fix cpu resource metric type by changing to counter

dashpole · dashpole · commit 203b1ddfdf75 · 2020-03-27T17:09:43.000-07:00
diff --git a/pkg/kubelet/metrics/collectors/resource_metrics.go b/pkg/kubelet/metrics/collectors/resource_metrics.go
@@ -26,7 +26,7 @@ import (
 )
 
 var (
-	nodeCPUUsageDesc = metrics.NewDesc("node_cpu_usage_seconds",
+	nodeCPUUsageDesc = metrics.NewDesc("node_cpu_usage_seconds_total",
 		"Cumulative cpu time consumed by the node in core-seconds",
 		nil,
 		nil,
@@ -40,7 +40,7 @@ var (
 		metrics.ALPHA,
 		"")
 
-	containerCPUUsageDesc = metrics.NewDesc("container_cpu_usage_seconds",
+	containerCPUUsageDesc = metrics.NewDesc("container_cpu_usage_seconds_total",
 		"Cumulative cpu time consumed by the container in core-seconds",
 		[]string{"container", "pod", "namespace"},
 		nil,
@@ -120,7 +120,7 @@ func (rc *resourceMetricsCollector) collectNodeCPUMetrics(ch chan<- metrics.Metr
 	}
 
 	ch <- metrics.NewLazyMetricWithTimestamp(s.CPU.Time.Time,
-		metrics.NewLazyConstMetric(nodeCPUUsageDesc, metrics.GaugeValue, float64(*s.CPU.UsageCoreNanoSeconds)/float64(time.Second)))
+		metrics.NewLazyConstMetric(nodeCPUUsageDesc, metrics.CounterValue, float64(*s.CPU.UsageCoreNanoSeconds)/float64(time.Second)))
 }
 
 func (rc *resourceMetricsCollector) collectNodeMemoryMetrics(ch chan<- metrics.Metric, s summary.NodeStats) {
@@ -138,7 +138,7 @@ func (rc *resourceMetricsCollector) collectContainerCPUMetrics(ch chan<- metrics
 	}
 
 	ch <- metrics.NewLazyMetricWithTimestamp(s.CPU.Time.Time,
-		metrics.NewLazyConstMetric(containerCPUUsageDesc, metrics.GaugeValue,
+		metrics.NewLazyConstMetric(containerCPUUsageDesc, metrics.CounterValue,
 			float64(*s.CPU.UsageCoreNanoSeconds)/float64(time.Second), s.Name, pod.PodRef.Name, pod.PodRef.Namespace))
 }
 
diff --git a/pkg/kubelet/metrics/collectors/resource_metrics_test.go b/pkg/kubelet/metrics/collectors/resource_metrics_test.go
@@ -47,9 +47,9 @@ func TestCollectResourceMetrics(t *testing.T) {
 	testTime := metav1.NewTime(time.Unix(2, 0)) // a static timestamp: 2000
 	interestedMetrics := []string{
 		"scrape_error",
-		"node_cpu_usage_seconds",
+		"node_cpu_usage_seconds_total",
 		"node_memory_working_set_bytes",
-		"container_cpu_usage_seconds",
+		"container_cpu_usage_seconds_total",
 		"container_memory_working_set_bytes",
 	}
 
@@ -85,9 +85,9 @@ func TestCollectResourceMetrics(t *testing.T) {
 			},
 			summaryErr: nil,
 			expectedMetrics: `
-				# HELP node_cpu_usage_seconds [ALPHA] Cumulative cpu time consumed by the node in core-seconds
-				# TYPE node_cpu_usage_seconds gauge
-				node_cpu_usage_seconds 10 2000
+				# HELP node_cpu_usage_seconds_total [ALPHA] Cumulative cpu time consumed by the node in core-seconds
+				# TYPE node_cpu_usage_seconds_total counter
+				node_cpu_usage_seconds_total 10 2000
 				# HELP node_memory_working_set_bytes [ALPHA] Current working set of the node in bytes
 				# TYPE node_memory_working_set_bytes gauge
 				node_memory_working_set_bytes 1000 2000
@@ -156,11 +156,11 @@ func TestCollectResourceMetrics(t *testing.T) {
 				# HELP scrape_error [ALPHA] 1 if there was an error while getting container metrics, 0 otherwise
 				# TYPE scrape_error gauge
 				scrape_error 0
-				# HELP container_cpu_usage_seconds [ALPHA] Cumulative cpu time consumed by the container in core-seconds
-				# TYPE container_cpu_usage_seconds gauge
-				container_cpu_usage_seconds{container="container_a",namespace="namespace_a",pod="pod_a"} 10 2000
-				container_cpu_usage_seconds{container="container_a",namespace="namespace_b",pod="pod_b"} 10 2000
-				container_cpu_usage_seconds{container="container_b",namespace="namespace_a",pod="pod_a"} 10 2000
+				# HELP container_cpu_usage_seconds_total [ALPHA] Cumulative cpu time consumed by the container in core-seconds
+				# TYPE container_cpu_usage_seconds_total counter
+				container_cpu_usage_seconds_total{container="container_a",namespace="namespace_a",pod="pod_a"} 10 2000
+				container_cpu_usage_seconds_total{container="container_a",namespace="namespace_b",pod="pod_b"} 10 2000
+				container_cpu_usage_seconds_total{container="container_b",namespace="namespace_a",pod="pod_a"} 10 2000
 				# HELP container_memory_working_set_bytes [ALPHA] Current working set of the container in bytes
 				# TYPE container_memory_working_set_bytes gauge
 				container_memory_working_set_bytes{container="container_a",namespace="namespace_a",pod="pod_a"} 1000 2000

Original file line number	Diff line number	Diff line change
`@@ -26,7 +26,7 @@ import (`
`26`	`26`	`)`
`27`	`27`
`28`	`28`	`var (`
`29`		`- nodeCPUUsageDesc = metrics.NewDesc("node_cpu_usage_seconds",`
	`29`	`+ nodeCPUUsageDesc = metrics.NewDesc("node_cpu_usage_seconds_total",`
`30`	`30`	`"Cumulative cpu time consumed by the node in core-seconds",`
`31`	`31`	`nil,`
`32`	`32`	`nil,`
`@@ -40,7 +40,7 @@ var (`
`40`	`40`	`metrics.ALPHA,`
`41`	`41`	`"")`
`42`	`42`
`43`		`- containerCPUUsageDesc = metrics.NewDesc("container_cpu_usage_seconds",`
	`43`	`+ containerCPUUsageDesc = metrics.NewDesc("container_cpu_usage_seconds_total",`
`44`	`44`	`"Cumulative cpu time consumed by the container in core-seconds",`
`45`	`45`	`[]string{"container", "pod", "namespace"},`
`46`	`46`	`nil,`
`@@ -120,7 +120,7 @@ func (rc *resourceMetricsCollector) collectNodeCPUMetrics(ch chan<- metrics.Metr`
`120`	`120`	`}`
`121`	`121`
`122`	`122`	`ch <- metrics.NewLazyMetricWithTimestamp(s.CPU.Time.Time,`
`123`		`- metrics.NewLazyConstMetric(nodeCPUUsageDesc, metrics.GaugeValue, float64(*s.CPU.UsageCoreNanoSeconds)/float64(time.Second)))`
	`123`	`+ metrics.NewLazyConstMetric(nodeCPUUsageDesc, metrics.CounterValue, float64(*s.CPU.UsageCoreNanoSeconds)/float64(time.Second)))`
`124`	`124`	`}`
`125`	`125`
`126`	`126`	`func (rc *resourceMetricsCollector) collectNodeMemoryMetrics(ch chan<- metrics.Metric, s summary.NodeStats) {`
`@@ -138,7 +138,7 @@ func (rc *resourceMetricsCollector) collectContainerCPUMetrics(ch chan<- metrics`
`138`	`138`	`}`
`139`	`139`
`140`	`140`	`ch <- metrics.NewLazyMetricWithTimestamp(s.CPU.Time.Time,`
`141`		`- metrics.NewLazyConstMetric(containerCPUUsageDesc, metrics.GaugeValue,`
	`141`	`+ metrics.NewLazyConstMetric(containerCPUUsageDesc, metrics.CounterValue,`
`142`	`142`	`float64(*s.CPU.UsageCoreNanoSeconds)/float64(time.Second), s.Name, pod.PodRef.Name, pod.PodRef.Namespace))`
`143`	`143`	`}`
`144`	`144`