Extract ReplicaSet name from pod name (#1202)

* Extract ReplicaSet name from pod name * Fix k8s test * Fix disabledinformers test
grafana · Sep 26, 2024 · 9d41b5c · 9d41b5c
1 parent 6960d77
commit 9d41b5c
Show file tree

Hide file tree

Showing 33 changed files with 116 additions and 85 deletions.
diff --git a/pkg/export/attributes/attr_defs.go b/pkg/export/attributes/attr_defs.go
@@ -107,6 +107,7 @@ func getDefinitions(groups AttrGroups) map[Section]AttrReportGroup {
 			attr.K8sPodUID:          true,
 			attr.K8sPodStartTime:    true,
 			attr.K8sClusterName:     true,
+			attr.K8sOwnerName:       true,
 		},
 	}
 

diff --git a/pkg/export/attributes/names/attrs.go b/pkg/export/attributes/names/attrs.go
@@ -54,16 +54,16 @@ const (
 	MessagingSystem        = Name(semconv.MessagingSystemKey)
 	MessagingDestination   = Name(semconv.MessagingDestinationNameKey)
 
-	K8sNamespaceName    = Name("k8s.namespace.name")
-	K8sPodName          = Name("k8s.pod.name")
-	K8sDeploymentName   = Name("k8s.deployment.name")
-	K8sReplicaSetName   = Name("k8s.replicaset.name")
-	K8sDaemonSetName    = Name("k8s.daemonset.name")
-	K8sStatefulSetName  = Name("k8s.statefulset.name")
-	K8sUnknownOwnerName = Name("k8s.owner.name")
-	K8sNodeName         = Name("k8s.node.name")
-	K8sPodUID           = Name("k8s.pod.uid")
-	K8sPodStartTime     = Name("k8s.pod.start_time")
+	K8sNamespaceName   = Name("k8s.namespace.name")
+	K8sPodName         = Name("k8s.pod.name")
+	K8sDeploymentName  = Name("k8s.deployment.name")
+	K8sReplicaSetName  = Name("k8s.replicaset.name")
+	K8sDaemonSetName   = Name("k8s.daemonset.name")
+	K8sStatefulSetName = Name("k8s.statefulset.name")
+	K8sOwnerName       = Name("k8s.owner.name")
+	K8sNodeName        = Name("k8s.node.name")
+	K8sPodUID          = Name("k8s.pod.uid")
+	K8sPodStartTime    = Name("k8s.pod.start_time")
 )
 
 // Beyla-specific network attributes

diff --git a/pkg/internal/discover/watcher_kube.go b/pkg/internal/discover/watcher_kube.go
@@ -9,6 +9,7 @@ import (
 	"github.com/mariomac/pipes/pipe"
 	"k8s.io/client-go/tools/cache"
 
+	attr "github.com/grafana/beyla/pkg/export/attributes/names"
 	"github.com/grafana/beyla/pkg/internal/helpers/container"
 	"github.com/grafana/beyla/pkg/internal/helpers/maps"
 	"github.com/grafana/beyla/pkg/internal/imetrics"
@@ -338,20 +339,12 @@ func withMetadata(pp processAttrs, info *kube.PodInfo) processAttrs {
 		services.AttrPodName:   info.Name,
 	}
 	ret.podLabels = info.Labels
-	owner := info.Owner
-	for owner != nil {
-		ret.metadata[services.AttrOwnerName] = owner.Name
-		switch owner.LabelName {
-		case kube.OwnerDaemonSet:
-			ret.metadata[services.AttrDaemonSetName] = owner.Name
-		case kube.OwnerReplicaSet:
-			ret.metadata[services.AttrReplicaSetName] = owner.Name
-		case kube.OwnerDeployment:
-			ret.metadata[services.AttrDeploymentName] = owner.Name
-		case kube.OwnerStatefulSet:
-			ret.metadata[services.AttrStatefulSetName] = owner.Name
-		}
-		owner = owner.Owner
+
+	if info.Owner != nil {
+		ret.metadata[attr.Name(info.Owner.LabelName).Prom()] = info.Owner.Name
+		topName, topLabel := info.Owner.TopOwnerNameLabel()
+		ret.metadata[attr.Name(topLabel).Prom()] = topName
+		ret.metadata[services.AttrOwnerName] = topName
 	}
 	return ret
 }
diff --git a/pkg/internal/discover/watcher_kube_test.go b/pkg/internal/discover/watcher_kube_test.go
@@ -30,9 +30,9 @@ const (
 	containerPID   = 123
 	containerID    = "container-123"
 	containerPort  = 332
-	replicaSetName = "rs-3321"
+	replicaSetName = "the-deployment-123456789"
 	deploymentName = "the-deployment"
-	podName        = "the-pod"
+	podName        = "the-deployment-123456789-abcde"
 )
 
 func TestWatcherKubeEnricher(t *testing.T) {
@@ -155,8 +155,8 @@ func TestWatcherKubeEnricherWithMatcher(t *testing.T) {
 		newProcess(inputCh, 123, []uint32{777})
 		newProcess(inputCh, 456, []uint32{})
 		newProcess(inputCh, 789, []uint32{443})
-		deployOwnedPod(t, k8sClient, namespace, "pod-789", "rs-789", "container-789")
-		deployReplicaSet(t, k8sClient, namespace, "rs-789", "ouyeah")
+		deployOwnedPod(t, k8sClient, namespace, "depl-rsid-podid", "depl-rsid", "container-789")
+		deployReplicaSet(t, k8sClient, namespace, "depl-rsid", "depl")
 	})
 
 	// sending events that will match and will be forwarded
@@ -205,8 +205,8 @@ func TestWatcherKubeEnricherWithMatcher(t *testing.T) {
 
 	t.Run("both process and metadata match", func(t *testing.T) {
 		newProcess(inputCh, 56, []uint32{443})
-		deployOwnedPod(t, k8sClient, namespace, "pod-56", "rs-56", "container-56")
-		deployReplicaSet(t, k8sClient, namespace, "rs-56", "chacha")
+		deployOwnedPod(t, k8sClient, namespace, "chacha-rsid-podid", "chacha-rsid", "container-56")
+		deployReplicaSet(t, k8sClient, namespace, "chacha-rsid", "chacha")
 		matches := testutil.ReadChannel(t, outputCh, timeout)
 		require.Len(t, matches, 1)
 		m := matches[0]

diff --git a/pkg/internal/ebpf/httpssl/bpf_x86_bpfel.o b/pkg/internal/ebpf/httpssl/bpf_x86_bpfel.o
diff --git a/pkg/internal/kube/informer.go b/pkg/internal/kube/informer.go
@@ -430,14 +430,8 @@ func (k *Metadata) AddNodeEventHandler(h cache.ResourceEventHandler) error {
 }
 
 func (i *PodInfo) ServiceName() string {
-	if i.Owner != nil {
-		// we have two levels of ownership at most
-		if i.Owner.Owner != nil {
-			return i.Owner.Owner.Name
-		}
-
-		return i.Owner.Name
+	if on, _ := i.Owner.TopOwnerNameLabel(); on != "" {
+		return on
 	}
-
 	return i.Name
 }
diff --git a/pkg/internal/kube/informer_test.go b/pkg/internal/kube/informer_test.go
@@ -16,26 +16,14 @@ func TestServiceName(t *testing.T) {
 
 	pod2 := PodInfo{
 		Owner: &Owner{
+			LabelName: OwnerReplicaSet,
 			Owner: &Owner{
 				Name: "nested_two",
 			},
 		},
 	}
 
 	pod3 := PodInfo{
-		Owner: &Owner{
-			Owner: &Owner{
-				Owner: &Owner{
-					Name: "nested_three",
-				},
-			},
-		},
-		ObjectMeta: metav1.ObjectMeta{
-			Name: "too_nested",
-		},
-	}
-
-	pod4 := PodInfo{
 		ObjectMeta: metav1.ObjectMeta{
 			Name: "not_nested",
 		},
@@ -45,7 +33,6 @@ func TestServiceName(t *testing.T) {
 
 	assert.Equal(t, "nested_one", pod.ServiceName())
 	assert.Equal(t, "nested_two", pod2.ServiceName())
-	assert.Equal(t, "", pod3.ServiceName())
-	assert.Equal(t, "not_nested", pod4.ServiceName())
+	assert.Equal(t, "not_nested", pod3.ServiceName())
 	assert.Equal(t, "", pod5.ServiceName())
 }
diff --git a/pkg/internal/kube/owner.go b/pkg/internal/kube/owner.go
@@ -15,7 +15,7 @@ const (
 	OwnerDeployment  = OwnerLabel(attr.K8sDeploymentName)
 	OwnerStatefulSet = OwnerLabel(attr.K8sStatefulSetName)
 	OwnerDaemonSet   = OwnerLabel(attr.K8sDaemonSetName)
-	OwnerUnknown     = OwnerLabel(attr.K8sUnknownOwnerName)
+	OwnerGeneric     = OwnerLabel(attr.K8sOwnerName)
 )
 
 type Owner struct {
@@ -57,11 +57,37 @@ func OwnerFrom(orefs []metav1.OwnerReference) *Owner {
 
 func unrecognizedOwner(or *metav1.OwnerReference) *Owner {
 	return &Owner{
-		LabelName: OwnerLabel(attr.K8sUnknownOwnerName),
+		LabelName: OwnerLabel(attr.K8sOwnerName),
 		Name:      or.Name,
 	}
 }
 
+// TopOwnerNameLabel returns the top-level name and metadata label in the owner chain.
+// For example, if the owner is a ReplicaSet, it will return the Deployment name.
+func (o *Owner) TopOwnerNameLabel() (string, OwnerLabel) {
+	if o == nil {
+		return "", ""
+	}
+	if o.LabelName == OwnerReplicaSet {
+		// we have two levels of ownership at most
+		if o.Owner != nil {
+			return o.Owner.Name, o.Owner.LabelName
+		}
+		// if the replicaset informer is disabled, we can't get the owner deployment,
+		// so we will heuristically extract it from the ReplicaSet Name (and cache it)
+		topOwnerName := o.Name
+		if idx := strings.LastIndexByte(topOwnerName, '-'); idx > 0 {
+			topOwnerName = topOwnerName[:idx]
+			o.Owner = &Owner{
+				Name:      topOwnerName,
+				LabelName: OwnerDeployment,
+			}
+			return topOwnerName, OwnerDeployment
+		}
+	}
+	return o.Name, o.LabelName
+}
+
 func (o *Owner) String() string {
 	sb := strings.Builder{}
 	o.string(&sb)

diff --git a/pkg/internal/kube/owner_test.go b/pkg/internal/kube/owner_test.go
@@ -41,7 +41,7 @@ func TestOwnerFrom_Unrecognized(t *testing.T) {
 	})
 	require.NotNil(t, owner)
 	assert.Equal(t, &Owner{
-		LabelName: OwnerUnknown,
+		LabelName: OwnerGeneric,
 		Name:      "theowner",
 	}, owner)
 }
@@ -52,7 +52,36 @@ func TestOwnerFrom_Unrecognized_AppsV1(t *testing.T) {
 	})
 	require.NotNil(t, owner)
 	assert.Equal(t, &Owner{
-		LabelName: OwnerUnknown,
+		LabelName: OwnerGeneric,
 		Name:      "theowner",
 	}, owner)
 }
+
+func TestTopOwnerLabel(t *testing.T) {
+	type testCase struct {
+		expectedLabel OwnerLabel
+		expectedName  string
+		owner         *Owner
+	}
+	for _, tc := range []testCase{
+		{owner: nil},
+		{expectedLabel: OwnerDaemonSet, expectedName: "ds",
+			owner: &Owner{LabelName: OwnerDaemonSet, Name: "ds"}},
+		{expectedLabel: OwnerDeployment, expectedName: "rs-without-dep-meta",
+			owner: &Owner{LabelName: OwnerReplicaSet, Name: "rs-without-dep-meta-34fb1fa3a"}},
+		{expectedLabel: OwnerDeployment, expectedName: "dep",
+			owner: &Owner{LabelName: OwnerReplicaSet, Name: "dep-34fb1fa3a",
+				Owner: &Owner{LabelName: OwnerDeployment, Name: "dep"}}},
+	} {
+		t.Run(tc.expectedName, func(t *testing.T) {
+			name, label := tc.owner.TopOwnerNameLabel()
+			assert.Equal(t, tc.expectedName, name)
+			assert.Equal(t, tc.expectedLabel, label)
+
+			// check that the output is consistent (e.g. after ReplicaSet owner data is cached)
+			name, label = tc.owner.TopOwnerNameLabel()
+			assert.Equal(t, tc.expectedName, name)
+			assert.Equal(t, tc.expectedLabel, label)
+		})
+	}
+}
diff --git a/pkg/internal/netolly/ebpf/net_arm64_bpfel.o b/pkg/internal/netolly/ebpf/net_arm64_bpfel.o
diff --git a/pkg/internal/netolly/ebpf/netsk_x86_bpfel.o b/pkg/internal/netolly/ebpf/netsk_x86_bpfel.o
diff --git a/pkg/transform/k8s.go b/pkg/transform/k8s.go
@@ -128,10 +128,10 @@ func (md *metadataDecorator) appendMetadata(span *request.Span, info *kube.PodIn
 		attr.K8sPodStartTime:  info.StartTimeStr,
 		attr.K8sClusterName:   md.clusterName,
 	}
-	owner := info.Owner
-	for owner != nil {
-		span.ServiceID.Metadata[attr.Name(owner.LabelName)] = owner.Name
-		owner = owner.Owner
+	if info.Owner != nil {
+		span.ServiceID.Metadata[attr.Name(info.Owner.LabelName)] = info.Owner.Name
+		topName, topLabel := info.Owner.TopOwnerNameLabel()
+		span.ServiceID.Metadata[attr.Name(topLabel)] = topName
 	}
 	// override hostname by the Pod name
 	span.ServiceID.HostName = info.Name

diff --git a/pkg/transform/k8s_test.go b/pkg/transform/k8s_test.go
@@ -76,11 +76,12 @@ func TestDecoration(t *testing.T) {
 		deco := testutil.ReadChannel(t, outputhCh, timeout)
 		require.Len(t, deco, 1)
 		assert.Equal(t, "the-ns", deco[0].ServiceID.Namespace)
-		assert.Equal(t, "rs-34", deco[0].ServiceID.Name)
+		assert.Equal(t, "rs", deco[0].ServiceID.Name)
 		assert.Equal(t, map[attr.Name]string{
 			"k8s.node.name":       "the-node",
 			"k8s.namespace.name":  "the-ns",
 			"k8s.replicaset.name": "rs-34",
+			"k8s.deployment.name": "rs",
 			"k8s.pod.name":        "pod-34",
 			"k8s.pod.uid":         "uid-34",
 			"k8s.pod.start_time":  "2020-01-02 12:34:56",

diff --git a/test/integration/components/beyla/Dockerfile b/test/integration/components/beyla/Dockerfile
@@ -1,7 +1,7 @@
 # Development version of the beyla Dockerfile that compiles for coverage
 # and allows retrieving coverage files.
 # The production-ready minimal image is in the project root's dockerfile.
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/beyla/Dockerfile_dbg b/test/integration/components/beyla/Dockerfile_dbg
@@ -1,7 +1,7 @@
 # Development version of the beyla Dockerfile that compiles for coverage
 # and allows retrieving coverage files.
 # The production-ready minimal image is in the project root's dockerfile.
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/dotnetserver/Dockerfile b/test/integration/components/dotnetserver/Dockerfile
@@ -1,4 +1,4 @@
-FROM mcr.microsoft.com/dotnet/sdk:7.0 as builder
+FROM mcr.microsoft.com/dotnet/sdk:7.0 AS builder
 
 # Set the working directory to /build
 WORKDIR /build

diff --git a/test/integration/components/gohttp2/client/Dockerfile b/test/integration/components/gohttp2/client/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/gohttp2/server/Dockerfile b/test/integration/components/gohttp2/server/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/gokafka/Dockerfile b/test/integration/components/gokafka/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/goredis/Dockerfile b/test/integration/components/goredis/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/grpcpinger/Dockerfile b/test/integration/components/grpcpinger/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/httppinger/Dockerfile b/test/integration/components/httppinger/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/old_grpc/backend/Dockerfile b/test/integration/components/old_grpc/backend/Dockerfile
@@ -1,4 +1,4 @@
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 WORKDIR /src
 

diff --git a/test/integration/components/old_grpc/worker/Dockerfile b/test/integration/components/old_grpc/worker/Dockerfile
@@ -1,4 +1,4 @@
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 WORKDIR /src
 

diff --git a/test/integration/components/pingclient/Dockerfile b/test/integration/components/pingclient/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH
 

diff --git a/test/integration/components/sqlclient/Dockerfile b/test/integration/components/sqlclient/Dockerfile
@@ -1,6 +1,6 @@
 # Build the testserver binary
 # Docker command must be invoked from the projec root directory
-FROM golang:1.23 as builder
+FROM golang:1.23 AS builder
 
 ARG TARGETARCH