.Empty, error) {
 	klog.Infof("Got registration request from device plugin with resource name %q", r.ResourceName)
 	metrics.DevicePluginRegistrationCount.WithLabelValues(r.ResourceName).Inc()
+	metrics.DeprecatedDevicePluginRegistrationCount.WithLabelValues(r.ResourceName).Inc()
 	var versionCompatible bool
 	for _, v := range pluginapi.SupportedVersions {
 		if r.Version == v {
@@ -696,7 +697,8 @@ func (m *ManagerImpl) allocateContainerResources(pod *v1.Pod, container *v1.Cont
 		// in a passed in AllocateRequest pointer, and issues a single Allocate call per pod.
 		klog.V(3).Infof("Making allocation request for devices %v for device plugin %s", devs, resource)
 		resp, err := eI.e.allocate(devs)
-		metrics.DevicePluginAllocationLatency.WithLabelValues(resource).Observe(metrics.SinceInMicroseconds(startRPCTime))
+		metrics.DevicePluginAllocationDuration.WithLabelValues(resource).Observe(metrics.SinceInSeconds(startRPCTime))
+		metrics.DeprecatedDevicePluginAllocationLatency.WithLabelValues(resource).Observe(metrics.SinceInMicroseconds(startRPCTime))
 		if err != nil {
 			// In case of allocation failure, we want to restore m.allocatedDevices
 			// to the actual allocated state from m.podDevices.

pkg/kubelet/eviction/eviction_manager.go

@@ -361,7 +361,8 @@ func (m *managerImpl) synchronize(diskInfoProvider DiskInfoProvider, podFunc Act
 	for _, t := range thresholds {
 		timeObserved := observations[t.Signal].time
 		if !timeObserved.IsZero() {
-			metrics.EvictionStatsAge.WithLabelValues(string(t.Signal)).Observe(metrics.SinceInMicroseconds(timeObserved.Time))
+			metrics.EvictionStatsAge.WithLabelValues(string(t.Signal)).Observe(metrics.SinceInSeconds(timeObserved.Time))
+			metrics.DeprecatedEvictionStatsAge.WithLabelValues(string(t.Signal)).Observe(metrics.SinceInMicroseconds(timeObserved.Time))
 		}
 	}
 

pkg/kubelet/kubelet.go

@@ -1500,7 +1500,8 @@ func (kl *Kubelet) syncPod(o syncPodOptions) error {
 		if !firstSeenTime.IsZero() {
 			// This is the first time we are syncing the pod. Record the latency
 			// since kubelet first saw the pod if firstSeenTime is set.
-			metrics.PodWorkerStartLatency.Observe(metrics.SinceInMicroseconds(firstSeenTime))
+			metrics.PodWorkerStartDuration.Observe(metrics.SinceInSeconds(firstSeenTime))
+			metrics.DeprecatedPodWorkerStartLatency.Observe(metrics.SinceInMicroseconds(firstSeenTime))
 		} else {
 			klog.V(3).Infof("First seen time not recorded for pod %q", pod.UID)
 		}
@@ -1517,7 +1518,8 @@ func (kl *Kubelet) syncPod(o syncPodOptions) error {
 	existingStatus, ok := kl.statusManager.GetPodStatus(pod.UID)
 	if !ok || existingStatus.Phase == v1.PodPending && apiPodStatus.Phase == v1.PodRunning &&
 		!firstSeenTime.IsZero() {
-		metrics.PodStartLatency.Observe(metrics.SinceInMicroseconds(firstSeenTime))
+		metrics.PodStartDuration.Observe(metrics.SinceInSeconds(firstSeenTime))
+		metrics.DeprecatedPodStartLatency.Observe(metrics.SinceInMicroseconds(firstSeenTime))
 	}
 
 	runnable := kl.canRunPod(pod)
@@ -1996,7 +1998,8 @@ func (kl *Kubelet) dispatchWork(pod *v1.Pod, syncType kubetypes.SyncPodType, mir
 		UpdateType: syncType,
 		OnCompleteFunc: func(err error) {
 			if err != nil {
-				metrics.PodWorkerLatency.WithLabelValues(syncType.String()).Observe(metrics.SinceInMicroseconds(start))
+				metrics.PodWorkerDuration.WithLabelValues(syncType.String()).Observe(metrics.SinceInSeconds(start))
+				metrics.DeprecatedPodWorkerLatency.WithLabelValues(syncType.String()).Observe(metrics.SinceInMicroseconds(start))
 			}
 		},
 	})

pkg/kubelet/kuberuntime/instrumented_services.go

@@ -49,13 +49,16 @@ func newInstrumentedImageManagerService(service internalapi.ImageManagerService)
 // recordOperation records the duration of the operation.
 func recordOperation(operation string, start time.Time) {
 	metrics.RuntimeOperations.WithLabelValues(operation).Inc()
-	metrics.RuntimeOperationsLatency.WithLabelValues(operation).Observe(metrics.SinceInMicroseconds(start))
+	metrics.DeprecatedRuntimeOperations.WithLabelValues(operation).Inc()
+	metrics.RuntimeOperationsDuration.WithLabelValues(operation).Observe(metrics.SinceInSeconds(start))
+	metrics.DeprecatedRuntimeOperationsLatency.WithLabelValues(operation).Observe(metrics.SinceInMicroseconds(start))
 }
 
 // recordError records error for metric if an error occurred.
 func recordError(operation string, err error) {
 	if err != nil {
 		metrics.RuntimeOperationsErrors.WithLabelValues(operation).Inc()
+		metrics.DeprecatedRuntimeOperationsErrors.WithLabelValues(operation).Inc()
 	}
 }
 

pkg/kubelet/kuberuntime/instrumented_services_test.go

@@ -30,7 +30,7 @@ import (
 
 func TestRecordOperation(t *testing.T) {
 	prometheus.MustRegister(metrics.RuntimeOperations)
-	prometheus.MustRegister(metrics.RuntimeOperationsLatency)
+	prometheus.MustRegister(metrics.RuntimeOperationsDuration)
 	prometheus.MustRegister(metrics.RuntimeOperationsErrors)
 
 	temporalServer := "127.0.0.1:1234"
@@ -50,8 +50,8 @@ func TestRecordOperation(t *testing.T) {
 	}()
 
 	recordOperation("create_container", time.Now())
-	runtimeOperationsCounterExpected := "kubelet_runtime_operations{operation_type=\"create_container\"} 1"
-	runtimeOperationsLatencyExpected := "kubelet_runtime_operations_latency_microseconds_count{operation_type=\"create_container\"} 1"
+	runtimeOperationsCounterExpected := "kubelet_runtime_operations_total{operation_type=\"create_container\"} 1"
+	runtimeOperationsDurationExpected := "kubelet_runtime_operations_duration_seconds_count{operation_type=\"create_container\"} 1"
 
 	assert.HTTPBodyContains(t, http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		mux.ServeHTTP(w, r)
@@ -59,7 +59,7 @@ func TestRecordOperation(t *testing.T) {
 
 	assert.HTTPBodyContains(t, http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		mux.ServeHTTP(w, r)
-	}), "GET", prometheusURL, nil, runtimeOperationsLatencyExpected)
+	}), "GET", prometheusURL, nil, runtimeOperationsDurationExpected)
 }
 
 func TestInstrumentedVersion(t *testing.T) {

pkg/kubelet/pleg/generic.go

@@ -189,12 +189,14 @@ func (g *GenericPLEG) relist() {
 	klog.V(5).Infof("GenericPLEG: Relisting")
 
 	if lastRelistTime := g.getRelistTime(); !lastRelistTime.IsZero() {
-		metrics.PLEGRelistInterval.Observe(metrics.SinceInMicroseconds(lastRelistTime))
+		metrics.PLEGRelistInterval.Observe(metrics.SinceInSeconds(lastRelistTime))
+		metrics.DeprecatedPLEGRelistInterval.Observe(metrics.SinceInMicroseconds(lastRelistTime))
 	}
 
 	timestamp := g.clock.Now()
 	defer func() {
-		metrics.PLEGRelistLatency.Observe(metrics.SinceInMicroseconds(timestamp))
+		metrics.PLEGRelistDuration.Observe(metrics.SinceInSeconds(timestamp))
+		metrics.DeprecatedPLEGRelistLatency.Observe(metrics.SinceInMicroseconds(timestamp))
 	}()
 
 	// Get all the pods.

test/e2e/framework/kubelet_stats.go

@@ -102,13 +102,13 @@ func getKubeletMetrics(c clientset.Interface, nodeName string) (metrics.KubeletM
 // Note that the KubeletMetrics passed in should not contain subsystem prefix.
 func GetDefaultKubeletLatencyMetrics(ms metrics.KubeletMetrics) KubeletLatencyMetrics {
 	latencyMetricNames := sets.NewString(
-		kubeletmetrics.PodWorkerLatencyKey,
-		kubeletmetrics.PodWorkerStartLatencyKey,
-		kubeletmetrics.PodStartLatencyKey,
+		kubeletmetrics.PodWorkerDurationKey,
+		kubeletmetrics.PodWorkerStartDurationKey,
+		kubeletmetrics.PodStartDurationKey,
 		kubeletmetrics.CgroupManagerOperationsKey,
 		dockermetrics.DockerOperationsLatencyKey,
-		kubeletmetrics.PodWorkerStartLatencyKey,
-		kubeletmetrics.PLEGRelistLatencyKey,
+		kubeletmetrics.PodWorkerStartDurationKey,
+		kubeletmetrics.PLEGRelistDurationKey,
 	)
 	return GetKubeletLatencyMetrics(ms, latencyMetricNames)
 }

test/e2e/framework/metrics_util.go

@@ -168,9 +168,9 @@ var InterestingKubeletMetrics = []string{
 	"kubelet_docker_errors",
 	"kubelet_docker_operations_latency_seconds",
 	"kubelet_generate_pod_status_latency_microseconds",
-	"kubelet_pod_start_latency_microseconds",
-	"kubelet_pod_worker_latency_microseconds",
-	"kubelet_pod_worker_start_latency_microseconds",
+	"kubelet_pod_start_duration_seconds",
+	"kubelet_pod_worker_duration_seconds",
+	"kubelet_pod_worker_start_duration_seconds",
 	"kubelet_sync_pods_latency_microseconds",
 }
 

test/e2e_node/density_test.go

@@ -459,12 +459,12 @@ func getPodStartLatency(node string) (framework.KubeletLatencyMetrics, error) {
 
 	for _, samples := range ms {
 		for _, sample := range samples {
-			if sample.Metric["__name__"] == kubemetrics.KubeletSubsystem+"_"+kubemetrics.PodStartLatencyKey {
+			if sample.Metric["__name__"] == kubemetrics.KubeletSubsystem+"_"+kubemetrics.PodStartDurationKey {
 				quantile, _ := strconv.ParseFloat(string(sample.Metric["quantile"]), 64)
 				latencyMetrics = append(latencyMetrics,
 					framework.KubeletLatencyMetric{
 						Quantile: quantile,
-						Method:   kubemetrics.PodStartLatencyKey,
+						Method:   kubemetrics.PodStartDurationKey,
 						Latency:  time.Duration(int(sample.Value)) * time.Microsecond})
 			}
 		}

test/e2e_node/gpu_device_plugin.go

@@ -156,7 +156,7 @@ func logDevicePluginMetrics() {
 	framework.ExpectNoError(err)
 	for msKey, samples := range ms {
 		switch msKey {
-		case kubeletmetrics.KubeletSubsystem + "_" + kubeletmetrics.DevicePluginAllocationLatencyKey:
+		case kubeletmetrics.KubeletSubsystem + "_" + kubeletmetrics.DevicePluginAllocationDurationKey:
 			for _, sample := range samples {
 				latency := sample.Value
 				resource := string(sample.Metric["resource_name"])