Change the way resource attributes are populated (unbreaks a 2.0 chan…

…ge) (#1565)

* Change the way resource attributes is populated

* fix integration tests

* fixed tests

docs/sources/configure/service-discovery.md

@@ -321,39 +321,20 @@ Beyla uses the following criteria in this order to automatically set the service
     - `k8s.container.name`
 5. The executable name of the instrumented process.
 
-The Kubernetes annotations and labels from the previous bullets 2 and 3 can be overridden and via
-configuration.
+The Kubernetes labels from the previous bullet 3 can be overridden via configuration.
 
 In YAML:
 ```yaml
 kubernetes:
-  meta_naming_sources:
-      annotations:
-        service_name:
-          # gets service name from the first existing Pod annotation
-          - my.domain.com/override-service-name
-          - resource.opentelemetry.io/service.name
-        service_namespace:
-          # gets service namespace from the first existing Pod annotation
-          - my.domain.com/override-service-namespace
-          - resource.opentelemetry.io/service.namespace
-      labels:
-        service_name:
-          # gets service name from the first existing Pod label
-          - override-svc-name
-          - app.kubernetes.io/name
-        service_namespace:
-          # gets service namespace from the first existing Pod label
-          - override-svc-ns
-          - app.kubernetes.io/part-of
+  resource_labels:
+    service.name:
+      # gets service name from the first existing Pod label
+      - override-svc-name
+      - app.kubernetes.io/name
+    service.namespace:
+      # gets service namespace from the first existing Pod label
+      - override-svc-ns
+      - app.kubernetes.io/part-of
 ```
 
-The equivalent environment variables for the labels and annotation overriding
-properties are:
-
-* `BEYLA_KUBE_ANNOTATIONS_SERVICE_NAME`
-* `BEYLA_KUBE_ANNOTATIONS_SERVICE_NAMESPACE`
-* `BEYLA_KUBE_LABELS_SERVICE_NAME`
-* `BEYLA_KUBE_LABELS_SERVICE_NAMESPACE`
-
 They accept a comma-separated list of annotation and label names.

pkg/beyla/config.go

@@ -115,7 +115,7 @@ var DefaultConfig = Config{
 			Enable:                kubeflags.EnabledDefault,
 			InformersSyncTimeout:  30 * time.Second,
 			InformersResyncPeriod: 30 * time.Minute,
-			MetadataSources:       kube.DefaultMetadataSources,
+			ResourceLabels:        kube.DefaultResourceLabels,
 		},
 		HostID: HostIDConfig{
 			FetchTimeout: 500 * time.Millisecond,

pkg/beyla/config_test.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"log/slog"
+	"maps"
 	"os"
 	"regexp"
 	"strings"
@@ -55,11 +56,8 @@ attributes:
     kubeconfig_path: /foo/bar
     enable: true
     informers_sync_timeout: 30s
-    meta_naming_sources:
-      annotations:
-        service_namespace: ["huha.com/yeah"]
-      labels:
-        service_name: ["titi.com/lala"]
+    resource_labels:
+      service.namespace: ["huha.com/yeah"]
   instance_id:
     dns: true
   host_id:
@@ -108,9 +106,8 @@ network:
 	nc.AgentIP = "1.2.3.4"
 	nc.CIDRs = cidr.Definitions{"10.244.0.0/16"}
 
-	metaSources := kube.DefaultMetadataSources
-	metaSources.Annotations.ServiceNamespace = []string{"huha.com/yeah"}
-	metaSources.Labels.ServiceName = []string{"titi.com/lala"}
+	metaSources := maps.Clone(kube.DefaultResourceLabels)
+	metaSources["service.namespace"] = []string{"huha.com/yeah"}
 
 	assert.Equal(t, &Config{
 		Exec:             cfg.Exec,
@@ -190,7 +187,7 @@ network:
 				Enable:                kubeflags.EnabledTrue,
 				InformersSyncTimeout:  30 * time.Second,
 				InformersResyncPeriod: 30 * time.Minute,
-				MetadataSources:       metaSources,
+				ResourceLabels:        metaSources,
 			},
 			HostID: HostIDConfig{
 				Override:     "the-host-id",

pkg/components/beyla.go

@@ -116,6 +116,26 @@ func mustSkip(cfg *beyla.Config) string {
 func buildCommonContextInfo(
 	ctx context.Context, config *beyla.Config,
 ) (*global.ContextInfo, error) {
+
+	// merging deprecated resource labels definition for backwards compatibility
+	resourceLabels := config.Attributes.Kubernetes.ResourceLabels
+	if resourceLabels == nil {
+		resourceLabels = map[string][]string{}
+	}
+	showDeprecation := sync.OnceFunc(func() {
+		slog.Warn("The meta_source_labels (BEYLA_KUBE_META_SOURCE_LABEL_* environment variables) is deprecated." +
+			" Check the documentation for more information about replacing it by the resource_labels kubernetes" +
+			" YAML property")
+	})
+	if svc := config.Attributes.Kubernetes.MetaSourceLabels.ServiceName; svc != "" {
+		resourceLabels["service.name"] = append([]string{svc}, resourceLabels["service.name"]...)
+		showDeprecation()
+	}
+	if ns := config.Attributes.Kubernetes.MetaSourceLabels.ServiceNamespace; ns != "" {
+		resourceLabels["service.namespace"] = append([]string{ns}, resourceLabels["service.namespace"]...)
+		showDeprecation()
+	}
+
 	promMgr := &connector.PrometheusManager{}
 	ctxInfo := &global.ContextInfo{
 		Prometheus: promMgr,
@@ -126,7 +146,7 @@ func buildCommonContextInfo(
 			ResyncPeriod:      config.Attributes.Kubernetes.InformersResyncPeriod,
 			DisabledInformers: config.Attributes.Kubernetes.DisableInformers,
 			MetaCacheAddr:     config.Attributes.Kubernetes.MetaCacheAddress,
-			MetadataSources:   config.Attributes.Kubernetes.MetadataSources,
+			ResourceLabels:    resourceLabels,
 			RestrictLocalNode: config.Attributes.Kubernetes.MetaRestrictLocalNode,
 		}),
 	}

pkg/internal/discover/watcher_kube_test.go

@@ -67,7 +67,7 @@ func TestWatcherKubeEnricher(t *testing.T) {
 
 			// Setup a fake K8s API connected to the watcherKubeEnricher
 			fInformer := &fakeInformer{}
-			store := kube.NewStore(fInformer, kube.MetadataSources{})
+			store := kube.NewStore(fInformer, kube.ResourceLabels{})
 			wkeNodeFunc, err := WatcherKubeEnricherProvider(context.TODO(), &fakeMetadataProvider{store: store})()
 			require.NoError(t, err)
 			inputCh, outputCh := make(chan []Event[processAttrs], 10), make(chan []Event[processAttrs], 10)
@@ -107,7 +107,7 @@ func TestWatcherKubeEnricherWithMatcher(t *testing.T) {
 	processInfo = fakeProcessInfo
 	// Setup a fake K8s API connected to the watcherKubeEnricher
 	fInformer := &fakeInformer{}
-	store := kube.NewStore(fInformer, kube.MetadataSources{})
+	store := kube.NewStore(fInformer, kube.ResourceLabels{})
 	wkeNodeFunc, err := WatcherKubeEnricherProvider(context.TODO(), &fakeMetadataProvider{store: store})()
 	require.NoError(t, err)
 	pipeConfig := beyla.Config{}

pkg/internal/kube/informer_provider.go

@@ -34,7 +34,7 @@ type MetadataConfig struct {
 	SyncTimeout       time.Duration
 	ResyncPeriod      time.Duration
 	MetaCacheAddr     string
-	MetadataSources   MetadataSources
+	ResourceLabels    ResourceLabels
 	RestrictLocalNode bool
 }
 
@@ -107,7 +107,7 @@ func (mp *MetadataProvider) Get(ctx context.Context) (*Store, error) {
 		return nil, err
 	}
 
-	mp.metadata = NewStore(informer, mp.cfg.MetadataSources)
+	mp.metadata = NewStore(informer, mp.cfg.ResourceLabels)
 
 	return mp.metadata, nil
 }

pkg/internal/kube/store.go

@@ -36,33 +36,26 @@ func qName(om *informer.ObjectMeta) qualifiedName {
 	return qualifiedName{name: om.Name, namespace: om.Namespace, kind: om.Kind}
 }
 
-// MetadataSources allow overriding some metadata from kubernetes Pod labels and annotations
-type MetadataSources struct {
-	Annotations AnnotationSources `yaml:"annotations"`
-	Labels      LabelSources      `yaml:"labels"`
+// MetaSourceLabels allow overriding some metadata from kubernetes labels
+// Deprecated. Left here for backwards-compatibility.
+type MetaSourceLabels struct {
+	ServiceName      string `yaml:"service_name" env:"BEYLA_KUBE_META_SOURCE_LABEL_SERVICE_NAME"`
+	ServiceNamespace string `yaml:"service_namespace" env:"BEYLA_KUBE_META_SOURCE_LABEL_SERVICE_NAMESPACE"`
 }
 
-type LabelSources struct {
-	ServiceName      []string `yaml:"service_name" env:"BEYLA_KUBE_LABELS_SERVICE_NAME" envSeparator:","`
-	ServiceNamespace []string `yaml:"service_namespace" env:"BEYLA_KUBE_LABELS_SERVICE_NAMESPACE" envSeparator:","`
-}
+type ResourceLabels map[string][]string
 
-type AnnotationSources struct {
-	ServiceName      []string `yaml:"service_name" env:"BEYLA_KUBE_ANNOTATIONS_SERVICE_NAME" envSeparator:","`
-	ServiceNamespace []string `yaml:"service_namespace" env:"BEYLA_KUBE_ANNOTATIONS_SERVICE_NAMESPACE" envSeparator:","`
-}
+const (
+	ResourceAttributesPrefix   = "resource.opentelemetry.io/"
+	ServiceNameAnnotation      = ResourceAttributesPrefix + "service.name"
+	ServiceNamespaceAnnotation = ResourceAttributesPrefix + "service.namespace"
+)
 
-var DefaultMetadataSources = MetadataSources{
-	Annotations: AnnotationSources{
-		ServiceName:      []string{"resource.opentelemetry.io/service.name"},
-		ServiceNamespace: []string{"resource.opentelemetry.io/service.namespace"},
-	},
-	// If a user sets useLabelsForResourceAttributes: false it its OTEL operator, is the task of the
+var DefaultResourceLabels = ResourceLabels{
+	// If a user sets useLabelsForResourceAttributes: false in its OTEL operator config, is the task of the
 	// OTEL operator to provide empty values for this.
-	Labels: LabelSources{
-		ServiceName:      []string{"app.kubernetes.io/name"},
-		ServiceNamespace: []string{"app.kubernetes.io/part-of"},
-	},
+	"service.name":      []string{"app.kubernetes.io/name"},
+	"service.namespace": []string{"app.kubernetes.io/part-of"},
 }
 
 // Store aggregates Kubernetes information from multiple sources:
@@ -102,10 +95,10 @@ type Store struct {
 	// they receive is already present in the store
 	meta.BaseNotifier
 
-	metadataSources MetadataSources
+	resourceLabels ResourceLabels
 }
 
-func NewStore(kubeMetadata meta.Notifier, metadataSources MetadataSources) *Store {
+func NewStore(kubeMetadata meta.Notifier, resourceLabels ResourceLabels) *Store {
 	log := dblog()
 	db := &Store{
 		log:                 log,
@@ -119,7 +112,7 @@ func NewStore(kubeMetadata meta.Notifier, metadataSources MetadataSources) *Stor
 		otelServiceInfoByIP: map[string]OTelServiceNamePair{},
 		metadataNotifier:    kubeMetadata,
 		BaseNotifier:        meta.NewBaseNotifier(log),
-		metadataSources:     metadataSources,
+		resourceLabels:      resourceLabels,
 	}
 	kubeMetadata.Subscribe(db)
 	return db
@@ -318,15 +311,13 @@ func (s *Store) serviceNameNamespaceForMetadata(om *informer.ObjectMeta) (string
 // function implemented to provide consistent service metadata naming across multiple
 // OTEL implementations: OTEL operator, Loki and Beyla
 // https://github.com/grafana/k8s-monitoring-helm/issues/942
-func (s *Store) valueFromMetadata(om *informer.ObjectMeta, annotationNames, labelNames []string) string {
+func (s *Store) valueFromMetadata(om *informer.ObjectMeta, annotationName string, labelNames []string) string {
 	// if this object meta is not a pod, we ignore the metadata
 	if om.Pod == nil {
 		return ""
 	}
-	for _, key := range annotationNames {
-		if val, ok := om.Annotations[key]; ok {
-			return val
-		}
+	if val, ok := om.Annotations[annotationName]; ok {
+		return val
 	}
 	for _, key := range labelNames {
 		if val, ok := om.Labels[key]; ok {
@@ -376,16 +367,16 @@ func (s *Store) serviceNameNamespaceOwnerID(om *informer.ObjectMeta, ownerName s
 	if envName, ok := s.serviceNameFromEnv(ownerKey); ok {
 		serviceName = envName
 	} else if nameFromMeta := s.valueFromMetadata(om,
-		s.metadataSources.Annotations.ServiceName,
-		s.metadataSources.Labels.ServiceName,
+		ServiceNameAnnotation,
+		s.resourceLabels["service.name"],
 	); nameFromMeta != "" {
 		serviceName = nameFromMeta
 	}
 	if envName, ok := s.serviceNamespaceFromEnv(ownerKey); ok {
 		serviceNamespace = envName
 	} else if nsFromMeta := s.valueFromMetadata(om,
-		s.metadataSources.Annotations.ServiceNamespace,
-		s.metadataSources.Labels.ServiceNamespace,
+		ServiceNamespaceAnnotation,
+		s.resourceLabels["service.namespace"],
 	); nsFromMeta != "" {
 		serviceNamespace = nsFromMeta
 	}

pkg/internal/kube/store_test.go

@@ -90,7 +90,7 @@ func TestContainerInfo(t *testing.T) {
 
 	fInformer := &fakeInformer{}
 
-	store := NewStore(fInformer, MetadataSources{})
+	store := NewStore(fInformer, ResourceLabels{})
 
 	_ = store.On(&informer.Event{Type: informer.EventType_CREATED, Resource: &service})
 	_ = store.On(&informer.Event{Type: informer.EventType_CREATED, Resource: &podMetaA})
@@ -257,7 +257,7 @@ func TestMemoryCleanedUp(t *testing.T) {
 
 	fInformer := &fakeInformer{}
 
-	store := NewStore(fInformer, MetadataSources{})
+	store := NewStore(fInformer, ResourceLabels{})
 
 	_ = store.On(&informer.Event{Type: informer.EventType_CREATED, Resource: &service})
 	_ = store.On(&informer.Event{Type: informer.EventType_CREATED, Resource: &podMetaA})
@@ -281,7 +281,7 @@ func TestMemoryCleanedUp(t *testing.T) {
 // Fixes a memory leak in the store where the objectMetaByIP map was not cleaned up
 func TestMetaByIPEntryRemovedIfIPGroupChanges(t *testing.T) {
 	// GIVEN a store with
-	store := NewStore(&fakeInformer{}, MetadataSources{})
+	store := NewStore(&fakeInformer{}, ResourceLabels{})
 	// WHEN an object is created with several IPs
 	_ = store.On(&informer.Event{
 		Type: informer.EventType_CREATED,
@@ -326,7 +326,7 @@ func TestMetaByIPEntryRemovedIfIPGroupChanges(t *testing.T) {
 }
 
 func TestNoLeakOnUpdateOrDeletion(t *testing.T) {
-	store := NewStore(&fakeInformer{}, MetadataSources{})
+	store := NewStore(&fakeInformer{}, ResourceLabels{})
 	topOwner := &informer.Owner{Name: "foo", Kind: "Deployment"}
 	require.NoError(t, store.On(&informer.Event{
 		Type: informer.EventType_CREATED,

pkg/transform/k8s.go

@@ -53,9 +53,13 @@ type KubernetesDecorator struct {
 	// node as the Beyla instance. It will also restrict the Node information to the local node.
 	MetaRestrictLocalNode bool `yaml:"meta_restrict_local_node" env:"BEYLA_KUBE_META_RESTRICT_LOCAL_NODE"`
 
-	// MetadataSources allows Beyla overriding the service name and namespace of an application from
+	// MetaSourceLabels allows Beyla overriding the service name and namespace of an application from
 	// the given labels.
-	MetadataSources kube.MetadataSources `yaml:"meta_naming_sources"`
+	// Deprecated: kept for backwards-compatibility with Beyla 1.9
+	MetaSourceLabels kube.MetaSourceLabels `yaml:"meta_source_labels"`
+
+	// ResourceLabels allows Beyla overriding the OTEL Resource attributes from a map of user-defined labels.
+	ResourceLabels kube.ResourceLabels `yaml:"resource_labels"`
 }
 
 const (

pkg/transform/k8s_test.go

@@ -22,15 +22,9 @@ const timeout = 5 * time.Second
 
 func TestDecoration(t *testing.T) {
 	inf := &fakeInformer{}
-	store := kube.NewStore(inf, kube.MetadataSources{
-		Annotations: kube.AnnotationSources{
-			ServiceName:      []string{"resource.opentelemetry.io/service.name"},
-			ServiceNamespace: []string{"resource.opentelemetry.io/service.namespace"},
-		},
-		Labels: kube.LabelSources{
-			ServiceName:      []string{"app.kubernetes.io/name"},
-			ServiceNamespace: []string{"app.kubernetes.io/part-of"},
-		},
+	store := kube.NewStore(inf, kube.ResourceLabels{
+		"service.name":      []string{"app.kubernetes.io/name"},
+		"service.namespace": []string{"app.kubernetes.io/part-of"},
 	})
 	// pre-populated kubernetes metadata database
 	inf.Notify(&informer.Event{Type: informer.EventType_CREATED, Resource: &informer.ObjectMeta{

pkg/transform/name_resolver_test.go

@@ -32,7 +32,7 @@ func TestSuffixPrefix(t *testing.T) {
 
 func TestResolvePodsFromK8s(t *testing.T) {
 	inf := &fakeInformer{}
-	db := kube2.NewStore(inf, kube2.MetadataSources{})
+	db := kube2.NewStore(inf, kube2.ResourceLabels{})
 	pod1 := &informer.ObjectMeta{Name: "pod1", Kind: "Pod", Ips: []string{"10.0.0.1", "10.1.0.1"}}
 	pod2 := &informer.ObjectMeta{Name: "pod2", Namespace: "something", Kind: "Pod", Ips: []string{"10.0.0.2", "10.1.0.2"}}
 	pod3 := &informer.ObjectMeta{Name: "pod3", Kind: "Pod", Ips: []string{"10.0.0.3", "10.1.0.3"}}
@@ -104,7 +104,7 @@ func TestResolvePodsFromK8s(t *testing.T) {
 
 func TestResolveServiceFromK8s(t *testing.T) {
 	inf := &fakeInformer{}
-	db := kube2.NewStore(inf, kube2.MetadataSources{})
+	db := kube2.NewStore(inf, kube2.ResourceLabels{})
 	pod1 := &informer.ObjectMeta{Name: "pod1", Kind: "Service", Ips: []string{"10.0.0.1", "10.1.0.1"}}
 	pod2 := &informer.ObjectMeta{Name: "pod2", Namespace: "something", Kind: "Service", Ips: []string{"10.0.0.2", "10.1.0.2"}}
 	pod3 := &informer.ObjectMeta{Name: "pod3", Kind: "Service", Ips: []string{"10.0.0.3", "10.1.0.3"}}
@@ -196,7 +196,7 @@ func TestCleanName(t *testing.T) {
 
 func TestResolveNodesFromK8s(t *testing.T) {
 	inf := &fakeInformer{}
-	db := kube2.NewStore(inf, kube2.MetadataSources{})
+	db := kube2.NewStore(inf, kube2.ResourceLabels{})
 	node1 := &informer.ObjectMeta{Name: "node1", Kind: "Node", Ips: []string{"10.0.0.1", "10.1.0.1"}}
 	node2 := &informer.ObjectMeta{Name: "node2", Namespace: "something", Kind: "Node", Ips: []string{"10.0.0.2", "10.1.0.2"}}
 	node3 := &informer.ObjectMeta{Name: "node3", Kind: "Node", Ips: []string{"10.0.0.3", "10.1.0.3"}}