Update span enrichment to align with existing APM data logic (#237)

* Add enrichment for `service.instance.id` which follows existing APM logic for `service.node.name`

* Update the `elasticattr` package to include supported resource attributes

* Update span enrichment to preserve existing values for: span.type, span.subtype, service.target.type, service.target.name, span.destination.service.resource.

This allows any events that have been received using the `elasticapmintake receiver` to retain their original values.

* Fix faas trigger id attribute name

Author: isaacaflores2

elasticattr/attributes.go

@@ -19,8 +19,40 @@ package elasticattr
 
 const (
 	// resource s
-	AgentName    = "agent.name"
-	AgentVersion = "agent.version"
+	AgentName              = "agent.name"
+	AgentVersion           = "agent.version"
+	AgentEphemeralID       = "agent.ephemeral_id"
+	AgentActivationMethod  = "agent.activation_method"
+	CloudOriginAccountID   = "cloud.origin.account.id"
+	CloudOriginProvider    = "cloud.origin.provider"
+	CloudOriginRegion      = "cloud.origin.region"
+	CloudOriginServiceName = "cloud.origin.service.name"
+	CloudAccountName       = "cloud.account.name"
+	CloudInstanceID        = "cloud.instance.id"
+	CloudInstanceName      = "cloud.instance.name"
+	CloudMachineType       = "cloud.machine.type"
+	CloudProjectID         = "cloud.project.id"
+	CloudProjectName       = "cloud.project.name"
+	ContainerImageTag      = "container.image.tag"
+	DeviceManufacturer     = "device.manufacturer"
+	DataStreamDataset      = "data_stream.dataset"
+	DataStreamNamespace    = "data_stream.namespace"
+	DestinationIP          = "destination.ip"
+	SourceNATIP            = "source.nat.ip"
+	FaaSExecution          = "faas.execution"
+	FaaSTriggerRequestID   = "faas.trigger.request_id"
+	HostHostName           = "host.hostname"
+	HostOSPlatform         = "host.os.platform"
+	ProcessRuntimeName     = "process.runtime.name"
+	ProcessRuntimeVersion  = "process.runtime.version"
+	ServiceLanguageName    = "service.language.name"
+	ServiceLanguageVersion = "service.language.version"
+	ServiceRuntimeName     = "service.runtime.name"
+	ServiceRuntimeVersion  = "service.runtime.version"
+	ServiceOriginID        = "service.origin.id"
+	ServiceOriginName      = "service.origin.name"
+	ServiceOriginVersion   = "service.origin.version"
+	UserDomain             = "user.domain"
 
 	// scope s
 	ServiceFrameworkName    = "service.framework.name"

enrichments/config/config.go

@@ -34,6 +34,7 @@ type ResourceConfig struct {
 	AgentVersion          AttributeConfig `mapstructure:"agent_version"`
 	OverrideHostName      AttributeConfig `mapstructure:"override_host_name"`
 	DeploymentEnvironment AttributeConfig `mapstructure:"deployment_environment"`
+	ServiceInstanceID     AttributeConfig `mapstructure:"service_instance_id"`
 }
 
 // ScopeConfig configures the enrichment of scope attributes.
@@ -122,6 +123,7 @@ func Enabled() Config {
 			AgentVersion:          AttributeConfig{Enabled: true},
 			OverrideHostName:      AttributeConfig{Enabled: true},
 			DeploymentEnvironment: AttributeConfig{Enabled: true},
+			ServiceInstanceID:     AttributeConfig{Enabled: true},
 		},
 		Scope: ScopeConfig{
 			ServiceFrameworkName:    AttributeConfig{Enabled: true},

enrichments/internal/elastic/resource.go

@@ -20,11 +20,12 @@ package elastic
 import (
 	"fmt"
 
-	"github.com/elastic/opentelemetry-lib/elasticattr"
-	"github.com/elastic/opentelemetry-lib/enrichments/config"
 	"go.opentelemetry.io/collector/pdata/pcommon"
 	semconv25 "go.opentelemetry.io/otel/semconv/v1.25.0"
 	semconv "go.opentelemetry.io/otel/semconv/v1.27.0"
+
+	"github.com/elastic/opentelemetry-lib/elasticattr"
+	"github.com/elastic/opentelemetry-lib/enrichments/config"
 )
 
 // EnrichResource derives and adds Elastic specific resource attributes.
@@ -45,6 +46,9 @@ type resourceEnrichmentContext struct {
 
 	deploymentEnvironment     string
 	deploymentEnvironmentName string
+
+	serviceInstanceID string
+	containerID       string
 }
 
 func (s *resourceEnrichmentContext) Enrich(resource pcommon.Resource, cfg config.ResourceConfig) {
@@ -68,6 +72,10 @@ func (s *resourceEnrichmentContext) Enrich(resource pcommon.Resource, cfg config
 			s.deploymentEnvironment = v.Str()
 		case string(semconv.DeploymentEnvironmentNameKey):
 			s.deploymentEnvironmentName = v.Str()
+		case string(semconv25.ServiceInstanceIDKey):
+			s.serviceInstanceID = v.Str()
+		case string(semconv.ContainerIDKey):
+			s.containerID = v.Str()
 		}
 		return true
 	})
@@ -91,6 +99,10 @@ func (s *resourceEnrichmentContext) Enrich(resource pcommon.Resource, cfg config
 	if cfg.DeploymentEnvironment.Enabled {
 		s.setDeploymentEnvironment(resource)
 	}
+
+	if cfg.ServiceInstanceID.Enabled {
+		s.setServiceInstanceID(resource)
+	}
 }
 
 // SemConv v1.27.0 deprecated `deployment.environment` and added `deployment.environment.name` in favor of it.
@@ -161,3 +173,23 @@ func (s *resourceEnrichmentContext) overrideHostNameWithK8sNodeName(resource pco
 		s.k8sNodeName,
 	)
 }
+
+// setServiceInstanceID sets service.instance.id from container.id or host.name
+// if service.instance.id is not already set. This follows the existing APM logic for
+// `service.node.name`.
+func (s *resourceEnrichmentContext) setServiceInstanceID(resource pcommon.Resource) {
+	if s.serviceInstanceID != "" {
+		return
+	}
+
+	switch {
+	case s.containerID != "":
+		s.serviceInstanceID = s.containerID
+	case s.hostName != "":
+		s.serviceInstanceID = s.hostName
+	default:
+		// no instance id could be derived
+		return
+	}
+	resource.Attributes().PutStr(string(semconv25.ServiceInstanceIDKey), s.serviceInstanceID)
+}

enrichments/internal/elastic/resource_test.go

@@ -20,13 +20,14 @@ package elastic
 import (
 	"testing"
 
-	"github.com/elastic/opentelemetry-lib/elasticattr"
-	"github.com/elastic/opentelemetry-lib/enrichments/config"
 	"github.com/google/go-cmp/cmp"
 	"github.com/stretchr/testify/assert"
 	"go.opentelemetry.io/collector/pdata/pcommon"
 	semconv25 "go.opentelemetry.io/otel/semconv/v1.25.0"
 	semconv "go.opentelemetry.io/otel/semconv/v1.27.0"
+
+	"github.com/elastic/opentelemetry-lib/elasticattr"
+	"github.com/elastic/opentelemetry-lib/enrichments/config"
 )
 
 func TestResourceEnrich(t *testing.T) {
@@ -174,10 +175,11 @@ func TestResourceEnrich(t *testing.T) {
 			}(),
 			config: config.Enabled().Resource,
 			enrichedAttrs: map[string]any{
-				string(semconv.HostNameKey):    "k8s-node",
-				string(semconv.K8SNodeNameKey): "k8s-node",
-				elasticattr.AgentName:          "otlp",
-				elasticattr.AgentVersion:       "unknown",
+				string(semconv.HostNameKey):            "k8s-node",
+				string(semconv.K8SNodeNameKey):         "k8s-node",
+				elasticattr.AgentName:                  "otlp",
+				elasticattr.AgentVersion:               "unknown",
+				string(semconv25.ServiceInstanceIDKey): string("test-host"),
 			},
 		},
 		{
@@ -245,6 +247,58 @@ func TestResourceEnrich(t *testing.T) {
 				elasticattr.AgentVersion:                     "unknown",
 			},
 		},
+		{
+			name: "service_instance_id_derived_from_container_id",
+			input: func() pcommon.Resource {
+				res := pcommon.NewResource()
+				res.Attributes().PutStr(string(semconv.ServiceInstanceIDKey), "")
+				res.Attributes().PutStr(string(semconv25.ContainerIDKey), "container-id")
+				res.Attributes().PutStr(string(semconv25.HostNameKey), "k8s-node")
+				return res
+			}(),
+			config: config.Enabled().Resource,
+			enrichedAttrs: map[string]any{
+				string(semconv25.ServiceInstanceIDKey): "container-id",
+				string(semconv.ContainerIDKey):         "container-id",
+				string(semconv.HostNameKey):            "k8s-node",
+				elasticattr.AgentName:                  "otlp",
+				elasticattr.AgentVersion:               "unknown",
+			},
+		},
+		{
+			name: "service_instance_id_derived_from_host_name",
+			input: func() pcommon.Resource {
+				res := pcommon.NewResource()
+				res.Attributes().PutStr(string(semconv.ServiceInstanceIDKey), "")
+				res.Attributes().PutStr(string(semconv25.HostNameKey), "k8s-node")
+				return res
+			}(),
+			config: config.Enabled().Resource,
+			enrichedAttrs: map[string]any{
+				string(semconv25.ServiceInstanceIDKey): "k8s-node",
+				string(semconv.HostNameKey):            "k8s-node",
+				elasticattr.AgentName:                  "otlp",
+				elasticattr.AgentVersion:               "unknown",
+			},
+		},
+		{
+			name: "service_instance_id_already_set",
+			input: func() pcommon.Resource {
+				res := pcommon.NewResource()
+				res.Attributes().PutStr(string(semconv.ServiceInstanceIDKey), "node-name")
+				res.Attributes().PutStr(string(semconv25.ContainerIDKey), "container-id")
+				res.Attributes().PutStr(string(semconv25.HostNameKey), "k8s-node")
+				return res
+			}(),
+			config: config.Enabled().Resource,
+			enrichedAttrs: map[string]any{
+				string(semconv25.ServiceInstanceIDKey): "node-name",
+				string(semconv.ContainerIDKey):         "container-id",
+				string(semconv.HostNameKey):            "k8s-node",
+				elasticattr.AgentName:                  "otlp",
+				elasticattr.AgentVersion:               "unknown",
+			},
+		},
 	} {
 		t.Run(tc.name, func(t *testing.T) {
 			// Merge existing resource attrs with the attrs added

enrichments/internal/elastic/span.go

@@ -29,8 +29,6 @@ import (
 	"net/url"
 	"strconv"
 
-	"github.com/elastic/opentelemetry-lib/elasticattr"
-	"github.com/elastic/opentelemetry-lib/enrichments/config"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/pkg/sampling"
 	"github.com/ua-parser/uap-go/uaparser"
 	"go.opentelemetry.io/collector/pdata/pcommon"
@@ -40,6 +38,9 @@ import (
 	semconv37 "go.opentelemetry.io/otel/semconv/v1.37.0"
 	tracepb "go.opentelemetry.io/proto/otlp/trace/v1"
 	"google.golang.org/grpc/codes"
+
+	"github.com/elastic/opentelemetry-lib/elasticattr"
+	"github.com/elastic/opentelemetry-lib/enrichments/config"
 )
 
 // defaultRepresentativeCount is the representative count to use for adjusting
@@ -444,9 +445,14 @@ func (s *spanEnrichmentContext) setSpanTypeSubtype(span ptrace.Span) (spanType s
 		}
 	}
 
-	span.Attributes().PutStr(elasticattr.SpanType, spanType)
+	// do not overwrite existing span.type and span.subtype attributes
+	if existingSpanType, _ := span.Attributes().Get(elasticattr.SpanType); existingSpanType.Str() == "" {
+		span.Attributes().PutStr(elasticattr.SpanType, spanType)
+	}
 	if spanSubtype != "" {
-		span.Attributes().PutStr(elasticattr.SpanSubtype, spanSubtype)
+		if existingSpanSubtype, _ := span.Attributes().Get(elasticattr.SpanSubtype); existingSpanSubtype.Str() == "" {
+			span.Attributes().PutStr(elasticattr.SpanSubtype, spanSubtype)
+		}
 	}
 
 	return spanType, spanSubtype
@@ -494,9 +500,15 @@ func (s *spanEnrichmentContext) setServiceTarget(span ptrace.Span) {
 		}
 	}
 
+	// set either target.type or target.name if at least one is available
 	if targetType != "" || targetName != "" {
-		span.Attributes().PutStr(elasticattr.ServiceTargetType, targetType)
-		span.Attributes().PutStr(elasticattr.ServiceTargetName, targetName)
+		// do not overwrite existing target.type and target.name attributes
+		if existingTargetType, _ := span.Attributes().Get(elasticattr.ServiceTargetType); existingTargetType.Str() == "" {
+			span.Attributes().PutStr(elasticattr.ServiceTargetType, targetType)
+		}
+		if existingTargetName, _ := span.Attributes().Get(elasticattr.ServiceTargetName); existingTargetName.Str() == "" {
+			span.Attributes().PutStr(elasticattr.ServiceTargetName, targetName)
+		}
 	}
 }
 
@@ -536,7 +548,10 @@ func (s *spanEnrichmentContext) setDestinationService(span ptrace.Span) {
 	}
 
 	if destnResource != "" {
-		span.Attributes().PutStr(elasticattr.SpanDestinationServiceResource, destnResource)
+		// do not overwrite existing span.destination.service.resource attribute
+		if existingDestnResource, _ := span.Attributes().Get(elasticattr.SpanDestinationServiceResource); existingDestnResource.Str() == "" {
+			span.Attributes().PutStr(elasticattr.SpanDestinationServiceResource, destnResource)
+		}
 	}
 }