Split out guest cpu metrics on Linux. (#744)

Linux "guest" metrics for VMs are already accounted for in node_cpu
`user` and `nice` metrics.  Separate these into their own metric to
avoid duplication of data.

CHANGELOG.md

@@ -1,3 +1,9 @@
+## master / unreleased
+
+**Breaking changes**
+
+Linux node_cpu metrics now break out `guest` values into separate metrics.  See Issue #737
+
 ## v0.15.0 / 2017-10-06
 
 **Breaking changes**

collector/cpu_linux.go

@@ -38,6 +38,7 @@ var (
 
 type cpuCollector struct {
 	cpu                *prometheus.Desc
+	cpuGuest           *prometheus.Desc
 	cpuFreq            *prometheus.Desc
 	cpuFreqMin         *prometheus.Desc
 	cpuFreqMax         *prometheus.Desc
@@ -57,6 +58,11 @@ func NewCPUCollector() (Collector, error) {
 			"Seconds the cpus spent in each mode.",
 			[]string{"cpu", "mode"}, nil,
 		),
+		cpuGuest: prometheus.NewDesc(
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "guest_seconds_total"),
+			"Seconds the cpus spent in guests (VMs) for each mode.",
+			[]string{"cpu", "mode"}, nil,
+		),
 		cpuFreq: prometheus.NewDesc(
 			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "frequency_hertz"),
 			"Current cpu thread frequency in hertz.",
@@ -196,6 +202,7 @@ func (c *cpuCollector) updateStat(ch chan<- prometheus.Metric) error {
 
 	for cpuID, cpuStat := range stats.CPU {
 		cpuName := fmt.Sprintf("cpu%d", cpuID)
+		cpuNum := fmt.Sprintf("%d", cpuID)
 		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.User, cpuName, "user")
 		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.Nice, cpuName, "nice")
 		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.System, cpuName, "system")
@@ -204,8 +211,10 @@ func (c *cpuCollector) updateStat(ch chan<- prometheus.Metric) error {
 		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.IRQ, cpuName, "irq")
 		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.SoftIRQ, cpuName, "softirq")
 		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.Steal, cpuName, "steal")
-		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.Guest, cpuName, "guest")
+
-		ch <- prometheus.MustNewConstMetric(c.cpu, prometheus.CounterValue, cpuStat.GuestNice, cpuName, "guest_nice")
+		// Guest CPU is also accounted for in cpuStat.User and cpuStat.Nice, expose these as separate metrics.
+		ch <- prometheus.MustNewConstMetric(c.cpuGuest, prometheus.CounterValue, cpuStat.Guest, cpuNum, "user")
+		ch <- prometheus.MustNewConstMetric(c.cpuGuest, prometheus.CounterValue, cpuStat.GuestNice, cpuNum, "nice")
 	}
 
 	return nil

collector/fixtures/e2e-output.txt

@@ -201,8 +201,6 @@ node_buddyinfo_count{node="0",size="9",zone="Normal"} 0
 node_context_switches 3.8014093e+07
 # HELP node_cpu Seconds the cpus spent in each mode.
 # TYPE node_cpu counter
-node_cpu{cpu="cpu0",mode="guest"} 0
-node_cpu{cpu="cpu0",mode="guest_nice"} 0
 node_cpu{cpu="cpu0",mode="idle"} 10870.69
 node_cpu{cpu="cpu0",mode="iowait"} 2.2
 node_cpu{cpu="cpu0",mode="irq"} 0.01
@@ -211,8 +209,6 @@ node_cpu{cpu="cpu0",mode="softirq"} 34.1
 node_cpu{cpu="cpu0",mode="steal"} 0
 node_cpu{cpu="cpu0",mode="system"} 210.45
 node_cpu{cpu="cpu0",mode="user"} 444.9
-node_cpu{cpu="cpu1",mode="guest"} 0
-node_cpu{cpu="cpu1",mode="guest_nice"} 0
 node_cpu{cpu="cpu1",mode="idle"} 11107.87
 node_cpu{cpu="cpu1",mode="iowait"} 5.91
 node_cpu{cpu="cpu1",mode="irq"} 0
@@ -221,8 +217,6 @@ node_cpu{cpu="cpu1",mode="softirq"} 0.46
 node_cpu{cpu="cpu1",mode="steal"} 0
 node_cpu{cpu="cpu1",mode="system"} 164.74
 node_cpu{cpu="cpu1",mode="user"} 478.69
-node_cpu{cpu="cpu2",mode="guest"} 0
-node_cpu{cpu="cpu2",mode="guest_nice"} 0
 node_cpu{cpu="cpu2",mode="idle"} 11123.21
 node_cpu{cpu="cpu2",mode="iowait"} 4.41
 node_cpu{cpu="cpu2",mode="irq"} 0
@@ -231,8 +225,6 @@ node_cpu{cpu="cpu2",mode="softirq"} 3.26
 node_cpu{cpu="cpu2",mode="steal"} 0
 node_cpu{cpu="cpu2",mode="system"} 159.16
 node_cpu{cpu="cpu2",mode="user"} 465.04
-node_cpu{cpu="cpu3",mode="guest"} 0
-node_cpu{cpu="cpu3",mode="guest_nice"} 0
 node_cpu{cpu="cpu3",mode="idle"} 11132.3
 node_cpu{cpu="cpu3",mode="iowait"} 5.33
 node_cpu{cpu="cpu3",mode="irq"} 0
@@ -241,8 +233,6 @@ node_cpu{cpu="cpu3",mode="softirq"} 0.6
 node_cpu{cpu="cpu3",mode="steal"} 0
 node_cpu{cpu="cpu3",mode="system"} 156.83
 node_cpu{cpu="cpu3",mode="user"} 470.54
-node_cpu{cpu="cpu4",mode="guest"} 0
-node_cpu{cpu="cpu4",mode="guest_nice"} 0
 node_cpu{cpu="cpu4",mode="idle"} 11403.21
 node_cpu{cpu="cpu4",mode="iowait"} 2.17
 node_cpu{cpu="cpu4",mode="irq"} 0
@@ -251,8 +241,6 @@ node_cpu{cpu="cpu4",mode="softirq"} 0.08
 node_cpu{cpu="cpu4",mode="steal"} 0
 node_cpu{cpu="cpu4",mode="system"} 107.76
 node_cpu{cpu="cpu4",mode="user"} 284.13
-node_cpu{cpu="cpu5",mode="guest"} 0
-node_cpu{cpu="cpu5",mode="guest_nice"} 0
 node_cpu{cpu="cpu5",mode="idle"} 11362.7
 node_cpu{cpu="cpu5",mode="iowait"} 6.72
 node_cpu{cpu="cpu5",mode="irq"} 0
@@ -261,8 +249,6 @@ node_cpu{cpu="cpu5",mode="softirq"} 0.3
 node_cpu{cpu="cpu5",mode="steal"} 0
 node_cpu{cpu="cpu5",mode="system"} 115.86
 node_cpu{cpu="cpu5",mode="user"} 292.71
-node_cpu{cpu="cpu6",mode="guest"} 0
-node_cpu{cpu="cpu6",mode="guest_nice"} 0
 node_cpu{cpu="cpu6",mode="idle"} 11397.21
 node_cpu{cpu="cpu6",mode="iowait"} 3.19
 node_cpu{cpu="cpu6",mode="irq"} 0
@@ -271,8 +257,6 @@ node_cpu{cpu="cpu6",mode="softirq"} 0.29
 node_cpu{cpu="cpu6",mode="steal"} 0
 node_cpu{cpu="cpu6",mode="system"} 102.76
 node_cpu{cpu="cpu6",mode="user"} 291.52
-node_cpu{cpu="cpu7",mode="guest"} 0
-node_cpu{cpu="cpu7",mode="guest_nice"} 0
 node_cpu{cpu="cpu7",mode="idle"} 11392.82
 node_cpu{cpu="cpu7",mode="iowait"} 5.55
 node_cpu{cpu="cpu7",mode="irq"} 0
@@ -301,6 +285,24 @@ node_cpu_frequency_max_hertz{cpu="cpu3"} 4.2e+09
 node_cpu_frequency_min_hertz{cpu="cpu0"} 8e+08
 node_cpu_frequency_min_hertz{cpu="cpu1"} 8e+08
 node_cpu_frequency_min_hertz{cpu="cpu3"} 1e+06
+# HELP node_cpu_guest_seconds_total Seconds the cpus spent for guests (VMs) in each mode.
+# TYPE node_cpu_guest_seconds_total counter
+node_cpu_guest_seconds_total{cpu="0",mode="nice"} 0.01
+node_cpu_guest_seconds_total{cpu="0",mode="user"} 0.02
+node_cpu_guest_seconds_total{cpu="1",mode="nice"} 0.02
+node_cpu_guest_seconds_total{cpu="1",mode="user"} 0.03
+node_cpu_guest_seconds_total{cpu="2",mode="nice"} 0.03
+node_cpu_guest_seconds_total{cpu="2",mode="user"} 0.04
+node_cpu_guest_seconds_total{cpu="3",mode="nice"} 0.04
+node_cpu_guest_seconds_total{cpu="3",mode="user"} 0.05
+node_cpu_guest_seconds_total{cpu="4",mode="nice"} 0.05
+node_cpu_guest_seconds_total{cpu="4",mode="user"} 0.06
+node_cpu_guest_seconds_total{cpu="5",mode="nice"} 0.06
+node_cpu_guest_seconds_total{cpu="5",mode="user"} 0.07
+node_cpu_guest_seconds_total{cpu="6",mode="nice"} 0.07
+node_cpu_guest_seconds_total{cpu="6",mode="user"} 0.08
+node_cpu_guest_seconds_total{cpu="7",mode="nice"} 0.08
+node_cpu_guest_seconds_total{cpu="7",mode="user"} 0.09
 # HELP node_cpu_package_throttles_total Number of times this cpu package has been throttled.
 # TYPE node_cpu_package_throttles_total counter
 node_cpu_package_throttles_total{node="0"} 30

collector/fixtures/proc/stat

@@ -1,12 +1,12 @@
-cpu  301854 612 111922 8979004 3552 2 3944 0 0 0
+cpu  301854 612 111922 8979004 3552 2 3944 0 44 36
-cpu0 44490 19 21045 1087069 220 1 3410 0 0 0
+cpu0 44490 19 21045 1087069 220 1 3410 0 2 1
-cpu1 47869 23 16474 1110787 591 0 46 0 0 0
+cpu1 47869 23 16474 1110787 591 0 46 0 3 2
-cpu2 46504 36 15916 1112321 441 0 326 0 0 0
+cpu2 46504 36 15916 1112321 441 0 326 0 4 3
-cpu3 47054 102 15683 1113230 533 0 60 0 0 0
+cpu3 47054 102 15683 1113230 533 0 60 0 5 4
-cpu4 28413 25 10776 1140321 217 0 8 0 0 0
+cpu4 28413 25 10776 1140321 217 0 8 0 6 5
-cpu5 29271 101 11586 1136270 672 0 30 0 0 0
+cpu5 29271 101 11586 1136270 672 0 30 0 7 6
-cpu6 29152 36 10276 1139721 319 0 29 0 0 0
+cpu6 29152 36 10276 1139721 319 0 29 0 8 7
-cpu7 29098 268 10164 1139282 555 0 31 0 0 0
+cpu7 29098 268 10164 1139282 555 0 31 0 9 8
 intr 8885917 17 0 0 0 0 0 0 0 1 79281 0 0 0 0 0 0 0 231237 0 0 0 0 250586 103 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 223424 190745 13 906 1283803 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 ctxt 38014093
 btime 1418183276