add kcp operator custom metrics

cmd/main.go

@@ -41,6 +41,7 @@ import (
 	"github.com/kcp-dev/kcp-operator/internal/controller/kubeconfig"
 	"github.com/kcp-dev/kcp-operator/internal/controller/rootshard"
 	"github.com/kcp-dev/kcp-operator/internal/controller/shard"
+	"github.com/kcp-dev/kcp-operator/internal/metrics"
 	"github.com/kcp-dev/kcp-operator/internal/reconciling"
 	operatorv1alpha1 "github.com/kcp-dev/kcp-operator/sdk/apis/operator/v1alpha1"
 )
@@ -190,6 +191,12 @@ func main() {
 	}
 	// +kubebuilder:scaffold:builder
 
+	metrics.RegisterMetrics()
+
+	metricsCollector := metrics.NewMetricsCollector(mgr.GetClient())
+	ctx := ctrl.SetupSignalHandler()
+	go metricsCollector.Start(ctx)
+
 	if err := mgr.AddHealthzCheck("healthz", healthz.Ping); err != nil {
 		setupLog.Error(err, "unable to set up health check")
 		os.Exit(1)
@@ -200,7 +207,7 @@ func main() {
 	}
 
 	setupLog.Info("starting manager")
-	if err := mgr.Start(ctrl.SetupSignalHandler()); err != nil {
+	if err := mgr.Start(ctx); err != nil {
 		setupLog.Error(err, "problem running manager")
 		os.Exit(1)
 	}

go.mod

@@ -13,6 +13,7 @@ require (
 	github.com/kcp-dev/kcp-operator/sdk v0.0.0-00010101000000-000000000000
 	github.com/kcp-dev/kcp/sdk v0.27.1
 	github.com/kcp-dev/logicalcluster/v3 v3.0.5
+	github.com/prometheus/client_golang v1.20.5
 	github.com/stretchr/testify v1.10.0
 	go.uber.org/zap v1.27.0
 	k8c.io/reconciler v0.5.0
@@ -68,7 +69,6 @@ require (
 	github.com/onsi/gomega v1.35.1 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
 	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
-	github.com/prometheus/client_golang v1.20.5 // indirect
 	github.com/prometheus/client_model v0.6.1 // indirect
 	github.com/prometheus/common v0.61.0 // indirect
 	github.com/prometheus/procfs v0.15.1 // indirect

internal/controller/frontproxy/controller.go

@@ -19,6 +19,7 @@ package frontproxy
 import (
 	"context"
 	"fmt"
+	"time"
 
 	certmanagerv1 "github.com/cert-manager/cert-manager/pkg/apis/certmanager/v1"
 
@@ -38,6 +39,7 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/reconcile"
 
 	"github.com/kcp-dev/kcp-operator/internal/controller/util"
+	"github.com/kcp-dev/kcp-operator/internal/metrics"
 	"github.com/kcp-dev/kcp-operator/internal/resources"
 	"github.com/kcp-dev/kcp-operator/internal/resources/frontproxy"
 	operatorv1alpha1 "github.com/kcp-dev/kcp-operator/sdk/apis/operator/v1alpha1"
@@ -89,12 +91,19 @@ func (r *FrontProxyReconciler) SetupWithManager(mgr ctrl.Manager) error {
 // +kubebuilder:rbac:groups=core,resources=services;configmaps;secrets,verbs=get;list;watch;create;update;patch;delete
 
 func (r *FrontProxyReconciler) Reconcile(ctx context.Context, req ctrl.Request) (res ctrl.Result, recErr error) {
+	startTime := time.Now()
+	defer func() {
+		duration := time.Since(startTime)
+		metrics.RecordReconciliationMetrics(metrics.FrontProxyResourceType, duration.Seconds(), recErr)
+	}()
+
 	logger := log.FromContext(ctx)
 	logger.V(4).Info("Reconciling")
 
 	var frontProxy operatorv1alpha1.FrontProxy
 	if err := r.Get(ctx, req.NamespacedName, &frontProxy); err != nil {
 		if ctrlruntimeclient.IgnoreNotFound(err) != nil {
+			metrics.RecordReconciliationError(metrics.FrontProxyResourceType, err.Error())
 			return ctrl.Result{}, fmt.Errorf("failed to get FrontProxy object: %w", err)
 		}
 
@@ -108,6 +117,14 @@ func (r *FrontProxyReconciler) Reconcile(ctx context.Context, req ctrl.Request)
 		recErr = kerrors.NewAggregate([]error{recErr, err})
 	}
 
+	metrics.RecordObjectMetrics(
+		metrics.FrontProxyResourceType,
+		frontProxy.Name,
+		req.Namespace,
+		string(frontProxy.Status.Phase),
+		frontProxy.Status.Conditions,
+	)
+
 	return ctrl.Result{}, recErr
 }
 

internal/controller/kubeconfig/controller.go

@@ -40,6 +40,7 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/log"
 
 	"github.com/kcp-dev/kcp-operator/internal/controller/util"
+	"github.com/kcp-dev/kcp-operator/internal/metrics"
 	"github.com/kcp-dev/kcp-operator/internal/reconciling"
 	"github.com/kcp-dev/kcp-operator/internal/resources"
 	"github.com/kcp-dev/kcp-operator/internal/resources/kubeconfig"
@@ -74,6 +75,12 @@ func (r *KubeconfigReconciler) SetupWithManager(mgr ctrl.Manager) error {
 // Reconcile is part of the main kubernetes reconciliation loop which aims to
 // move the current state of the cluster closer to the desired state.
 func (r *KubeconfigReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ctrl.Result, error) {
+	startTime := time.Now()
+	defer func() {
+		duration := time.Since(startTime)
+		metrics.RecordReconciliationMetrics(metrics.KubeconfigResourceType, duration.Seconds(), nil)
+	}()
+
 	logger := log.FromContext(ctx)
 	logger.V(4).Info("Reconciling")
 
@@ -83,6 +90,7 @@ func (r *KubeconfigReconciler) Reconcile(ctx context.Context, req ctrl.Request)
 		if apierrors.IsNotFound(err) {
 			return ctrl.Result{}, nil
 		}
+		metrics.RecordReconciliationError(metrics.KubeconfigResourceType, err.Error())
 		return ctrl.Result{}, err
 	}
 
@@ -113,6 +121,14 @@ func (r *KubeconfigReconciler) Reconcile(ctx context.Context, req ctrl.Request)
 		recErr = kerrors.NewAggregate([]error{recErr, err})
 	}
 
+	metrics.RecordObjectMetrics(
+		metrics.KubeconfigResourceType,
+		kc.Name,
+		req.Namespace,
+		string(kc.Status.Phase),
+		kc.Status.Conditions,
+	)
+
 	return ctrl.Result{}, recErr
 }
 

internal/controller/rootshard/controller.go

@@ -20,6 +20,7 @@ import (
 	"context"
 	"fmt"
 	"sort"
+	"time"
 
 	certmanagerv1 "github.com/cert-manager/cert-manager/pkg/apis/certmanager/v1"
 	k8creconciling "k8c.io/reconciler/pkg/reconciling"
@@ -40,6 +41,7 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/reconcile"
 
 	"github.com/kcp-dev/kcp-operator/internal/controller/util"
+	"github.com/kcp-dev/kcp-operator/internal/metrics"
 	"github.com/kcp-dev/kcp-operator/internal/reconciling"
 	"github.com/kcp-dev/kcp-operator/internal/resources"
 	"github.com/kcp-dev/kcp-operator/internal/resources/frontproxy"
@@ -99,12 +101,19 @@ func (r *RootShardReconciler) SetupWithManager(mgr ctrl.Manager) error {
 // For more details, check Reconcile and its Result here:
 // - https://pkg.go.dev/sigs.k8s.io/[email protected]/pkg/reconcile
 func (r *RootShardReconciler) Reconcile(ctx context.Context, req ctrl.Request) (res ctrl.Result, recErr error) {
+	startTime := time.Now()
+	defer func() {
+		duration := time.Since(startTime)
+		metrics.RecordReconciliationMetrics(metrics.RootShardResourceType, duration.Seconds(), recErr)
+	}()
+
 	logger := log.FromContext(ctx)
 	logger.V(4).Info("Reconciling")
 
 	var rootShard operatorv1alpha1.RootShard
 	if err := r.Get(ctx, req.NamespacedName, &rootShard); err != nil {
 		if ctrlruntimeclient.IgnoreNotFound(err) != nil {
+			metrics.RecordReconciliationError(metrics.RootShardResourceType, err.Error())
 			return ctrl.Result{}, fmt.Errorf("failed to find %s/%s: %w", req.Namespace, req.Name, err)
 		}
 
@@ -118,6 +127,14 @@ func (r *RootShardReconciler) Reconcile(ctx context.Context, req ctrl.Request) (
 		recErr = kerrors.NewAggregate([]error{recErr, err})
 	}
 
+	metrics.RecordObjectMetrics(
+		metrics.RootShardResourceType,
+		rootShard.Name,
+		req.Namespace,
+		string(rootShard.Status.Phase),
+		rootShard.Status.Conditions,
+	)
+
 	return ctrl.Result{}, recErr
 }
 

internal/controller/shard/controller.go

@@ -19,6 +19,7 @@ package shard
 import (
 	"context"
 	"fmt"
+	"time"
 
 	certmanagerv1 "github.com/cert-manager/cert-manager/pkg/apis/certmanager/v1"
 	k8creconciling "k8c.io/reconciler/pkg/reconciling"
@@ -39,6 +40,7 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/reconcile"
 
 	"github.com/kcp-dev/kcp-operator/internal/controller/util"
+	"github.com/kcp-dev/kcp-operator/internal/metrics"
 	"github.com/kcp-dev/kcp-operator/internal/reconciling"
 	"github.com/kcp-dev/kcp-operator/internal/resources"
 	"github.com/kcp-dev/kcp-operator/internal/resources/shard"
@@ -90,12 +92,19 @@ func (r *ShardReconciler) SetupWithManager(mgr ctrl.Manager) error {
 // +kubebuilder:rbac:groups=core,resources=secrets;services,verbs=get;list;watch;create;update;patch;delete
 
 func (r *ShardReconciler) Reconcile(ctx context.Context, req ctrl.Request) (res ctrl.Result, recErr error) {
+	startTime := time.Now()
+	defer func() {
+		duration := time.Since(startTime)
+		metrics.RecordReconciliationMetrics(metrics.ShardResourceType, duration.Seconds(), recErr)
+	}()
+
 	logger := log.FromContext(ctx)
 	logger.V(4).Info("Reconciling Shard object")
 
 	var s operatorv1alpha1.Shard
 	if err := r.Get(ctx, req.NamespacedName, &s); err != nil {
 		if ctrlruntimeclient.IgnoreNotFound(err) != nil {
+			metrics.RecordReconciliationError(metrics.ShardResourceType, err.Error())
 			return ctrl.Result{}, fmt.Errorf("failed to get shard: %w", err)
 		}
 
@@ -108,6 +117,14 @@ func (r *ShardReconciler) Reconcile(ctx context.Context, req ctrl.Request) (res
 		recErr = kerrors.NewAggregate([]error{recErr, err})
 	}
 
+	metrics.RecordObjectMetrics(
+		metrics.ShardResourceType,
+		s.Name,
+		req.Namespace,
+		string(s.Status.Phase),
+		s.Status.Conditions,
+	)
+
 	return ctrl.Result{}, recErr
 }