Add Go and Process metrics

Signed-off-by: Richard Wall <richard.wall@venafi.com>

cmd/app/app.go

@@ -23,16 +23,20 @@ import (
 	"crypto/x509"
 	"encoding/pem"
 	"fmt"
+	"time"
 
+	"github.com/cert-manager/cert-manager/pkg/server"
 	"github.com/cert-manager/csi-lib/driver"
 	"github.com/cert-manager/csi-lib/manager"
 	"github.com/cert-manager/csi-lib/manager/util"
 	"github.com/cert-manager/csi-lib/metadata"
 	"github.com/cert-manager/csi-lib/storage"
 	"github.com/spf13/cobra"
+	"golang.org/x/sync/errgroup"
 	"k8s.io/utils/clock"
 
 	"github.com/cert-manager/csi-driver/cmd/app/options"
+	"github.com/cert-manager/csi-driver/internal/metrics"
 	"github.com/cert-manager/csi-driver/internal/version"
 	csiapi "github.com/cert-manager/csi-driver/pkg/apis/v1alpha1"
 	"github.com/cert-manager/csi-driver/pkg/filestore"
@@ -96,18 +100,46 @@ func NewCommand(ctx context.Context) *cobra.Command {
 				return fmt.Errorf("failed to setup driver: " + err.Error())
 			}
 
-			go func() {
+			const metricsListenAddress = ":8080"
+			// Start metrics server
+			metricsLn, err := server.Listen("tcp", metricsListenAddress)
+			if err != nil {
+				return fmt.Errorf("failed to listen on prometheus address %s: %v", metricsListenAddress, err)
+			}
+			metricsServer := metrics.NewServer(metricsLn)
+
+			g, _ := errgroup.WithContext(ctx)
+			g.Go(func() error {
 				<-ctx.Done()
 				log.Info("shutting down driver", "context", ctx.Err())
 				d.Stop()
-			}()
+				return nil
+			})
 
-			log.Info("running driver")
-			if err := d.Run(); err != nil {
-				return fmt.Errorf("failed running driver: " + err.Error())
-			}
+			g.Go(func() error {
+				log.Info("running driver")
+				if err := d.Run(); err != nil {
+					return fmt.Errorf("failed running driver: " + err.Error())
+				}
+				return nil
+			})
+
+			g.Go(func() error {
+				<-ctx.Done()
+				// allow a timeout for graceful shutdown
+				shutdownCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+				defer cancel()
+
+				// nolint: contextcheck
+				return metricsServer.Shutdown(shutdownCtx)
+			})
+
+			g.Go(func() error {
+				log.V(3).Info("starting metrics server", "address", metricsLn.Addr())
+				return metricsServer.Serve(metricsLn)
+			})
 
-			return nil
+			return g.Wait()
 		},
 	}
 

go.mod

@@ -8,9 +8,11 @@ require (
 	github.com/go-logr/logr v1.4.2
 	github.com/onsi/ginkgo/v2 v2.19.0
 	github.com/onsi/gomega v1.33.1
+	github.com/prometheus/client_golang v1.18.0
 	github.com/spf13/cobra v1.8.1
 	github.com/spf13/pflag v1.0.5
 	github.com/stretchr/testify v1.9.0
+	golang.org/x/sync v0.7.0
 	k8s.io/api v0.30.2
 	k8s.io/apimachinery v0.30.2
 	k8s.io/cli-runtime v0.30.2
@@ -70,7 +72,6 @@ require (
 	github.com/peterbourgon/diskv v2.0.1+incompatible // 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 v1.18.0 // indirect
 	github.com/prometheus/client_model v0.6.1 // indirect
 	github.com/prometheus/common v0.46.0 // indirect
 	github.com/prometheus/procfs v0.15.0 // indirect
@@ -81,7 +82,6 @@ require (
 	golang.org/x/crypto v0.23.0 // indirect
 	golang.org/x/net v0.25.0 // indirect
 	golang.org/x/oauth2 v0.20.0 // indirect
-	golang.org/x/sync v0.7.0 // indirect
 	golang.org/x/sys v0.20.0 // indirect
 	golang.org/x/term v0.20.0 // indirect
 	golang.org/x/text v0.15.0 // indirect

internal/metrics/metrics.go

@@ -0,0 +1,37 @@
+package metrics
+
+import (
+	"net"
+	"net/http"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/collectors"
+	"github.com/prometheus/client_golang/prometheus/promhttp"
+)
+
+const (
+	prometheusMetricsServerReadTimeout    = 8 * time.Second
+	prometheusMetricsServerWriteTimeout   = 8 * time.Second
+	prometheusMetricsServerMaxHeaderBytes = 1 << 20 // 1 MiB
+)
+
+func NewServer(ln net.Listener) *http.Server {
+	registry := prometheus.NewRegistry()
+	registry.MustRegister(
+		collectors.NewProcessCollector(collectors.ProcessCollectorOpts{}),
+		collectors.NewGoCollector(),
+	)
+
+	mux := http.NewServeMux()
+	mux.Handle("/metrics", promhttp.HandlerFor(registry, promhttp.HandlerOpts{}))
+
+	server := &http.Server{
+		Addr:           ln.Addr().String(),
+		ReadTimeout:    prometheusMetricsServerReadTimeout,
+		WriteTimeout:   prometheusMetricsServerWriteTimeout,
+		MaxHeaderBytes: prometheusMetricsServerMaxHeaderBytes,
+		Handler:        mux,
+	}
+	return server
+}