From 37d49746bc012054ecdf3438a5e49ebaa4eeba12 Mon Sep 17 00:00:00 2001
From: Haoyu Sun <hasun@redhat.com>
Date: Fri, 17 Feb 2023 18:47:40 +0100
Subject: [PATCH] Remove metrics of offline CPUs in CPU collector

Signed-off-by: Haoyu Sun <hasun@redhat.com>
---
 CHANGELOG.md                |  4 +-
 collector/cpu_linux.go      | 10 +++++
 collector/cpu_linux_test.go | 85 +++++++++++++++++++++++++++++++++++++
 go.mod                      |  1 +
 go.sum                      |  2 +
 5 files changed, 101 insertions(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 31abbbfb..aa5ee7bf 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,8 @@
 * [ENHANCEMENT]
 * [BUGFIX]
 
+* [BUGFIX] Remove metrics of offline CPUs in CPU collector #2605
+
 ## 1.5.0 / 2022-11-29
 
 NOTE: This changes the Go runtime "GOMAXPROCS" to 1. This is done to limit the
@@ -139,7 +141,7 @@ NOTE: Filesystem collector flags have been renamed. `--collector.filesystem.igno
 * [BUGFIX] Fix wrong value for OpenBSD memory buffer cache #2015
 * [BUGFIX] Only initiate collectors once #2048
 * [BUGFIX] Handle small backwards jumps in CPU idle #2067
-  
+
 ## 1.1.2 / 2021-03-05
 
 * [BUGFIX] Handle errors from disabled PSI subsystem #1983
diff --git a/collector/cpu_linux.go b/collector/cpu_linux.go
index e16764d4..c323ec14 100644
--- a/collector/cpu_linux.go
+++ b/collector/cpu_linux.go
@@ -29,6 +29,8 @@ import (
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/procfs"
 	"github.com/prometheus/procfs/sysfs"
+	"golang.org/x/exp/maps"
+	"golang.org/x/exp/slices"
 	"gopkg.in/alecthomas/kingpin.v2"
 )
 
@@ -424,4 +426,12 @@ func (c *cpuCollector) updateCPUStats(newStats map[int64]procfs.CPUStat) {
 
 		c.cpuStats[i] = cpuStats
 	}
+
+	// Remove offline CPUs.
+	if len(newStats) != len(c.cpuStats) {
+		onlineCPUIds := maps.Keys(newStats)
+		maps.DeleteFunc(c.cpuStats, func(key int64, item procfs.CPUStat) bool {
+			return !slices.Contains(onlineCPUIds, key)
+		})
+	}
 }
diff --git a/collector/cpu_linux_test.go b/collector/cpu_linux_test.go
index 586087f3..b148ffcd 100644
--- a/collector/cpu_linux_test.go
+++ b/collector/cpu_linux_test.go
@@ -115,3 +115,88 @@ func TestCPU(t *testing.T) {
 		t.Fatalf("should have %v CPU Stat: got %v", resetIdle, got)
 	}
 }
+
+func TestCPUOffline(t *testing.T) {
+	// CPU 1 goes offline.
+	firstCPUStat := map[int64]procfs.CPUStat{
+		0: {
+			User:      100.0,
+			Nice:      100.0,
+			System:    100.0,
+			Idle:      100.0,
+			Iowait:    100.0,
+			IRQ:       100.0,
+			SoftIRQ:   100.0,
+			Steal:     100.0,
+			Guest:     100.0,
+			GuestNice: 100.0,
+		},
+		1: {
+			User:      101.0,
+			Nice:      101.0,
+			System:    101.0,
+			Idle:      101.0,
+			Iowait:    101.0,
+			IRQ:       101.0,
+			SoftIRQ:   101.0,
+			Steal:     101.0,
+			Guest:     101.0,
+			GuestNice: 101.0,
+		},
+	}
+
+	c := makeTestCPUCollector(firstCPUStat)
+	want := map[int64]procfs.CPUStat{
+		0: {
+			User:      100.0,
+			Nice:      100.0,
+			System:    100.0,
+			Idle:      100.0,
+			Iowait:    100.0,
+			IRQ:       100.0,
+			SoftIRQ:   100.0,
+			Steal:     100.0,
+			Guest:     100.0,
+			GuestNice: 100.0,
+		},
+	}
+	c.updateCPUStats(want)
+	got := c.cpuStats
+	if !reflect.DeepEqual(want, got) {
+		t.Fatalf("should have %v CPU Stat: got %v", want, got)
+	}
+
+	// CPU 1 comes back online.
+	want = map[int64]procfs.CPUStat{
+		0: {
+			User:      100.0,
+			Nice:      100.0,
+			System:    100.0,
+			Idle:      100.0,
+			Iowait:    100.0,
+			IRQ:       100.0,
+			SoftIRQ:   100.0,
+			Steal:     100.0,
+			Guest:     100.0,
+			GuestNice: 100.0,
+		},
+		1: {
+			User:      101.0,
+			Nice:      101.0,
+			System:    101.0,
+			Idle:      101.0,
+			Iowait:    101.0,
+			IRQ:       101.0,
+			SoftIRQ:   101.0,
+			Steal:     101.0,
+			Guest:     101.0,
+			GuestNice: 101.0,
+		},
+	}
+	c.updateCPUStats(want)
+	got = c.cpuStats
+	if !reflect.DeepEqual(want, got) {
+		t.Fatalf("should have %v CPU Stat: got %v", want, got)
+	}
+
+}
diff --git a/go.mod b/go.mod
index 93fc7ccc..b0ed690b 100644
--- a/go.mod
+++ b/go.mod
@@ -27,6 +27,7 @@ require (
 	github.com/prometheus/procfs v0.9.0
 	github.com/safchain/ethtool v0.2.0
 	github.com/soundcloud/go-runit v0.0.0-20150630195641-06ad41a06c4a
+	golang.org/x/exp v0.0.0-20230213192124-5e25df0256eb
 	golang.org/x/sys v0.5.0
 	gopkg.in/alecthomas/kingpin.v2 v2.2.6
 )
diff --git a/go.sum b/go.sum
index dda6e32a..d1fce499 100644
--- a/go.sum
+++ b/go.sum
@@ -116,6 +116,8 @@ go.uber.org/multierr v1.6.0/go.mod h1:cdWPpRnG4AhwMwsgIHip0KRBQjJy5kYEpYjJxpXp9i
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
 golang.org/x/crypto v0.0.0-20221012134737-56aed061732a h1:NmSIgad6KjE6VvHciPZuNRTKxGhlPfD6OA87W/PLkqg=
 golang.org/x/crypto v0.0.0-20221012134737-56aed061732a/go.mod h1:IxCIyHEi3zRg3s0A5j5BB6A9Jmi73HwBIUl50j+osU4=
+golang.org/x/exp v0.0.0-20230213192124-5e25df0256eb h1:PaBZQdo+iSDyHT053FjUCgZQ/9uqVwPOcl7KSWhKn6w=
+golang.org/x/exp v0.0.0-20230213192124-5e25df0256eb/go.mod h1:CxIveKay+FTh1D0yPZemJVgC/95VzuuOLq5Qi4xnoYc=
 golang.org/x/net v0.0.0-20190311183353-d8887717615a/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190603091049-60506f45cf65/go.mod h1:HSz+uSET+XFnRR8LxR5pz3Of3rY3CfYBVs4xY44aLks=
 golang.org/x/net v0.0.0-20190827160401-ba9fcec4b297/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=