feat: Tolerate collector failures (#1769)

Signed-off-by: Jan-Otto Kröpke <[email protected]>

.golangci.yaml

@@ -11,7 +11,6 @@ linters:
     - exportloopref
     - fatcontext
     - funlen
-    - gochecknoglobals
     - gocognit
     - goconst
     - gocyclo
@@ -88,7 +87,3 @@ issues:
   - text: "don't use ALL_CAPS in Go names; use CamelCase"
     linters:
     - revive
-  - path: internal/perfdata/v1/
-    linters:
-      - godox
-      - stylecheck

.run/all.run.xml

@@ -0,0 +1,13 @@
+<component name="ProjectRunConfigurationManager">
+  <configuration default="false" name="all" type="GoApplicationRunConfiguration" factoryName="Go Application" folderName="run">
+    <module name="windows_exporter" />
+    <working_directory value="$PROJECT_DIR$" />
+    <parameters value="--web.listen-address=127.0.0.1:9182 --log.level=debug --collectors.enabled=ad,adcs,adfs,cache,container,cpu,cpu_info,cs,dfsr,dhcp,diskdrive,dns,exchange,filetime,fsrmquota,hyperv,iis,license,logical_disk,logon,memory,mscluster,msmq,mssql,net,netframework,nps,os,pagefile,perfdata,physical_disk,printer,process,remote_fx,scheduled_task,service,smb,smbclient,smtp,system,tcp,terminal_services,textfile,thermalzone,time,udp,update,vmware" />
+    <sudo value="true" />
+    <kind value="PACKAGE" />
+    <package value="github.com/prometheus-community/windows_exporter/cmd/windows_exporter" />
+    <directory value="$PROJECT_DIR$" />
+    <filePath value="$PROJECT_DIR$/exporter.go" />
+    <method v="2" />
+  </configuration>
+</component>

cmd/windows_exporter/main.go

@@ -42,6 +42,7 @@ import (
 	"github.com/prometheus-community/windows_exporter/internal/httphandler"
 	"github.com/prometheus-community/windows_exporter/internal/log"
 	"github.com/prometheus-community/windows_exporter/internal/log/flag"
+	"github.com/prometheus-community/windows_exporter/internal/utils"
 	"github.com/prometheus-community/windows_exporter/pkg/collector"
 	"github.com/prometheus/common/version"
 	"github.com/prometheus/exporter-toolkit/web"
@@ -64,6 +65,8 @@ func main() {
 }
 
 func run() int {
+	startTime := time.Now()
+
 	app := kingpin.New("windows_exporter", "A metrics collector for Windows.")
 
 	var (
@@ -191,7 +194,7 @@ func run() int {
 
 	enabledCollectorList := expandEnabledCollectors(*enabledCollectors)
 	if err := collectors.Enable(enabledCollectorList); err != nil {
-		logger.Error("Couldn't enable collectors",
+		logger.Error("couldn't enable collectors",
 			slog.Any("err", err),
 		)
 
@@ -200,11 +203,11 @@ func run() int {
 
 	// Initialize collectors before loading
 	if err = collectors.Build(logger); err != nil {
-		logger.Error("Couldn't load collectors",
-			slog.Any("err", err),
-		)
-
-		return 1
+		for _, err := range utils.SplitError(err) {
+			logger.Warn("couldn't initialize collector",
+				slog.Any("err", err),
+			)
+		}
 	}
 
 	logCurrentUser(logger)
@@ -228,7 +231,7 @@ func run() int {
 		mux.HandleFunc("GET /debug/pprof/trace", pprof.Trace)
 	}
 
-	logger.Info("Starting windows_exporter",
+	logger.Info(fmt.Sprintf("starting windows_exporter in %s", time.Since(startTime)),
 		slog.String("version", version.Version),
 		slog.String("branch", version.Branch),
 		slog.String("revision", version.GetRevision()),

internal/collector/ad/ad.go

@@ -16,7 +16,6 @@
 package ad
 
 import (
-	"errors"
 	"fmt"
 	"log/slog"
 
@@ -31,6 +30,7 @@ const Name = "ad"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 type Collector struct {
@@ -671,7 +671,7 @@ func (c *Collector) Collect(ch chan<- prometheus.Metric) error {
 	data, ok := perfData["NTDS"]
 
 	if !ok {
-		return errors.New("perflib query for DirectoryServices (AD) returned empty result set")
+		return fmt.Errorf("failed to collect DirectoryServices (AD) metrics: %w", types.ErrNoData)
 	}
 
 	ch <- prometheus.MustNewConstMetric(

internal/collector/adcs/adcs.go

@@ -16,7 +16,6 @@
 package adcs
 
 import (
-	"errors"
 	"fmt"
 	"log/slog"
 
@@ -32,6 +31,7 @@ const Name = "adcs"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 type Collector struct {
@@ -191,7 +191,7 @@ func (c *Collector) Collect(ch chan<- prometheus.Metric) error {
 	}
 
 	if len(perfData) == 0 {
-		return errors.New("perflib query for Certification Authority (ADCS) returned empty result set")
+		return fmt.Errorf("failed to collect Certification Authority (ADCS) metrics: %w", types.ErrNoData)
 	}
 
 	for name, data := range perfData {

internal/collector/adfs/adfs.go

@@ -16,7 +16,6 @@
 package adfs
 
 import (
-	"errors"
 	"fmt"
 	"log/slog"
 	"maps"
@@ -34,6 +33,7 @@ const Name = "adfs"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 type Collector struct {
@@ -160,7 +160,7 @@ func (c *Collector) Build(_ *slog.Logger, _ *mi.Session) error {
 		avgConfigDBQueryTime,
 		federationMetadataRequests,
 	})
-	if err != nil && !errors.Is(err, perfdata.ErrNoData) {
+	if err != nil {
 		return fmt.Errorf("failed to create AD FS collector: %w", err)
 	}
 
@@ -435,13 +435,13 @@ func (c *Collector) Collect(ch chan<- prometheus.Metric) error {
 	instanceKey := slices.Collect(maps.Keys(data))
 
 	if len(instanceKey) == 0 {
-		return errors.New("perflib query for ADFS returned empty result set")
+		return fmt.Errorf("failed to collect ADFS metrics: %w", types.ErrNoData)
 	}
 
 	adfsData, ok := data[instanceKey[0]]
 
 	if !ok {
-		return errors.New("perflib query for ADFS returned empty result set")
+		return fmt.Errorf("failed to collect ADFS metrics: %w", types.ErrNoData)
 	}
 
 	ch <- prometheus.MustNewConstMetric(

internal/collector/cache/cache.go

@@ -16,7 +16,6 @@
 package cache
 
 import (
-	"errors"
 	"fmt"
 	"log/slog"
 
@@ -31,6 +30,7 @@ const Name = "cache"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector for Perflib Cache metrics.
@@ -322,7 +322,7 @@ func (c *Collector) Collect(ch chan<- prometheus.Metric) error {
 	cacheData, ok := data[perfdata.InstanceEmpty]
 
 	if !ok {
-		return errors.New("perflib query for Cache returned empty result set")
+		return fmt.Errorf("failed to collect Cache metrics: %w", types.ErrNoData)
 	}
 
 	ch <- prometheus.MustNewConstMetric(

internal/collector/container/container.go

@@ -33,6 +33,7 @@ const Name = "container"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector for containers metrics.

internal/collector/cpu/cpu.go

@@ -32,6 +32,7 @@ const Name = "cpu"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 type Collector struct {

internal/collector/cpu_info/cpu_info.go

@@ -32,6 +32,7 @@ const Name = "cpu_info"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector for a few WMI metrics in Win32_Processor.
@@ -147,6 +148,11 @@ func (c *Collector) Build(_ *slog.Logger, miSession *mi.Session) error {
 		nil,
 	)
 
+	var dst []miProcessor
+	if err := c.miSession.Query(&dst, mi.NamespaceRootCIMv2, c.miQuery); err != nil {
+		return fmt.Errorf("WMI query failed: %w", err)
+	}
+
 	return nil
 }
 

internal/collector/cs/cs.go

@@ -29,6 +29,7 @@ const Name = "cs"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector for WMI metrics.

internal/collector/dfsr/const.go

@@ -53,5 +53,5 @@ const (
 	databaseLookupsTotal           = "Database Lookups"
 	usnJournalRecordsReadTotal     = "USN Journal Records Read"
 	usnJournalRecordsAcceptedTotal = "USN Journal Records Accepted"
-	usnJournalUnreadPercentage     = "USN Journal Records Unread Percentage"
+	usnJournalUnreadPercentage     = "USN Journal Unread Percentage"
 )

internal/collector/dfsr/dfsr.go

@@ -35,6 +35,7 @@ type Config struct {
 	CollectorsEnabled []string `yaml:"collectors_enabled"`
 }
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{
 	CollectorsEnabled: []string{"connection", "folder", "volume"},
 }
@@ -542,7 +543,7 @@ func (c *Collector) collectPDHConnection(ch chan<- prometheus.Metric) error {
 	}
 
 	if len(perfData) == 0 {
-		return errors.New("perflib query for DFS Replication Connections returned empty result set")
+		return fmt.Errorf("failed to collect DFS Replication Connections metrics: %w", types.ErrNoData)
 	}
 
 	for name, connection := range perfData {
@@ -620,7 +621,7 @@ func (c *Collector) collectPDHFolder(ch chan<- prometheus.Metric) error {
 	}
 
 	if len(perfData) == 0 {
-		return errors.New("perflib query for DFS Replicated Folders returned empty result set")
+		return fmt.Errorf("failed to collect DFS Replicated Folders metrics: %w", types.ErrNoData)
 	}
 
 	for name, folder := range perfData {
@@ -824,7 +825,7 @@ func (c *Collector) collectPDHVolume(ch chan<- prometheus.Metric) error {
 	}
 
 	if len(perfData) == 0 {
-		return errors.New("perflib query for DFS Replication Volumes returned empty result set")
+		return fmt.Errorf("failed to collect DFS Replication Volumes metrics: %w", types.ErrNoData)
 	}
 
 	for name, volume := range perfData {

internal/collector/dhcp/dhcp.go

@@ -16,7 +16,6 @@
 package dhcp
 
 import (
-	"errors"
 	"fmt"
 	"log/slog"
 
@@ -31,6 +30,7 @@ const Name = "dhcp"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector perflib DHCP metrics.
@@ -288,7 +288,7 @@ func (c *Collector) Collect(ch chan<- prometheus.Metric) error {
 
 	data, ok := perfData[perfdata.InstanceEmpty]
 	if !ok {
-		return errors.New("perflib query for DHCP Server returned empty result set")
+		return fmt.Errorf("failed to collect DHCP Server metrics: %w", types.ErrNoData)
 	}
 
 	ch <- prometheus.MustNewConstMetric(

internal/collector/diskdrive/diskdrive.go

@@ -31,6 +31,7 @@ const Name = "diskdrive"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector for a few WMI metrics in Win32_DiskDrive.
@@ -119,6 +120,11 @@ func (c *Collector) Build(_ *slog.Logger, miSession *mi.Session) error {
 		nil,
 	)
 
+	var dst []diskDrive
+	if err := c.miSession.Query(&dst, mi.NamespaceRootCIMv2, c.miQuery); err != nil {
+		return fmt.Errorf("WMI query failed: %w", err)
+	}
+
 	return nil
 }
 
@@ -133,6 +139,7 @@ type diskDrive struct {
 	Availability uint16 `mi:"Availability"`
 }
 
+//nolint:gochecknoglobals
 var (
 	allDiskStatus = []string{
 		"OK",

internal/collector/dns/dns.go

@@ -16,7 +16,6 @@
 package dns
 
 import (
-	"errors"
 	"fmt"
 	"log/slog"
 
@@ -31,6 +30,7 @@ const Name = "dns"
 
 type Config struct{}
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{}
 
 // A Collector is a Prometheus Collector for WMI Win32_PerfRawData_DNS_DNS metrics.
@@ -284,7 +284,7 @@ func (c *Collector) Collect(ch chan<- prometheus.Metric) error {
 
 	data, ok := perfData[perfdata.InstanceEmpty]
 	if !ok {
-		return errors.New("perflib query for DNS returned empty result set")
+		return fmt.Errorf("failed to collect DNS metrics: %w", types.ErrNoData)
 	}
 
 	ch <- prometheus.MustNewConstMetric(

internal/collector/exchange/exchange.go

@@ -47,6 +47,7 @@ type Config struct {
 	CollectorsEnabled []string `yaml:"collectors_enabled"`
 }
 
+//nolint:gochecknoglobals
 var ConfigDefaults = Config{
 	CollectorsEnabled: []string{
 		adAccessProcesses,

internal/collector/exchange/exchange_active_sync.go

@@ -16,7 +16,6 @@
 package exchange
 
 import (
-	"errors"
 	"fmt"
 
 	"github.com/prometheus-community/windows_exporter/internal/perfdata"
@@ -73,7 +72,7 @@ func (c *Collector) collectActiveSync(ch chan<- prometheus.Metric) error {
 	}
 
 	if len(perfData) == 0 {
-		return errors.New("perflib query for MSExchange ActiveSync returned empty result set")
+		return fmt.Errorf("failed to collect MSExchange ActiveSync metrics: %w", types.ErrNoData)
 	}
 
 	for _, data := range perfData {