Add system CPU metrics to metrics-v3 (#19560)

endpoint: /minio/metrics/v3/system/cpu

metrics:
- minio_system_cpu_avg_idle
- minio_system_cpu_avg_iowait
- minio_system_cpu_load
- minio_system_cpu_load_perc
- minio_system_cpu_nice
- minio_system_cpu_steal
- minio_system_cpu_system
- minio_system_cpu_user

cmd/metrics-resource.go

@@ -162,14 +162,7 @@ func init() {
 	resourceCollector = newMinioResourceCollector(resourceMetricsGroups)
 }
 
-func updateResourceMetrics(subSys MetricSubsystem, name MetricName, val float64, labels map[string]string, isCumulative bool) {
-	resourceMetricsMapMu.Lock()
-	defer resourceMetricsMapMu.Unlock()
-	subsysMetrics, found := resourceMetricsMap[subSys]
-	if !found {
-		subsysMetrics = ResourceMetrics{}
-	}
-
+func getResourceKey(name MetricName, labels map[string]string) string {
 	// labels are used to uniquely identify a metric
 	// e.g. reads_per_sec_{drive} inside the map
 	sfx := ""
@@ -180,7 +173,18 @@ func updateResourceMetrics(subSys MetricSubsystem, name MetricName, val float64,
 		sfx += v
 	}
 
-	key := string(name) + "_" + sfx
+	return string(name) + "_" + sfx
+}
+
+func updateResourceMetrics(subSys MetricSubsystem, name MetricName, val float64, labels map[string]string, isCumulative bool) {
+	resourceMetricsMapMu.Lock()
+	defer resourceMetricsMapMu.Unlock()
+	subsysMetrics, found := resourceMetricsMap[subSys]
+	if !found {
+		subsysMetrics = ResourceMetrics{}
+	}
+
+	key := getResourceKey(name, labels)
 	metric, found := subsysMetrics[key]
 	if !found {
 		metric = ResourceMetric{

cmd/metrics-v3-cache.go

@@ -35,6 +35,7 @@ type metricsCache struct {
 	esetHealthResult    *cachevalue.Cache[HealthResult]
 	driveMetrics        *cachevalue.Cache[storageMetrics]
 	memoryMetrics       *cachevalue.Cache[madmin.MemInfo]
+	cpuMetrics          *cachevalue.Cache[madmin.CPUMetrics]
 	clusterDriveMetrics *cachevalue.Cache[storageMetrics]
 	nodesUpDown         *cachevalue.Cache[nodesOnline]
 }
@@ -45,6 +46,7 @@ func newMetricsCache() *metricsCache {
 		esetHealthResult:    newESetHealthResultCache(),
 		driveMetrics:        newDriveMetricsCache(),
 		memoryMetrics:       newMemoryMetricsCache(),
+		cpuMetrics:          newCPUMetricsCache(),
 		clusterDriveMetrics: newClusterStorageInfoCache(),
 		nodesUpDown:         newNodesUpDownCache(),
 	}
@@ -200,6 +202,31 @@ func newDriveMetricsCache() *cachevalue.Cache[storageMetrics] {
 		loadDriveMetrics)
 }
 
+func newCPUMetricsCache() *cachevalue.Cache[madmin.CPUMetrics] {
+	loadCPUMetrics := func() (v madmin.CPUMetrics, err error) {
+		var types madmin.MetricType = madmin.MetricsCPU
+
+		m := collectLocalMetrics(types, collectMetricsOpts{
+			hosts: map[string]struct{}{
+				globalLocalNodeName: {},
+			},
+		})
+
+		for _, hm := range m.ByHost {
+			if hm.CPU != nil {
+				v = *hm.CPU
+				break
+			}
+		}
+
+		return
+	}
+
+	return cachevalue.NewFromFunc(1*time.Minute,
+		cachevalue.Opts{ReturnLastGood: true},
+		loadCPUMetrics)
+}
+
 func newMemoryMetricsCache() *cachevalue.Cache[madmin.MemInfo] {
 	loadMemoryMetrics := func() (v madmin.MemInfo, err error) {
 		var types madmin.MetricType = madmin.MetricsMem

cmd/metrics-v3-system-cpu.go

@@ -0,0 +1,82 @@
+// Copyright (c) 2015-2024 MinIO, Inc.
+//
+// # This file is part of MinIO Object Storage stack
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+package cmd
+
+import (
+	"context"
+	"math"
+)
+
+const (
+	sysCPUAvgIdle   = "avg_idle"
+	sysCPUAvgIOWait = "avg_iowait"
+	sysCPULoad      = "load"
+	sysCPULoadPerc  = "load_perc"
+	sysCPUNice      = "nice"
+	sysCPUSteal     = "steal"
+	sysCPUSystem    = "system"
+	sysCPUUser      = "user"
+)
+
+var (
+	sysCPUAvgIdleMD   = NewGaugeMD(sysCPUAvgIdle, "Average CPU idle time")
+	sysCPUAvgIOWaitMD = NewGaugeMD(sysCPUAvgIOWait, "Average CPU IOWait time")
+	sysCPULoadMD      = NewGaugeMD(sysCPULoad, "CPU load average 1min")
+	sysCPULoadPercMD  = NewGaugeMD(sysCPULoadPerc, "CPU load average 1min (percentage)")
+	sysCPUNiceMD      = NewGaugeMD(sysCPUNice, "CPU nice time")
+	sysCPUStealMD     = NewGaugeMD(sysCPUSteal, "CPU steal time")
+	sysCPUSystemMD    = NewGaugeMD(sysCPUSystem, "CPU system time")
+	sysCPUUserMD      = NewGaugeMD(sysCPUUser, "CPU user time")
+)
+
+// loadCPUMetrics - `MetricsLoaderFn` for system CPU metrics.
+func loadCPUMetrics(ctx context.Context, m MetricValues, c *metricsCache) error {
+	cpuMetrics, _ := c.cpuMetrics.Get()
+
+	if cpuMetrics.LoadStat != nil {
+		m.Set(sysCPULoad, cpuMetrics.LoadStat.Load1)
+		perc := cpuMetrics.LoadStat.Load1 * 100 / float64(cpuMetrics.CPUCount)
+		m.Set(sysCPULoadPerc, math.Round(perc*100)/100)
+	}
+
+	ts := cpuMetrics.TimesStat
+	tot := ts.User + ts.System + ts.Idle + ts.Iowait + ts.Nice + ts.Steal
+	cpuUserVal := math.Round(ts.User/tot*100*100) / 100
+	m.Set(sysCPUUser, cpuUserVal)
+	cpuSystemVal := math.Round(ts.System/tot*100*100) / 100
+	m.Set(sysCPUSystem, cpuSystemVal)
+	cpuNiceVal := math.Round(ts.Nice/tot*100*100) / 100
+	m.Set(sysCPUNice, cpuNiceVal)
+	cpuStealVal := math.Round(ts.Steal/tot*100*100) / 100
+	m.Set(sysCPUSteal, cpuStealVal)
+
+	// metrics-resource.go runs a job to collect resource metrics including their Avg values and
+	// stores them in resourceMetricsMap. We can use it to get the Avg values of CPU idle and IOWait.
+	cpuResourceMetrics, found := resourceMetricsMap[cpuSubsystem]
+	if found {
+		if cpuIdleMetric, ok := cpuResourceMetrics[getResourceKey(cpuIdle, nil)]; ok {
+			avgVal := math.Round(cpuIdleMetric.Avg*100) / 100
+			m.Set(sysCPUAvgIdle, avgVal)
+		}
+		if cpuIOWaitMetric, ok := cpuResourceMetrics[getResourceKey(cpuIOWait, nil)]; ok {
+			avgVal := math.Round(cpuIOWaitMetric.Avg*100) / 100
+			m.Set(sysCPUAvgIOWait, avgVal)
+		}
+	}
+	return nil
+}

cmd/metrics-v3.go

@@ -36,6 +36,7 @@ const (
 	systemNetworkInternodeCollectorPath collectorPath = "/system/network/internode"
 	systemDriveCollectorPath            collectorPath = "/system/drive"
 	systemMemoryCollectorPath           collectorPath = "/system/memory"
+	systemCPUCollectorPath              collectorPath = "/system/cpu"
 	systemProcessCollectorPath          collectorPath = "/system/process"
 	systemGoCollectorPath               collectorPath = "/system/go"
 
@@ -128,6 +129,20 @@ func newMetricGroups(r *prometheus.Registry) *metricsV3Collection {
 		loadMemoryMetrics,
 	)
 
+	systemCPUMG := NewMetricsGroup(systemCPUCollectorPath,
+		[]MetricDescriptor{
+			sysCPUAvgIdleMD,
+			sysCPUAvgIOWaitMD,
+			sysCPULoadMD,
+			sysCPULoadPercMD,
+			sysCPUNiceMD,
+			sysCPUStealMD,
+			sysCPUSystemMD,
+			sysCPUUserMD,
+		},
+		loadCPUMetrics,
+	)
+
 	systemDriveMG := NewMetricsGroup(systemDriveCollectorPath,
 		[]MetricDescriptor{
 			driveUsedBytesMD,
@@ -235,6 +250,7 @@ func newMetricGroups(r *prometheus.Registry) *metricsV3Collection {
 		systemNetworkInternodeMG,
 		systemDriveMG,
 		systemMemoryMG,
+		systemCPUMG,
 
 		clusterHealthMG,
 		clusterUsageObjectsMG,

docs/metrics/v3.md

@@ -139,6 +139,18 @@ The standard metrics groups for ProcessCollector and GoCollector are not shown b
 | `minio_system_memory_shared`     | `gauge` | Shared memory on the node          | `server` |
 | `minio_system_memory_available`  | `gauge` | Available memory on the node       | `server` |
 
+### `/system/cpu`
+
+| Name                          | Type    | Help                               | Labels   |
+|-------------------------------|---------|------------------------------------|----------|
+| `minio_system_cpu_avg_idle`   | `gauge` | Average CPU idle time              | `server` |
+| `minio_system_cpu_avg_iowait` | `gauge` | Average CPU IOWait time            | `server` |
+| `minio_system_cpu_load`       | `gauge` | CPU load average 1min              | `server` |
+| `minio_system_cpu_load_perc`  | `gauge` | CPU load average 1min (percentage) | `server` |
+| `minio_system_cpu_nice`       | `gauge` | CPU nice time                      | `server` |
+| `minio_system_cpu_steal`      | `gauge` | CPU steal time                     | `server` |
+| `minio_system_cpu_system`     | `gauge` | CPU system time                    | `server` |
+| `minio_system_cpu_user`       | `gauge` | CPU user time                      | `server` |
 
 ### `/system/network/internode`