Replace --collectors.enabled with per-collector flags (prometheus#640)

* Move NodeCollector into package collector

* Refactor collector enabling

* Update README with new collector enabled flags

* Fix out-of-date inline flag reference syntax

* Use new flags in end-to-end tests

* Add flag to disable all default collectors

* Track if a flag has been set explicitly

* Add --collectors.disable-defaults to README

* Revert disable-defaults flag

* Shorten flags

* Fixup timex collector registration

* Fix end-to-end tests

* Change procfs and sysfs path flags

* Fix review comments

README.md

@@ -19,7 +19,8 @@ This fork adds HTTP Basic authentication and TLS support using [Percona's shared
 There is varying support for collectors on each operating system. The tables
 below list all existing collectors and the supported systems.
 
-Which collectors are used is controlled by the `--collectors.enabled` flag.
+Collectors are enabled by providing a `--collector.<name>` flag.
+Collectors that are enabled by default can be disabled by providing a `--no-collector.<name>` flag.
 
 ### Enabled by default
 
@@ -164,8 +165,8 @@ docker run -d -p 9100:9100 \
   -v "/:/rootfs:ro" \
   --net="host" \
   quay.io/prometheus/node-exporter \
-    --collector.procfs /host/proc \
-    --collector.sysfs /host/sys \
+    --path.procfs /host/proc \
+    --path.sysfs /host/sys \
     --collector.filesystem.ignored-mount-points "^/(sys|proc|dev|host|etc)($|/)"
 ```
 

collector/arp_linux.go

@@ -30,14 +30,14 @@ type arpCollector struct {
 }
 
 func init() {
-	Factories["arp"] = NewARPCollector
+	registerCollector("arp", defaultEnabled, NewARPCollector)
 }
 
 // NewARPCollector returns a new Collector exposing ARP stats.
 func NewARPCollector() (Collector, error) {
 	return &arpCollector{
 		entries: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "arp", "entries"),
+			prometheus.BuildFQName(namespace, "arp", "entries"),
 			"ARP entries by device",
 			[]string{"device"}, nil,
 		),

collector/bcache_linux.go

@@ -25,7 +25,7 @@ import (
 )
 
 func init() {
-	Factories["bcache"] = NewBcacheCollector
+	registerCollector("bcache", defaultEnabled, NewBcacheCollector)
 }
 
 // A bcacheCollector is a Collector which gathers metrics from Linux bcache.
@@ -283,7 +283,7 @@ func (c *bcacheCollector) updateBcacheStats(ch chan<- prometheus.Metric, s *bcac
 		labels := append(devLabel, m.extraLabel...)
 
 		desc := prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, subsystem, m.name),
+			prometheus.BuildFQName(namespace, subsystem, m.name),
 			m.desc,
 			labels,
 			nil,

collector/bonding_linux.go

@@ -31,20 +31,20 @@ type bondingCollector struct {
 }
 
 func init() {
-	Factories["bonding"] = NewBondingCollector
+	registerCollector("bonding", defaultDisabled, NewBondingCollector)
 }
 
 // NewBondingCollector returns a newly allocated bondingCollector.
 // It exposes the number of configured and active slave of linux bonding interfaces.
 func NewBondingCollector() (Collector, error) {
 	return &bondingCollector{
 		slaves: typedDesc{prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "bonding", "slaves"),
+			prometheus.BuildFQName(namespace, "bonding", "slaves"),
 			"Number of configured slaves per bonding interface.",
 			[]string{"master"}, nil,
 		), prometheus.GaugeValue},
 		active: typedDesc{prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "bonding", "active"),
+			prometheus.BuildFQName(namespace, "bonding", "active"),
 			"Number of active slaves per bonding interface.",
 			[]string{"master"}, nil,
 		), prometheus.GaugeValue},

collector/buddyinfo.go

@@ -34,13 +34,13 @@ type buddyinfoCollector struct {
 }
 
 func init() {
-	Factories["buddyinfo"] = NewBuddyinfoCollector
+	registerCollector("buddyinfo", defaultDisabled, NewBuddyinfoCollector)
 }
 
 // NewBuddyinfoCollector returns a new Collector exposing buddyinfo stats.
 func NewBuddyinfoCollector() (Collector, error) {
 	desc := prometheus.NewDesc(
-		prometheus.BuildFQName(Namespace, buddyInfoSubsystem, "count"),
+		prometheus.BuildFQName(namespace, buddyInfoSubsystem, "count"),
 		"Count of free blocks according to size.",
 		[]string{"node", "zone", "size"}, nil,
 	)

collector/collector.go

@@ -15,20 +15,121 @@
 package collector
 
 import (
+	"fmt"
+	"sync"
+	"time"
+
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/common/log"
+	"gopkg.in/alecthomas/kingpin.v2"
 )
 
 // Namespace defines the common namespace to be used by all metrics.
-const Namespace = "node"
+const namespace = "node"
 
-// Factories contains the list of all available collectors.
-var Factories = make(map[string]func() (Collector, error))
+var (
+	scrapeDurationDesc = prometheus.NewDesc(
+		prometheus.BuildFQName(namespace, "scrape", "collector_duration_seconds"),
+		"node_exporter: Duration of a collector scrape.",
+		[]string{"collector"},
+		nil,
+	)
+	scrapeSuccessDesc = prometheus.NewDesc(
+		prometheus.BuildFQName(namespace, "scrape", "collector_success"),
+		"node_exporter: Whether a collector succeeded.",
+		[]string{"collector"},
+		nil,
+	)
+)
 
 func warnDeprecated(collector string) {
 	log.Warnf("The %s collector is deprecated and will be removed in the future!", collector)
 }
 
+const (
+	defaultEnabled  = true
+	defaultDisabled = false
+)
+
+var (
+	factories      = make(map[string]func() (Collector, error))
+	collectorState = make(map[string]*bool)
+)
+
+func registerCollector(collector string, isDefaultEnabled bool, factory func() (Collector, error)) {
+	var helpDefaultState string
+	if isDefaultEnabled {
+		helpDefaultState = "enabled"
+	} else {
+		helpDefaultState = "disabled"
+	}
+
+	flagName := fmt.Sprintf("collector.%s", collector)
+	flagHelp := fmt.Sprintf("Enable the %s collector (default: %s).", collector, helpDefaultState)
+	defaultValue := fmt.Sprintf("%v", isDefaultEnabled)
+
+	flag := kingpin.Flag(flagName, flagHelp).Default(defaultValue).Bool()
+	collectorState[collector] = flag
+
+	factories[collector] = factory
+}
+
+// NodeCollector implements the prometheus.Collector interface.
+type nodeCollector struct {
+	Collectors map[string]Collector
+}
+
+// NewNodeCollector creates a new NodeCollector
+func NewNodeCollector() (*nodeCollector, error) {
+	collectors := make(map[string]Collector)
+	for key, enabled := range collectorState {
+		if *enabled {
+			collector, err := factories[key]()
+			if err != nil {
+				return nil, err
+			}
+			collectors[key] = collector
+		}
+	}
+	return &nodeCollector{Collectors: collectors}, nil
+}
+
+// Describe implements the prometheus.Collector interface.
+func (n nodeCollector) Describe(ch chan<- *prometheus.Desc) {
+	ch <- scrapeDurationDesc
+	ch <- scrapeSuccessDesc
+}
+
+// Collect implements the prometheus.Collector interface.
+func (n nodeCollector) Collect(ch chan<- prometheus.Metric) {
+	wg := sync.WaitGroup{}
+	wg.Add(len(n.Collectors))
+	for name, c := range n.Collectors {
+		go func(name string, c Collector) {
+			execute(name, c, ch)
+			wg.Done()
+		}(name, c)
+	}
+	wg.Wait()
+}
+
+func execute(name string, c Collector, ch chan<- prometheus.Metric) {
+	begin := time.Now()
+	err := c.Update(ch)
+	duration := time.Since(begin)
+	var success float64
+
+	if err != nil {
+		log.Errorf("ERROR: %s collector failed after %fs: %s", name, duration.Seconds(), err)
+		success = 0
+	} else {
+		log.Debugf("OK: %s collector succeeded after %fs.", name, duration.Seconds())
+		success = 1
+	}
+	ch <- prometheus.MustNewConstMetric(scrapeDurationDesc, prometheus.GaugeValue, duration.Seconds(), name)
+	ch <- prometheus.MustNewConstMetric(scrapeSuccessDesc, prometheus.GaugeValue, success, name)
+}
+
 // Collector is the interface a collector has to implement.
 type Collector interface {
 	// Get new metrics and expose them via prometheus registry.

collector/conntrack_linux.go

@@ -25,19 +25,19 @@ type conntrackCollector struct {
 }
 
 func init() {
-	Factories["conntrack"] = NewConntrackCollector
+	registerCollector("conntrack", defaultEnabled, NewConntrackCollector)
 }
 
 // NewConntrackCollector returns a new Collector exposing conntrack stats.
 func NewConntrackCollector() (Collector, error) {
 	return &conntrackCollector{
 		current: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "", "nf_conntrack_entries"),
+			prometheus.BuildFQName(namespace, "", "nf_conntrack_entries"),
 			"Number of currently allocated flow entries for connection tracking.",
 			nil, nil,
 		),
 		limit: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "", "nf_conntrack_entries_limit"),
+			prometheus.BuildFQName(namespace, "", "nf_conntrack_entries_limit"),
 			"Maximum size of connection tracking table.",
 			nil, nil,
 		),

collector/cpu_darwin.go

@@ -52,14 +52,14 @@ type statCollector struct {
 }
 
 func init() {
-	Factories["cpu"] = NewCPUCollector
+	registerCollector("cpu", defaultEnabled, NewCPUCollector)
 }
 
 // NewCPUCollector returns a new Collector exposing CPU stats.
 func NewCPUCollector() (Collector, error) {
 	return &statCollector{
 		cpu: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "", "cpu"),
+			prometheus.BuildFQName(namespace, "", "cpu"),
 			"Seconds the cpus spent in each mode.",
 			[]string{"cpu", "mode"}, nil,
 		),

collector/cpu_dragonfly.go

@@ -88,14 +88,14 @@ type statCollector struct {
 }
 
 func init() {
-	Factories["cpu"] = NewStatCollector
+	registerCollector("cpu", defaultEnabled, NewStatCollector)
 }
 
 // NewStatCollector returns a new Collector exposing CPU stats.
 func NewStatCollector() (Collector, error) {
 	return &statCollector{
 		cpu: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "", "cpu"),
+			prometheus.BuildFQName(namespace, "", "cpu"),
 			"Seconds the cpus spent in each mode.",
 			[]string{"cpu", "mode"}, nil,
 		),

collector/cpu_freebsd.go

@@ -86,19 +86,19 @@ type statCollector struct {
 }
 
 func init() {
-	Factories["cpu"] = NewStatCollector
+	registerCollector("cpu", defaultEnabled, NewStatCollector)
 }
 
 // NewStatCollector returns a new Collector exposing CPU stats.
 func NewStatCollector() (Collector, error) {
 	return &statCollector{
 		cpu: typedDesc{prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "cpu", "seconds_total"),
+			prometheus.BuildFQName(namespace, "cpu", "seconds_total"),
 			"Seconds the CPU spent in each mode.",
 			[]string{"cpu", "mode"}, nil,
 		), prometheus.CounterValue},
 		temp: typedDesc{prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "cpu", "temperature_celsius"),
+			prometheus.BuildFQName(namespace, "cpu", "temperature_celsius"),
 			"CPU temperature",
 			[]string{"cpu"}, nil,
 		), prometheus.GaugeValue},

collector/cpu_linux.go

@@ -29,7 +29,7 @@ import (
 )
 
 const (
-	cpuCollectorNamespace = "cpu"
+	cpuCollectorSubsystem = "cpu"
 )
 
 var (
@@ -46,40 +46,40 @@ type cpuCollector struct {
 }
 
 func init() {
-	Factories["cpu"] = NewCPUCollector
+	registerCollector("cpu", defaultEnabled, NewCPUCollector)
 }
 
 // NewCPUCollector returns a new Collector exposing kernel/system statistics.
 func NewCPUCollector() (Collector, error) {
 	return &cpuCollector{
 		cpu: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, "", cpuCollectorNamespace),
+			prometheus.BuildFQName(namespace, "", cpuCollectorSubsystem),
 			"Seconds the cpus spent in each mode.",
 			[]string{"cpu", "mode"}, nil,
 		),
 		cpuFreq: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, cpuCollectorNamespace, "frequency_hertz"),
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "frequency_hertz"),
 			"Current cpu thread frequency in hertz.",
 			[]string{"cpu"}, nil,
 		),
 		cpuFreqMin: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, cpuCollectorNamespace, "frequency_min_hertz"),
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "frequency_min_hertz"),
 			"Minimum cpu thread frequency in hertz.",
 			[]string{"cpu"}, nil,
 		),
 		cpuFreqMax: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, cpuCollectorNamespace, "frequency_max_hertz"),
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "frequency_max_hertz"),
 			"Maximum cpu thread frequency in hertz.",
 			[]string{"cpu"}, nil,
 		),
 		// FIXME: This should be a per core metric, not per cpu!
 		cpuCoreThrottle: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, cpuCollectorNamespace, "core_throttles_total"),
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "core_throttles_total"),
 			"Number of times this cpu core has been throttled.",
 			[]string{"cpu"}, nil,
 		),
 		cpuPackageThrottle: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, cpuCollectorNamespace, "package_throttles_total"),
+			prometheus.BuildFQName(namespace, cpuCollectorSubsystem, "package_throttles_total"),
 			"Number of times this cpu package has been throttled.",
 			[]string{"node"}, nil,
 		),

collector/devstat_dragonfly.go

@@ -99,24 +99,24 @@ type devstatCollector struct {
 }
 
 func init() {
-	Factories["devstat"] = NewDevstatCollector
+	registerCollector("devstat", defaultDisabled, NewDevstatCollector)
 }
 
 // NewDevstatCollector returns a new Collector exposing Device stats.
 func NewDevstatCollector() (Collector, error) {
 	return &devstatCollector{
 		bytesDesc: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, devstatSubsystem, "bytes_total"),
+			prometheus.BuildFQName(namespace, devstatSubsystem, "bytes_total"),
 			"The total number of bytes transferred for reads and writes on the device.",
 			[]string{"device"}, nil,
 		),
 		transfersDesc: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, devstatSubsystem, "transfers_total"),
+			prometheus.BuildFQName(namespace, devstatSubsystem, "transfers_total"),
 			"The total number of transactions completed.",
 			[]string{"device"}, nil,
 		),
 		blocksDesc: prometheus.NewDesc(
-			prometheus.BuildFQName(Namespace, devstatSubsystem, "blocks_total"),
+			prometheus.BuildFQName(namespace, devstatSubsystem, "blocks_total"),
 			"The total number of bytes given in terms of the devices blocksize.",
 			[]string{"device"}, nil,
 		),