Merge branch 'master' into marctc/refactor_kingpin

.circleci/config.yml

@@ -7,10 +7,10 @@ executors:
   # should also be updated.
   golang:
     docker:
-      - image: cimg/go:1.20
+      - image: cimg/go:1.21
   arm:
     machine:
-      image: ubuntu-2004:current
+      image: ubuntu-2204:current
     resource_class: arm.medium
 
 jobs:
@@ -42,7 +42,7 @@ jobs:
       - run: git diff --exit-code
   build:
     machine:
-      image: ubuntu-2004:202101-01
+      image: ubuntu-2204:current
     parallelism: 3
     steps:
       - prometheus/setup_environment
@@ -58,7 +58,7 @@ jobs:
           destination: /build
   test_docker:
     machine:
-      image: ubuntu-2204:2022.04.2
+      image: ubuntu-2204:current
     environment:
       DOCKER_TEST_IMAGE_NAME: quay.io/prometheus/golang-builder:1.18-base
       REPO_PATH: github.com/prometheus/node_exporter

.github/workflows/golangci-lint.yml

@@ -18,15 +18,15 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v3
+        uses: actions/checkout@3df4ab11eba7bda6032a0b82a6bb43b11571feac # v4.0.0
       - name: install Go
-        uses: actions/setup-go@v3
+        uses: actions/setup-go@6edd4406fa81c3da01a34fa6f6343087c207a568 # v3.5.0
         with:
           go-version: 1.20.x
       - name: Install snmp_exporter/generator dependencies
         run: sudo apt-get update && sudo apt-get -y install libsnmp-dev
         if: github.repository == 'prometheus/snmp_exporter'
       - name: Lint
-        uses: golangci/golangci-lint-action@v3.4.0
+        uses: golangci/golangci-lint-action@3a919529898de77ec3da873e3063ca4b10e7f5cc # v3.7.0
         with:
-          version: v1.53.3
+          version: v1.54.2

.promu-cgo.yml

@@ -1,7 +1,7 @@
 go:
     # Whenever the Go version is updated here, .circle/config.yml and
     # .promu.yml should also be updated.
-    version: 1.20
+    version: 1.21
     cgo: true
 repository:
     path: github.com/prometheus/node_exporter

.promu.yml

@@ -1,7 +1,7 @@
 go:
     # Whenever the Go version is updated here, .circle/config.yml and
     # .promu-cgo.yml should also be updated.
-    version: 1.20
+    version: 1.21
 repository:
     path: github.com/prometheus/node_exporter
 build:

Makefile.common

@@ -61,7 +61,7 @@ PROMU_URL     := https://github.com/prometheus/promu/releases/download/v$(PROMU_
 SKIP_GOLANGCI_LINT :=
 GOLANGCI_LINT :=
 GOLANGCI_LINT_OPTS ?=
-GOLANGCI_LINT_VERSION ?= v1.53.3
+GOLANGCI_LINT_VERSION ?= v1.54.2
 # golangci-lint only supports linux, darwin and windows platforms on i386/amd64.
 # windows isn't included here because of the path separator being different.
 ifeq ($(GOHOSTOS),$(filter $(GOHOSTOS),linux darwin))

README.md

@@ -186,6 +186,7 @@ buddyinfo | Exposes statistics of memory fragments as reported by /proc/buddyinf
 cgroups | A summary of the number of active and enabled cgroups | Linux
 cpu\_vulnerabilities | Exposes CPU vulnerability information from sysfs. | Linux
 devstat | Exposes device statistics | Dragonfly, FreeBSD
+drm | Expose GPU metrics using sysfs / DRM, `amdgpu` is the only driver which exposes this information through DRM | Linux
 drbd | Exposes Distributed Replicated Block Device statistics (to version 8.4) | Linux
 ethtool | Exposes network interface information and network driver statistics equivalent to `ethtool`, `ethtool -S`, and `ethtool -i`. | Linux
 interrupts | Exposes detailed interrupts statistics. | Linux, OpenBSD

collector/arp_linux.go

@@ -17,18 +17,23 @@
 package collector
 
 import (
+	"errors"
 	"fmt"
+	"net"
 
 	"github.com/go-kit/log"
+	"github.com/jsimonetti/rtnetlink"
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/procfs"
+	"golang.org/x/sys/unix"
 )
 
 type arpCollector struct {
 	fs           procfs.FS
 	deviceFilter deviceFilter
 	entries      *prometheus.Desc
 	logger       log.Logger
+	config       NodeCollectorConfig
 }
 
 func init() {
@@ -38,6 +43,7 @@ func init() {
 type ArpConfig struct {
 	DeviceInclude *string
 	DeviceExclude *string
+	Netlink       *bool
 }
 
 // NewARPCollector returns a new Collector exposing ARP stats.
@@ -69,13 +75,65 @@ func getTotalArpEntries(deviceEntries []procfs.ARPEntry) map[string]uint32 {
 	return entries
 }
 
-func (c *arpCollector) Update(ch chan<- prometheus.Metric) error {
-	entries, err := c.fs.GatherARPEntries()
+func getTotalArpEntriesRTNL() (map[string]uint32, error) {
+	conn, err := rtnetlink.Dial(nil)
+	if err != nil {
+		return nil, err
+	}
+	defer conn.Close()
+
+	neighbors, err := conn.Neigh.List()
 	if err != nil {
-		return fmt.Errorf("could not get ARP entries: %w", err)
+		return nil, err
+	}
+
+	ifIndexEntries := make(map[uint32]uint32)
+
+	for _, n := range neighbors {
+		// Neighbors will also contain IPv6 neighbors, but since this is purely an ARP collector,
+		// restrict to AF_INET. Also skip entries which have state NUD_NOARP to conform to output
+		// of /proc/net/arp.
+		if n.Family == unix.AF_INET && n.State&unix.NUD_NOARP == 0 {
+			ifIndexEntries[n.Index]++
+		}
+	}
+
+	enumEntries := make(map[string]uint32)
+
+	// Convert interface indexes to names.
+	for ifIndex, entryCount := range ifIndexEntries {
+		iface, err := net.InterfaceByIndex(int(ifIndex))
+		if err != nil {
+			if errors.Unwrap(err).Error() == "no such network interface" {
+				continue
+			}
+			return nil, err
+		}
+
+		enumEntries[iface.Name] = entryCount
 	}
 
-	enumeratedEntry := getTotalArpEntries(entries)
+	return enumEntries, nil
+}
+
+func (c *arpCollector) Update(ch chan<- prometheus.Metric) error {
+	var enumeratedEntry map[string]uint32
+
+	if *c.config.Arp.Netlink {
+		var err error
+
+		enumeratedEntry, err = getTotalArpEntriesRTNL()
+		if err != nil {
+			return fmt.Errorf("could not get ARP entries: %w", err)
+		}
+	} else {
+		entries, err := c.fs.GatherARPEntries()
+		if err != nil {
+			return fmt.Errorf("could not get ARP entries: %w", err)
+		}
+
+		enumeratedEntry = getTotalArpEntries(entries)
+	}
 
 	for device, entryCount := range enumeratedEntry {
 		if c.deviceFilter.ignored(device) {

collector/bonding_linux_test.go

@@ -11,6 +11,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+//go:build !nobonding
+// +build !nobonding
+
 package collector
 
 import (

collector/boot_time_solaris.go

@@ -11,8 +11,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build solaris && !noboottime
-// +build solaris,!noboottime
+//go:build !noboottime
+// +build !noboottime
 
 package collector
 

collector/btrfs_linux.go

@@ -137,6 +137,7 @@ func (c *btrfsCollector) getIoctlStats() (map[string]*btrfsIoctlFsStats, error)
 				"err", err)
 			continue
 		}
+		defer fs.Close()
 
 		fsInfo, err := fs.Info()
 		if err != nil {

collector/cpu_linux.go

@@ -37,6 +37,7 @@ type cpuCollector struct {
 	fs                 procfs.FS
 	cpu                *prometheus.Desc
 	cpuInfo            *prometheus.Desc
+	cpuFrequencyHz     *prometheus.Desc
 	cpuFlagsInfo       *prometheus.Desc
 	cpuBugsInfo        *prometheus.Desc
 	cpuGuest           *prometheus.Desc
@@ -100,6 +101,11 @@ func NewCPUCollector(config NodeCollectorConfig, logger log.Logger) (Collector,
 			"CPU information from /proc/cpuinfo.",
 			[]string{"package", "core", "cpu", "vendor", "family", "model", "model_name", "microcode", "stepping", "cachesize"}, nil,
 		),
+		cpuFrequencyHz: prometheus.NewDesc(
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "frequency_hertz"),
+			"CPU frequency in hertz from /proc/cpuinfo.",
+			[]string{"package", "core", "cpu"}, nil,
+		),
 		cpuFlagsInfo: prometheus.NewDesc(
 			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "flag_info"),
 			"The `flags` field of CPU information from /proc/cpuinfo taken from the first core.",
@@ -202,6 +208,20 @@ func (c *cpuCollector) updateInfo(ch chan<- prometheus.Metric) error {
 			cpu.CacheSize)
 	}
 
+	cpuFreqEnabled, ok := collectorState["cpufreq"]
+	if !ok || cpuFreqEnabled == nil {
+		level.Debug(c.logger).Log("cpufreq key missing or nil value in collectorState map", err)
+	} else if !*cpuFreqEnabled {
+		for _, cpu := range info {
+			ch <- prometheus.MustNewConstMetric(c.cpuFrequencyHz,
+				prometheus.GaugeValue,
+				cpu.CPUMHz*1e6,
+				cpu.PhysicalID,
+				cpu.CoreID,
+				strconv.Itoa(int(cpu.Processor)))
+		}
+	}
+
 	if len(info) != 0 {
 		cpu := info[0]
 		if err := updateFieldInfo(cpu.Flags, c.cpuFlagsIncludeRegexp, c.cpuFlagsInfo, ch); err != nil {

collector/cpu_solaris.go

@@ -11,8 +11,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build solaris && !nocpu
-// +build solaris,!nocpu
+//go:build !nocpu
+// +build !nocpu
 
 package collector
 

collector/cpufreq_solaris.go

@@ -11,8 +11,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build solaris && !nocpu
-// +build solaris,!nocpu
+//go:build !nocpu
+// +build !nocpu
 
 package collector
 

collector/diskstats_linux_test.go

@@ -11,6 +11,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+//go:build !nodiskstats
+// +build !nodiskstats
+
 package collector
 
 import (

collector/diskstats_openbsd.go

@@ -11,8 +11,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build openbsd && !amd64 && !nodiskstats
-// +build openbsd,!amd64,!nodiskstats
+//go:build !nodiskstats && !amd64
+// +build !nodiskstats,!amd64
 
 package collector
 

collector/ethtool_linux_test.go

@@ -11,6 +11,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+//go:build !noethtool
+// +build !noethtool
+
 package collector
 
 import (

collector/exec_bsd.go

@@ -49,31 +49,37 @@ func NewExecCollector(config NodeCollectorConfig, logger log.Logger) (Collector,
 				name:        "exec_context_switches_total",
 				description: "Context switches since system boot.  Resets at architecture unsigned integer.",
 				mib:         "vm.stats.sys.v_swtch",
+				labels:      nil,
 			},
 			{
 				name:        "exec_traps_total",
 				description: "Traps since system boot.  Resets at architecture unsigned integer.",
 				mib:         "vm.stats.sys.v_trap",
+				labels:      nil,
 			},
 			{
 				name:        "exec_system_calls_total",
 				description: "System calls since system boot.  Resets at architecture unsigned integer.",
 				mib:         "vm.stats.sys.v_syscall",
 			},
+			labels: nil,
 			{
 				name:        "exec_device_interrupts_total",
 				description: "Device interrupts since system boot.  Resets at architecture unsigned integer.",
 				mib:         "vm.stats.sys.v_intr",
+				labels:      nil,
 			},
 			{
 				name:        "exec_software_interrupts_total",
 				description: "Software interrupts since system boot.  Resets at architecture unsigned integer.",
 				mib:         "vm.stats.sys.v_soft",
+				labels:      nil,
 			},
 			{
 				name:        "exec_forks_total",
 				description: "Number of fork() calls since system boot.  Resets at architecture unsigned integer.",
 				mib:         "vm.stats.vm.v_forks",
+				labels:      nil,
 			},
 		},
 		logger: logger,

collector/fibrechannel_linux.go

@@ -11,8 +11,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build linux && !nofibrechannel
-// +build linux,!nofibrechannel
+//go:build !nofibrechannel
+// +build !nofibrechannel
 
 package collector
 

collector/filefd_linux_test.go

@@ -11,6 +11,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+//go:build !nofilefd
+// +build !nofilefd
+
 package collector
 
 import "testing"

collector/filesystem_common.go

@@ -181,6 +181,11 @@ func (c *filesystemCollector) Update(ch chan<- prometheus.Metric) error {
 			c.deviceErrorDesc, prometheus.GaugeValue,
 			s.deviceError, s.labels.device, s.labels.mountPoint, s.labels.fsType,
 		)
+		ch <- prometheus.MustNewConstMetric(
+			c.roDesc, prometheus.GaugeValue,
+			s.ro, s.labels.device, s.labels.mountPoint, s.labels.fsType,
+		)
+
 		if s.deviceError > 0 {
 			continue
 		}
@@ -205,10 +210,6 @@ func (c *filesystemCollector) Update(ch chan<- prometheus.Metric) error {
 			c.filesFreeDesc, prometheus.GaugeValue,
 			s.filesFree, s.labels.device, s.labels.mountPoint, s.labels.fsType,
 		)
-		ch <- prometheus.MustNewConstMetric(
-			c.roDesc, prometheus.GaugeValue,
-			s.ro, s.labels.device, s.labels.mountPoint, s.labels.fsType,
-		)
 	}
 	return nil
 }