Extend controller-runtime with metrics from issue spiffe#278

Signed-off-by: Alexandre Alvino <[email protected]>

cmd/main.go

@@ -42,11 +42,13 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/healthz"
 	"sigs.k8s.io/controller-runtime/pkg/log/zap"
 	"sigs.k8s.io/controller-runtime/pkg/manager"
+	k8sMetrics "sigs.k8s.io/controller-runtime/pkg/metrics"
 	"sigs.k8s.io/controller-runtime/pkg/webhook"
 
 	"github.com/spiffe/go-spiffe/v2/spiffeid"
 	spirev1alpha1 "github.com/spiffe/spire-controller-manager/api/v1alpha1"
 	"github.com/spiffe/spire-controller-manager/internal/controller"
+	"github.com/spiffe/spire-controller-manager/pkg/metrics"
 	"github.com/spiffe/spire-controller-manager/pkg/reconciler"
 	"github.com/spiffe/spire-controller-manager/pkg/spireapi"
 	"github.com/spiffe/spire-controller-manager/pkg/spireentry"
@@ -78,6 +80,10 @@ func init() {
 	utilruntime.Must(clientgoscheme.AddToScheme(scheme))
 
 	utilruntime.Must(spirev1alpha1.AddToScheme(scheme))
+
+	k8sMetrics.Registry.MustRegister(
+		metrics.PromCounters[metrics.StaticEntryFailures],
+	)
 	//+kubebuilder:scaffold:scheme
 }
 

go.mod

@@ -9,6 +9,7 @@ require (
 	github.com/jpillora/backoff v1.0.0
 	github.com/onsi/ginkgo/v2 v2.17.3
 	github.com/onsi/gomega v1.33.1
+	github.com/prometheus/client_golang v1.19.0
 	github.com/spiffe/go-spiffe/v2 v2.2.0
 	github.com/spiffe/spire-api-sdk v1.9.4
 	github.com/stretchr/testify v1.9.0
@@ -51,7 +52,6 @@ require (
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
-	github.com/prometheus/client_golang v1.19.0 // indirect
 	github.com/prometheus/client_model v0.6.0 // indirect
 	github.com/prometheus/common v0.51.1 // indirect
 	github.com/prometheus/procfs v0.13.0 // indirect

pkg/metrics/controller_runtime.go

@@ -0,0 +1,18 @@
+package metrics
+
+import "github.com/prometheus/client_golang/prometheus"
+
+const (
+	StaticEntryFailures = "cluster_static_entry_failures"
+)
+
+var (
+	PromCounters = map[string]prometheus.Counter{
+		StaticEntryFailures: prometheus.NewGauge(
+			prometheus.GaugeOpts{
+				Name: StaticEntryFailures,
+				Help: "Number of cluster static entry render failures",
+			},
+		),
+	}
+)

pkg/spireentry/reconciler.go

@@ -21,6 +21,7 @@ import (
 	"crypto/sha256"
 	"encoding/hex"
 	"fmt"
+	"github.com/prometheus/client_golang/prometheus"
 	"io"
 	"regexp"
 	"sort"
@@ -41,6 +42,7 @@ import (
 
 	spirev1alpha1 "github.com/spiffe/spire-controller-manager/api/v1alpha1"
 	"github.com/spiffe/spire-controller-manager/pkg/k8sapi"
+	"github.com/spiffe/spire-controller-manager/pkg/metrics"
 	"github.com/spiffe/spire-controller-manager/pkg/namespace"
 	"github.com/spiffe/spire-controller-manager/pkg/reconciler"
 	"github.com/spiffe/spire-controller-manager/pkg/spireapi"
@@ -80,7 +82,8 @@ type ReconcilerConfig struct {
 
 func Reconciler(config ReconcilerConfig) reconciler.Reconciler {
 	r := &entryReconciler{
-		config: config,
+		config:      config,
+		promCounter: metrics.PromCounters,
 	}
 	return reconciler.New(reconciler.Config{
 		Kind:       "entry",
@@ -93,6 +96,7 @@ type entryReconciler struct {
 	config ReconcilerConfig
 
 	unsupportedFields        map[spireapi.Field]struct{}
+	promCounter              map[string]prometheus.Counter
 	nextGetUnsupportedFields time.Time
 }
 
@@ -349,6 +353,7 @@ func (r *entryReconciler) addClusterStaticEntryEntriesState(ctx context.Context,
 		if err != nil {
 			log.Error(err, "Failed to render ClusterStaticEntry")
 			clusterStaticEntry.NextStatus.Rendered = false
+			r.promCounter[metrics.StaticEntryFailures].Add(1)
 			continue
 		}
 		clusterStaticEntry.NextStatus.Rendered = true