Added support for ServiceMonitors

Signed-off-by: Juraci Paixão Kröhling <[email protected]>

Makefile

@@ -69,7 +69,7 @@ clean:
 
 .PHONY: crd
 crd:
-	@kubectl create -f deploy/crds/opentelemetry_v1alpha1_opentelemetrycollector_crd.yaml 2>&1 | grep -v "already exists" || true
+	@kubectl create -f deploy/crds/opentelemetry.io_opentelemetrycollectors_crd.yaml 2>&1 | grep -v "already exists" || true
 
 .PHONY: generate
 generate:
@@ -94,3 +94,8 @@ ci: install-tools ensure-generate-is-noop all
 install-tools:
 	@curl -sfL https://raw.githubusercontent.com/securego/gosec/master/install.sh | sh -s -- -b ${GOPATH}/bin 2.0.0
 	@go install golang.org/x/tools/cmd/goimports
+
+.PHONY: install-prometheus-operator
+install-prometheus-operator:
+	@echo Installing Prometheus Operator bundle
+	@kubectl apply -f https://raw.githubusercontent.com/coreos/prometheus-operator/master/bundle.yaml

README.md

@@ -10,7 +10,7 @@ At this point, it has [OpenTelemetry Collector](https://github.com/open-telemetr
 
 To install the operator, run:
 ```
-kubectl create -f https://raw.githubusercontent.com/jpkrohling/opentelemetry-operator/master/deploy/crds/opentelemetry_v1alpha1_opentelemetrycollector_crd.yaml
+kubectl create -f https://raw.githubusercontent.com/jpkrohling/opentelemetry-operator/master/deploy/crds/opentelemetry.io_opentelemetrycollectors_crd.yaml
 kubectl create -f https://raw.githubusercontent.com/jpkrohling/opentelemetry-operator/master/deploy/service_account.yaml
 kubectl create -f https://raw.githubusercontent.com/jpkrohling/opentelemetry-operator/master/deploy/role.yaml
 kubectl create -f https://raw.githubusercontent.com/jpkrohling/opentelemetry-operator/master/deploy/role_binding.yaml
@@ -54,6 +54,27 @@ At this point, the Operator does *not* validate the contents of the configuratio
 
 The `CustomResource` for the `OpenTelemetryCollector` exposes a property named `.Spec.Mode`, which can be used to specify whether the collector should run as a `DaemonSet` or as a `Deployment` (default). Look at the `examples/daemonset.yaml` for reference.
 
+## Prometheus ServiceMonitor objects
+
+When the Prometheus Operator is available in the same cluster as the OpenTelemetry Operator, the OpenTelemetry Operator will automatically create the relevant `ServiceMonitor` objects:
+
+* One set for the OpenTelemetry Operator itself
+* One set for each managed OpenTelemetry instance
+
+Refer to the Prometheus Operator for complete instructions on how to do a production-quality installation. For development purposes, the following will do:
+
+```console
+$ kubectl apply -f https://raw.githubusercontent.com/coreos/prometheus-operator/v0.33.0/bundle.yaml
+```
+
+When deploying the example `simplest.yaml`, the following `ServiceMonitor` will be created once the Prometheus Operator is available:
+
+```console
+$ kubectl get servicemonitors simplest-collector
+NAME                 AGE
+simplest-collector   103s
+```
+
 ## Contributing and Developing
 
 Please see [CONTRIBUTING.md](CONTRIBUTING.md).

cmd/manager/main.go

@@ -18,6 +18,7 @@ import (
 	"github.com/operator-framework/operator-sdk/pkg/metrics"
 	"github.com/operator-framework/operator-sdk/pkg/restmapper"
 	"github.com/spf13/pflag"
+	"github.com/spf13/viper"
 	v1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/util/intstr"
 	"sigs.k8s.io/controller-runtime/pkg/client/config"
@@ -54,6 +55,8 @@ func printVersion() {
 }
 
 func main() {
+	ctx := context.Background()
+
 	// Add the zap logger flag set to the CLI. The flag set must
 	// be added before calling pflag.Parse().
 	pflag.CommandLine.AddFlagSet(zap.FlagSet())
@@ -92,7 +95,6 @@ func main() {
 		os.Exit(1)
 	}
 
-	ctx := context.Background()
 	// Become the leader before proceeding
 	err = leader.Become(ctx, "opentelemetry-operator-lock")
 	if err != nil {
@@ -148,15 +150,18 @@ func main() {
 	// CreateServiceMonitors will automatically create the prometheus-operator ServiceMonitor resources
 	// necessary to configure Prometheus to scrape metrics from this operator.
 	services := []*v1.Service{service}
+	svcMonitorAvail := true
 	_, err = metrics.CreateServiceMonitors(cfg, namespace, services)
 	if err != nil {
 		log.Info("Could not create ServiceMonitor object", "error", err.Error())
 		// If this operator is deployed to a cluster without the prometheus-operator running, it will return
 		// ErrServiceMonitorNotPresent, which can be used to safely skip ServiceMonitor creation.
 		if err == metrics.ErrServiceMonitorNotPresent {
+			svcMonitorAvail = false
 			log.Info("Install prometheus-operator in your cluster to create ServiceMonitor objects", "error", err.Error())
 		}
 	}
+	viper.Set(opentelemetry.SvcMonitorAvailable, svcMonitorAvail)
 
 	log.Info("Starting the Cmd.")
 

go.mod

@@ -2,6 +2,7 @@ module github.com/open-telemetry/opentelemetry-operator
 
 require (
 	github.com/Azure/go-autorest v13.0.1+incompatible // indirect
+	github.com/coreos/prometheus-operator v0.29.0
 	github.com/go-logr/logr v0.1.0
 	github.com/go-openapi/spec v0.19.0
 	github.com/golang/protobuf v1.3.2 // indirect
@@ -20,6 +21,7 @@ require (
 	gopkg.in/check.v1 v1.0.0-20190902080502-41f04d3bba15 // indirect
 	gotest.tools v2.2.0+incompatible
 	k8s.io/api v0.0.0-20190918155943-95b840bb6a1f
+	k8s.io/apiextensions-apiserver v0.0.0-20190918161926-8f644eb6e783
 	k8s.io/apimachinery v0.0.0-20190913080033-27d36303b655
 	k8s.io/client-go v11.0.1-0.20190409021438-1a26190bd76a+incompatible
 	k8s.io/kube-openapi v0.0.0-20190401085232-94e1e7b7574c

go.sum

@@ -857,6 +857,7 @@ k8s.io/api v0.0.0-20190409021203-6e4e0e4f393b h1:aBGgKJUM9Hk/3AE8WaZIApnTxG35kbu
 k8s.io/api v0.0.0-20190409021203-6e4e0e4f393b/go.mod h1:iuAfoD4hCxJ8Onx9kaTIt30j7jUFS00AXQi6QMi99vA=
 k8s.io/apiextensions-apiserver v0.0.0-20190228180357-d002e88f6236 h1:JfFtjaElBIgYKCWEtYQkcNrTpW+lMO4GJy8NP6SVQmM=
 k8s.io/apiextensions-apiserver v0.0.0-20190228180357-d002e88f6236/go.mod h1:IxkesAMoaCRoLrPJdZNZUQp9NfZnzqaVzLhb2VEQzXE=
+k8s.io/apiextensions-apiserver v0.0.0-20190409022649-727a075fdec8 h1:q1Qvjzs/iEdXF6A1a8H3AKVFDzJNcJn3nXMs6R6qFtA=
 k8s.io/apiextensions-apiserver v0.0.0-20190409022649-727a075fdec8/go.mod h1:IxkesAMoaCRoLrPJdZNZUQp9NfZnzqaVzLhb2VEQzXE=
 k8s.io/apimachinery v0.0.0-20190221213512-86fb29eff628 h1:UYfHH+KEF88OTg+GojQUwFTNxbxwmoktLwutUzR0GPg=
 k8s.io/apimachinery v0.0.0-20190221213512-86fb29eff628/go.mod h1:ccL7Eh7zubPUSh9A3USN90/OzHNSVN6zxzde07TDCL0=
@@ -919,7 +920,9 @@ sigs.k8s.io/controller-runtime v0.2.0/go.mod h1:ZHqrRDZi3f6BzONcvlUxkqCKgwasGk5F
 sigs.k8s.io/controller-tools v0.1.10/go.mod h1:6g08p9m9G/So3sBc1AOQifHfhxH/mb6Sc4z0LMI8XMw=
 sigs.k8s.io/controller-tools v0.1.11-0.20190411181648-9d55346c2bde h1:ZkaHf5rNYzIB6CB82keKMQNv7xxkqT0ylOBdfJPfi+k=
 sigs.k8s.io/controller-tools v0.1.11-0.20190411181648-9d55346c2bde/go.mod h1:ATWLRP3WGxuAN9HcT2LaKHReXIH+EZGzRuMHuxjXfhQ=
+sigs.k8s.io/controller-tools v0.2.0 h1:AmQ/0JKBJAjyAiPAkrAf9QW06jkx2lc5hpxMjamsFpw=
 sigs.k8s.io/controller-tools v0.2.0/go.mod h1:8t/X+FVWvk6TaBcsa+UKUBbn7GMtvyBKX30SGl4em6Y=
+sigs.k8s.io/controller-tools v0.2.1 h1:HoCik83vXOpPi7KSJWdPRmiGntyOzK0v0BTV4U+pl8o=
 sigs.k8s.io/kustomize v2.0.3+incompatible/go.mod h1:MkjgH3RdOWrievjo6c9T245dYlB5QeXV4WCbnt/PEpU=
 sigs.k8s.io/structured-merge-diff v0.0.0-20190525122527-15d366b2352e/go.mod h1:wWxsB5ozmmv/SG7nM11ayaAW51xMvak/t1r0CSlcokI=
 sigs.k8s.io/testing_frameworks v0.1.1 h1:cP2l8fkA3O9vekpy5Ks8mmA0NW/F7yBdXf8brkWhVrs=

pkg/apis/opentelemetry/const.go

@@ -26,4 +26,7 @@ const (
 
 	// OtelColImageConfigKey represents the key to override the OpenTelemetry Collector container image
 	OtelColImageConfigKey = "otelcol-image"
+
+	// SvcMonitorAvailable represents the key that indicates whether a ServiceMonitor CRD is known to the cluster
+	SvcMonitorAvailable = "svc-monitor-avail"
 )

pkg/controller/opentelemetrycollector/labels.go

@@ -11,9 +11,13 @@ import (
 // commonLabels return the common labels to all objects that are part of a managed OpenTelemetryCollector
 func commonLabels(ctx context.Context) map[string]string {
 	instance := ctx.Value(opentelemetry.ContextInstance).(*v1alpha1.OpenTelemetryCollector)
-	base := instance.Labels
-	if nil == base {
-		base = map[string]string{}
+
+	// new map every time, so that we don't touch the instance's label
+	base := map[string]string{}
+	if nil != instance.Labels {
+		for k, v := range instance.Labels {
+			base[k] = v
+		}
 	}
 
 	base["app.kubernetes.io/managed-by"] = "opentelemetry-operator"

pkg/controller/opentelemetrycollector/main_test.go

@@ -8,20 +8,16 @@ import (
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/runtime"
 	"k8s.io/client-go/kubernetes/scheme"
-	"sigs.k8s.io/controller-runtime/pkg/client"
-	"sigs.k8s.io/controller-runtime/pkg/client/fake"
 	logf "sigs.k8s.io/controller-runtime/pkg/runtime/log"
 
 	"github.com/open-telemetry/opentelemetry-operator/pkg/apis/opentelemetry"
 	"github.com/open-telemetry/opentelemetry-operator/pkg/apis/opentelemetry/v1alpha1"
 )
 
 var (
-	instance   *v1alpha1.OpenTelemetryCollector
-	ctx        context.Context
-	reconciler *ReconcileOpenTelemetryCollector
-	schem      *runtime.Scheme
-	cl         client.Client
+	instance *v1alpha1.OpenTelemetryCollector
+	ctx      context.Context
+	schem    *runtime.Scheme
 )
 
 // TestMain ensures that all tests in this package have a fresh and sane instance of the common resources
@@ -49,8 +45,5 @@ func TestMain(m *testing.M) {
 	ctx = context.WithValue(context.Background(), opentelemetry.ContextInstance, instance)
 	ctx = context.WithValue(ctx, opentelemetry.ContextLogger, logf.Log.WithName("unit-tests"))
 
-	cl = fake.NewFakeClient(instance)
-	reconciler = New(cl, schem)
-
 	os.Exit(m.Run())
 }

pkg/controller/opentelemetrycollector/opentelemetrycollector_controller.go

@@ -16,12 +16,7 @@ var log = logf.Log.WithName("controller_opentelemetrycollector")
 // Add creates a new OpenTelemetryCollector Controller and adds it to the Manager. The Manager will set fields on the Controller
 // and Start it when the Manager is Started.
 func Add(mgr manager.Manager) error {
-	return add(mgr, newReconciler(mgr))
-}
-
-// newReconciler returns a new reconcile.Reconciler
-func newReconciler(mgr manager.Manager) reconcile.Reconciler {
-	return New(mgr.GetClient(), mgr.GetScheme())
+	return add(mgr, WithManager(mgr))
 }
 
 // add adds a new Controller to mgr with r as the reconcile.Reconciler

pkg/controller/opentelemetrycollector/reconcile.go

@@ -5,11 +5,13 @@ import (
 	"fmt"
 	"time"
 
+	monclientv1 "github.com/coreos/prometheus-operator/pkg/client/versioned/typed/monitoring/v1"
 	"k8s.io/apimachinery/pkg/api/errors"
 	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/runtime"
 	"sigs.k8s.io/controller-runtime/pkg/client"
 	"sigs.k8s.io/controller-runtime/pkg/controller/controllerutil"
+	"sigs.k8s.io/controller-runtime/pkg/manager"
 	"sigs.k8s.io/controller-runtime/pkg/reconcile"
 
 	"github.com/open-telemetry/opentelemetry-operator/pkg/apis/opentelemetry"
@@ -23,24 +25,46 @@ var _ reconcile.Reconciler = &ReconcileOpenTelemetryCollector{}
 type ReconcileOpenTelemetryCollector struct {
 	// This client, initialized using mgr.Client() above, is a split client
 	// that reads objects from the cache and writes to the apiserver
-	client client.Client
 	scheme *runtime.Scheme
 
+	// the clients that compose this reconciler
+	clients *Clients
+
 	// the list of reconciliation functions to execute
 	reconcileFuncs []func(context.Context) error
 }
 
+// Clients holds all the clients that the reconciler might need to hold
+type Clients struct {
+	client client.Client
+
+	// some places are using the rich REST client
+	monclient monclientv1.MonitoringV1Interface
+}
+
+// WithManager creates a new reconciler based on the manager information
+func WithManager(manager manager.Manager) *ReconcileOpenTelemetryCollector {
+	monclient := monclientv1.NewForConfigOrDie(manager.GetConfig())
+	clients := &Clients{
+		client:    manager.GetClient(),
+		monclient: monclient,
+	}
+
+	return New(manager.GetScheme(), clients)
+}
+
 // New constructs a ReconcileOpenTelemetryCollector based on the client and scheme, with the default reconciliation functions
-func New(client client.Client, scheme *runtime.Scheme) *ReconcileOpenTelemetryCollector {
+func New(scheme *runtime.Scheme, clients *Clients) *ReconcileOpenTelemetryCollector {
 	r := &ReconcileOpenTelemetryCollector{
-		client: client,
-		scheme: scheme,
+		scheme:  scheme,
+		clients: clients,
 	}
 	r.reconcileFuncs = []func(context.Context) error{
 		r.reconcileConfigMap,
 		r.reconcileService,
 		r.reconcileDeployment,
 		r.reconcileDaemonSet,
+		r.reconcileServiceMonitor,
 	}
 
 	return r
@@ -58,7 +82,7 @@ func (r *ReconcileOpenTelemetryCollector) Reconcile(request reconcile.Request) (
 
 	// Fetch the OpenTelemetryCollector instance
 	instance := &v1alpha1.OpenTelemetryCollector{}
-	err := r.client.Get(context.Background(), request.NamespacedName, instance)
+	err := r.clients.client.Get(context.Background(), request.NamespacedName, instance)
 	if err != nil {
 		if errors.IsNotFound(err) {
 			// Request object not found, could have been deleted after reconcile request.
@@ -86,13 +110,13 @@ func (r *ReconcileOpenTelemetryCollector) Reconcile(request reconcile.Request) (
 	}
 
 	// update the status object, which might have also been updated
-	if err := r.client.Status().Update(ctx, instance); err != nil {
+	if err := r.clients.client.Status().Update(ctx, instance); err != nil {
 		reqLogger.Error(err, "failed to store the custom resource's status")
 		return reconcile.Result{}, err
 	}
 
 	// apply it back, as it might have been updated
-	if err := r.client.Update(ctx, instance); err != nil {
+	if err := r.clients.client.Update(ctx, instance); err != nil {
 		reqLogger.Error(err, "failed to store back the custom resource")
 		return reconcile.Result{}, err
 	}

pkg/controller/opentelemetrycollector/reconcile_configmap.go

@@ -61,9 +61,9 @@ func (r *ReconcileOpenTelemetryCollector) reconcileExpectedConfigMaps(ctx contex
 		r.setControllerReference(ctx, desired)
 
 		existing := &corev1.ConfigMap{}
-		err := r.client.Get(ctx, types.NamespacedName{Name: desired.Name, Namespace: desired.Namespace}, existing)
+		err := r.clients.client.Get(ctx, types.NamespacedName{Name: desired.Name, Namespace: desired.Namespace}, existing)
 		if err != nil && errors.IsNotFound(err) {
-			if err := r.client.Create(ctx, desired); err != nil {
+			if err := r.clients.client.Create(ctx, desired); err != nil {
 				return fmt.Errorf("failed to create: %v", err)
 			}
 
@@ -93,7 +93,7 @@ func (r *ReconcileOpenTelemetryCollector) reconcileExpectedConfigMaps(ctx contex
 			updated.ObjectMeta.Labels[k] = v
 		}
 
-		if err := r.client.Update(ctx, updated); err != nil {
+		if err := r.clients.client.Update(ctx, updated); err != nil {
 			return fmt.Errorf("failed to apply changes: %v", err)
 		}
 		logger.V(2).Info("applied", "configmap.name", desired.Name, "configmap.namespace", desired.Namespace)
@@ -114,7 +114,7 @@ func (r *ReconcileOpenTelemetryCollector) deleteConfigMaps(ctx context.Context,
 		}),
 	}
 	list := &corev1.ConfigMapList{}
-	if err := r.client.List(ctx, list, opts...); err != nil {
+	if err := r.clients.client.List(ctx, list, opts...); err != nil {
 		return fmt.Errorf("failed to list: %v", err)
 	}
 
@@ -127,7 +127,7 @@ func (r *ReconcileOpenTelemetryCollector) deleteConfigMaps(ctx context.Context,
 		}
 
 		if del {
-			if err := r.client.Delete(ctx, &existing); err != nil {
+			if err := r.clients.client.Delete(ctx, &existing); err != nil {
 				return fmt.Errorf("failed to delete: %v", err)
 			}
 			logger.V(2).Info("deleted", "configmap.name", existing.Name, "configmap.namespace", existing.Namespace)