Add logging labels with GCE instance's managed instance group. (#1849)

Co-authored-by: Francisco Valente <1435136+franciscovalentecastro@users.noreply.github.com>

confgenerator/confgenerator.go

@@ -185,7 +185,7 @@ func (p pipelineInstance) fluentBitComponents(ctx context.Context) (fbSource, er
 		}
 		components = append(components, processorComponents...)
 	}
-	components = append(components, setLogNameComponents(ctx, tag, p.rID, receiver.Type(), platform.FromContext(ctx).Hostname())...)
+	components = append(components, setLogNameComponents(ctx, tag, p.rID, receiver.Type())...)
 
 	// Logs ingested using the fluent_forward receiver must add the existing_tag
 	// on the record to the LogName. This is done with a Lua filter.
@@ -225,7 +225,7 @@ func (p pipelineInstance) otelComponents(ctx context.Context) (map[string]otel.R
 		if processors, ok := receiverPipeline.Processors["logs"]; ok {
 			receiverPipeline.Processors["logs"] = append(
 				processors,
-				otelSetLogNameComponents(ctx, p.rID, platform.FromContext(ctx).Hostname())...,
+				otelSetLogNameComponents(ctx, p.rID)...,
 			)
 		}
 

confgenerator/confgenerator_test.go

@@ -24,6 +24,7 @@ import (
 	"strings"
 	"testing"
 
+	"github.com/GoogleCloudPlatform/opentelemetry-operations-go/detectors/gcp"
 	"github.com/GoogleCloudPlatform/ops-agent/apps"
 	"github.com/GoogleCloudPlatform/ops-agent/confgenerator"
 	"github.com/GoogleCloudPlatform/ops-agent/confgenerator/resourcedetector"
@@ -72,6 +73,11 @@ var (
 		Metadata:      map[string]string{"test-key": "test-value", "test-escape": "$foo", "test-escape-parentheses": "${foo:bar}"},
 		Label:         map[string]string{"test-label-key": "test-label-value"},
 		InterfaceIPv4: map[string]string{"test-interface": "test-interface-ipv4"},
+		ManagedInstanceGroup: gcp.ManagedInstanceGroup{
+			Name:     "test-mig",
+			Type:     gcp.Zone,
+			Location: "test-zone",
+		},
 	}
 	linuxTestPlatform = platformConfig{
 		name:            "linux",

confgenerator/logging.go

@@ -28,38 +28,48 @@ import (
 const InstrumentationSourceLabel = `labels."logging.googleapis.com/instrumentation_source"`
 const HttpRequestKey = "logging.googleapis.com/httpRequest"
 
-// setLogNameComponents generates a series of components that rewrites the tag on log entries tagged `tag` to be `logName`.
-func setLogNameComponents(ctx context.Context, tag, logName, receiverType string, hostName string) []fluentbit.Component {
-	return LoggingProcessorModifyFields{
-		Fields: map[string]*ModifyField{
-			"logName": {
-				DefaultValue: &logName,
-			},
-			`labels."compute.googleapis.com/resource_name"`: {
-				DefaultValue: &hostName,
-			},
-			`labels."agent.googleapis.com/log_file_path"`: {
-				MoveFrom: `jsonPayload."agent.googleapis.com/log_file_path"`,
-			},
-			// `labels."agent.googleapis.com/receiver_type"`: {
-			// 	StaticValue: &receiverType,
-			// },
+func setLogNameProcessor(ctx context.Context, logName string) LoggingProcessorModifyFields {
+	p := platform.FromContext(ctx)
+	hostName := p.Hostname()
+	fields := map[string]*ModifyField{
+		"logName": {
+			DefaultValue: &logName,
 		},
-	}.Components(ctx, tag, "setlogname")
+		`labels."compute.googleapis.com/resource_name"`: {
+			DefaultValue: &hostName,
+		},
+	}
+	r, err := p.GetResource()
+	if err == nil {
+		for k, v := range r.ExtraLogLabels() {
+			v := v
+			fields[fmt.Sprintf("labels.%q", k)] = &ModifyField{
+				DefaultValue: &v,
+			}
+		}
+	}
+	return LoggingProcessorModifyFields{
+		Fields: fields,
+	}
 }
 
-func otelSetLogNameComponents(ctx context.Context, logName, hostName string) []otel.Component {
-	components, err := LoggingProcessorModifyFields{
-		Fields: map[string]*ModifyField{
-			// TODO: Prepend `receiver_id.` if it already exists, like the `fluent_forward` receiver?
-			"logName": {
-				DefaultValue: &logName,
-			},
-			`labels."compute.googleapis.com/resource_name"`: {
-				DefaultValue: &hostName,
-			},
-		},
-	}.Processors(ctx)
+// setLogNameComponents generates a series of components that rewrites the tag on log entries tagged `tag` to be `logName`.
+func setLogNameComponents(ctx context.Context, tag, logName, receiverType string) []fluentbit.Component {
+	p := setLogNameProcessor(ctx, logName)
+	p.Fields[`labels."agent.googleapis.com/log_file_path"`] = &ModifyField{
+		MoveFrom: `jsonPayload."agent.googleapis.com/log_file_path"`,
+	}
+	// TODO: b/397957994 - Cleanup "receiver_type" logic if no future intended use.
+	// p.Fields[`labels."agent.googleapis.com/receiver_type"`] = &ModifyField{
+	// 	StaticValue: &receiverType,
+	// }
+	return p.Components(ctx, tag, "setlogname")
+}
+
+func otelSetLogNameComponents(ctx context.Context, logName string) []otel.Component {
+	// TODO: Prepend `receiver_id.` if it already exists, like the `fluent_forward` receiver?
+	p := setLogNameProcessor(ctx, logName)
+	components, err := p.Processors(ctx)
 	if err != nil {
 		// We're generating a hard-coded config, so this should never fail.
 		panic(err)

confgenerator/resourcedetector/detector.go

@@ -26,6 +26,7 @@ type Resource interface {
 	MonitoredResource() *monitoredres.MonitoredResource
 	OTelResourceAttributes() map[string]string
 	PrometheusStyleMetadata() map[string]string
+	ExtraLogLabels() map[string]string
 }
 
 type resourceCache struct {
@@ -81,6 +82,10 @@ func (UnrecognizedPlatformResource) PrometheusStyleMetadata() map[string]string
 	return nil
 }
 
+func (UnrecognizedPlatformResource) ExtraLogLabels() map[string]string {
+	return nil
+}
+
 func GetUnrecognizedPlatformResource() (Resource, error) {
 	return UnrecognizedPlatformResource{}, nil
 }

confgenerator/resourcedetector/gce_detector.go

@@ -18,6 +18,7 @@ import (
 	"fmt"
 	"strings"
 
+	"github.com/GoogleCloudPlatform/opentelemetry-operations-go/detectors/gcp"
 	"github.com/prometheus/prometheus/util/strutil"
 	"google.golang.org/genproto/googleapis/api/monitoredres"
 )
@@ -41,6 +42,12 @@ const (
 	defaultScopes
 )
 
+const (
+	ManagedInstanceGroupNameLabel   = `compute.googleapis.com/instance_group_manager/name`
+	ManagedInstanceGroupZoneLabel   = `compute.googleapis.com/instance_group_manager/zone`
+	ManagedInstanceGroupRegionLabel = `compute.googleapis.com/instance_group_manager/region`
+)
+
 func GetGCEResource() (Resource, error) {
 	provider := NewGCEMetadataProvider()
 	dt := GCEResourceBuilder{provider: provider}
@@ -65,6 +72,7 @@ type gceDataProvider interface {
 	getLabels() (map[string]string, error)
 	getMetadata() (map[string]string, error)
 	getInterfaceIPv4s() (map[string]string, error)
+	getMIG() (gcp.ManagedInstanceGroup, error)
 }
 
 // List of single-valued attributes (non-nested)
@@ -95,20 +103,21 @@ var nestedAttributeSpec = map[gceAttribute]func(gceDataProvider) (map[string]str
 
 // GCEResource implements the Resource interface and provide attributes of the VM when on GCE
 type GCEResource struct {
-	Project       string
-	Zone          string
-	Network       string
-	Subnetwork    string
-	PublicIP      string
-	PrivateIP     string
-	InstanceID    string
-	InstanceName  string
-	Tags          string
-	MachineType   string
-	DefaultScopes []string
-	Metadata      map[string]string
-	Label         map[string]string
-	InterfaceIPv4 map[string]string
+	Project              string
+	Zone                 string
+	Network              string
+	Subnetwork           string
+	PublicIP             string
+	PrivateIP            string
+	InstanceID           string
+	InstanceName         string
+	Tags                 string
+	MachineType          string
+	DefaultScopes        []string
+	Metadata             map[string]string
+	Label                map[string]string
+	InterfaceIPv4        map[string]string
+	ManagedInstanceGroup gcp.ManagedInstanceGroup
 }
 
 func (r GCEResource) ProjectName() string {
@@ -184,6 +193,20 @@ func (r GCEResource) PrometheusStyleMetadata() map[string]string {
 	return metaLabels
 }
 
+func (r GCEResource) ExtraLogLabels() map[string]string {
+	l := make(map[string]string)
+	if r.ManagedInstanceGroup.Name != "" {
+		l[ManagedInstanceGroupNameLabel] = r.ManagedInstanceGroup.Name
+	}
+	switch r.ManagedInstanceGroup.Type {
+	case gcp.Zone:
+		l[ManagedInstanceGroupZoneLabel] = r.ManagedInstanceGroup.Location
+	case gcp.Region:
+		l[ManagedInstanceGroupRegionLabel] = r.ManagedInstanceGroup.Location
+	}
+	return l
+}
+
 type GCEResourceBuilderInterface interface {
 	GetResource() (Resource, error)
 }
@@ -220,21 +243,27 @@ func (gd *GCEResourceBuilder) GetResource() (Resource, error) {
 		nestedAttributes[attrName] = attr
 	}
 
+	mig, err := gd.provider.getMIG()
+	if err != nil {
+		return nil, err
+	}
+
 	res := GCEResource{
-		Project:       singleAttributes[project],
-		Zone:          singleAttributes[zone],
-		Network:       singleAttributes[network],
-		Subnetwork:    singleAttributes[subnetwork],
-		PublicIP:      singleAttributes[publicIP],
-		PrivateIP:     singleAttributes[privateIP],
-		InstanceID:    singleAttributes[instanceID],
-		InstanceName:  singleAttributes[instanceName],
-		Tags:          singleAttributes[tags],
-		MachineType:   singleAttributes[machineType],
-		DefaultScopes: multiAttributes[defaultScopes],
-		Metadata:      nestedAttributes[metadata],
-		Label:         nestedAttributes[label],
-		InterfaceIPv4: nestedAttributes[interfaceIPv4],
+		Project:              singleAttributes[project],
+		Zone:                 singleAttributes[zone],
+		Network:              singleAttributes[network],
+		Subnetwork:           singleAttributes[subnetwork],
+		PublicIP:             singleAttributes[publicIP],
+		PrivateIP:            singleAttributes[privateIP],
+		InstanceID:           singleAttributes[instanceID],
+		InstanceName:         singleAttributes[instanceName],
+		Tags:                 singleAttributes[tags],
+		MachineType:          singleAttributes[machineType],
+		DefaultScopes:        multiAttributes[defaultScopes],
+		Metadata:             nestedAttributes[metadata],
+		Label:                nestedAttributes[label],
+		InterfaceIPv4:        nestedAttributes[interfaceIPv4],
+		ManagedInstanceGroup: mig,
 	}
 	return res, nil
 }

confgenerator/resourcedetector/gce_detector_test.go

@@ -17,6 +17,8 @@ package resourcedetector
 import (
 	"fmt"
 	"testing"
+
+	"github.com/GoogleCloudPlatform/opentelemetry-operations-go/detectors/gcp"
 )
 
 func TestGettingResourceWithoutError(t *testing.T) {
@@ -140,3 +142,7 @@ func (fp *FakeProvider) getLabels() (map[string]string, error) {
 func (fp *FakeProvider) getInterfaceIPv4s() (map[string]string, error) {
 	return fp.getMap()
 }
+
+func (fp *FakeProvider) getMIG() (gcp.ManagedInstanceGroup, error) {
+	return gcp.ManagedInstanceGroup{}, nil
+}

confgenerator/resourcedetector/gce_metadata_provider.go

@@ -19,6 +19,7 @@ import (
 	"strings"
 
 	gcp_metadata "cloud.google.com/go/compute/metadata"
+	"github.com/GoogleCloudPlatform/opentelemetry-operations-go/detectors/gcp"
 )
 
 const notAvailable = "NOT AVAILABLE"
@@ -132,3 +133,7 @@ func (gmp *GCEMetadataProvider) getInterfaceIPv4s() (map[string]string, error) {
 	}
 	return interfaces, err
 }
+
+func (gmp *GCEMetadataProvider) getMIG() (gcp.ManagedInstanceGroup, error) {
+	return gcp.NewDetector().GCEManagedInstanceGroup()
+}

confgenerator/testdata/goldens/all-backward_compatible_with_explicit_exporters/golden/linux-gpu/adea349dc2d92cd07daa1d7847f5e96a.lua

@@ -0,0 +1,74 @@
+
+function process(tag, timestamp, record)
+local __field_0 = (function()
+return record["agent.googleapis.com/log_file_path"]
+end)();
+local __field_1 = (function()
+if record["logging.googleapis.com/labels"] == nil
+then
+return nil
+end
+return record["logging.googleapis.com/labels"]["compute.googleapis.com/instance_group_manager/name"]
+end)();
+local __field_2 = (function()
+if record["logging.googleapis.com/labels"] == nil
+then
+return nil
+end
+return record["logging.googleapis.com/labels"]["compute.googleapis.com/instance_group_manager/zone"]
+end)();
+local __field_3 = (function()
+if record["logging.googleapis.com/labels"] == nil
+then
+return nil
+end
+return record["logging.googleapis.com/labels"]["compute.googleapis.com/resource_name"]
+end)();
+local __field_4 = (function()
+return record["logging.googleapis.com/logName"]
+end)();
+(function(value)
+record["agent.googleapis.com/log_file_path"] = value
+end)(nil);
+local v = __field_0;
+(function(value)
+if record["logging.googleapis.com/labels"] == nil
+then
+record["logging.googleapis.com/labels"] = {}
+end
+record["logging.googleapis.com/labels"]["agent.googleapis.com/log_file_path"] = value
+end)(v)
+local v = __field_1;
+if v == nil then v = "test-mig" end;
+(function(value)
+if record["logging.googleapis.com/labels"] == nil
+then
+record["logging.googleapis.com/labels"] = {}
+end
+record["logging.googleapis.com/labels"]["compute.googleapis.com/instance_group_manager/name"] = value
+end)(v)
+local v = __field_2;
+if v == nil then v = "test-zone" end;
+(function(value)
+if record["logging.googleapis.com/labels"] == nil
+then
+record["logging.googleapis.com/labels"] = {}
+end
+record["logging.googleapis.com/labels"]["compute.googleapis.com/instance_group_manager/zone"] = value
+end)(v)
+local v = __field_3;
+if v == nil then v = "" end;
+(function(value)
+if record["logging.googleapis.com/labels"] == nil
+then
+record["logging.googleapis.com/labels"] = {}
+end
+record["logging.googleapis.com/labels"]["compute.googleapis.com/resource_name"] = value
+end)(v)
+local v = __field_4;
+if v == nil then v = "syslog" end;
+(function(value)
+record["logging.googleapis.com/logName"] = value
+end)(v)
+return 2, timestamp, record
+end