Don't count empty collection as success (#1613)

Many collectors depend on underlying features to be enabled. This causes
confusion about what "success" means. This changes the behavior of the
`node_scrape_collector_success` metric.

* When a collector is unable to find data don't return success.
* Catch the no data error and send to Debug log level to avoid log spam.
* Update collectors to support this new functionality.
* Fix copy-pasta mistake in infiband debug message.

Closes: https://github.com/prometheus/node_exporter/issues/1323

Signed-off-by: Ben Kochie <superq@gmail.com>

collector/bonding_linux.go

@@ -61,7 +61,7 @@ func (c *bondingCollector) Update(ch chan<- prometheus.Metric) error {
 	if err != nil {
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "Not collecting bonding, file does not exist", "file", statusfile)
-			return nil
+			return ErrNoData
 		}
 		return err
 	}

collector/collector.go

@@ -15,6 +15,7 @@
 package collector
 
 import (
+	"errors"
 	"fmt"
 	"sync"
 	"time"
@@ -131,7 +132,11 @@ func execute(name string, c Collector, ch chan<- prometheus.Metric, logger log.L
 	var success float64
 
 	if err != nil {
-		level.Error(logger).Log("msg", "collector failed", "name", name, "duration_seconds", duration.Seconds(), "err", err)
+		if IsNoDataError(err) {
+			level.Debug(logger).Log("msg", "collector returned no data", "name", name, "duration_seconds", duration.Seconds(), "err", err)
+		} else {
+			level.Error(logger).Log("msg", "collector failed", "name", name, "duration_seconds", duration.Seconds(), "err", err)
+		}
 		success = 0
 	} else {
 		level.Debug(logger).Log("msg", "collector succeeded", "name", name, "duration_seconds", duration.Seconds())
@@ -155,3 +160,10 @@ type typedDesc struct {
 func (d *typedDesc) mustNewConstMetric(value float64, labels ...string) prometheus.Metric {
 	return prometheus.MustNewConstMetric(d.desc, d.valueType, value, labels...)
 }
+
+// ErrNoData indicates the collector found no data to collect, but had no other error.
+var ErrNoData = errors.New("collector returned no data")
+
+func IsNoDataError(err error) bool {
+	return err == ErrNoData
+}

collector/drbd_linux.go

@@ -188,7 +188,7 @@ func (c *drbdCollector) Update(ch chan<- prometheus.Metric) error {
 	if err != nil {
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "stats file does not exist, skipping", "file", statsFile, "err", err)
-			return nil
+			return ErrNoData
 		}
 
 		return err

collector/hwmon_linux.go

@@ -426,7 +426,7 @@ func (c *hwMonCollector) Update(ch chan<- prometheus.Metric) error {
 	if err != nil {
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "hwmon collector metrics are not available for this system")
-			return nil
+			return ErrNoData
 		}
 
 		return err

collector/infiniband_linux.go

@@ -109,8 +109,8 @@ func (c *infinibandCollector) Update(ch chan<- prometheus.Metric) error {
 	devices, err := c.fs.InfiniBandClass()
 	if err != nil {
 		if os.IsNotExist(err) {
-			level.Debug(c.logger).Log("msg", "IPv4 sockstat statistics not found, skipping")
-			return nil
+			level.Debug(c.logger).Log("msg", "infiniband statistics not found, skipping")
+			return ErrNoData
 		}
 		return fmt.Errorf("error obtaining InfiniBand class info: %s", err)
 	}

collector/ipvs_linux.go

@@ -115,7 +115,7 @@ func (c *ipvsCollector) Update(ch chan<- prometheus.Metric) error {
 		// Cannot access ipvs metrics, report no error.
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "ipvs collector metrics are not available for this system")
-			return nil
+			return ErrNoData
 		}
 		return fmt.Errorf("could not get IPVS stats: %s", err)
 	}

collector/mdadm_linux.go

@@ -105,7 +105,7 @@ func (c *mdadmCollector) Update(ch chan<- prometheus.Metric) error {
 	if err != nil {
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "Not collecting mdstat, file does not exist", "file", *procPath)
-			return nil
+			return ErrNoData
 		}
 
 		return fmt.Errorf("error parsing mdstatus: %s", err)

collector/nfs_linux.go

@@ -97,7 +97,7 @@ func (c *nfsCollector) Update(ch chan<- prometheus.Metric) error {
 	if err != nil {
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "Not collecting NFS metrics", "err", err)
-			return nil
+			return ErrNoData
 		}
 		return fmt.Errorf("failed to retrieve nfs stats: %w", err)
 	}

collector/nfsd_linux.go

@@ -63,7 +63,7 @@ func (c *nfsdCollector) Update(ch chan<- prometheus.Metric) error {
 	if err != nil {
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "Not collecting NFSd metrics", "err", err)
-			return nil
+			return ErrNoData
 		}
 		return fmt.Errorf("failed to retrieve nfsd stats: %w", err)
 	}

collector/wifi_linux.go

@@ -167,11 +167,11 @@ func (c *wifiCollector) Update(ch chan<- prometheus.Metric) error {
 		// Cannot access wifi metrics, report no error.
 		if os.IsNotExist(err) {
 			level.Debug(c.logger).Log("msg", "wifi collector metrics are not available for this system")
-			return nil
+			return ErrNoData
 		}
 		if os.IsPermission(err) {
 			level.Debug(c.logger).Log("msg", "wifi collector got permission denied when accessing metrics")
-			return nil
+			return ErrNoData
 		}
 
 		return fmt.Errorf("failed to access wifi data: %w", err)