Low-cardinality route from path mode

internal/test/integration/configs/obi-config-no-route-lc.yml

@@ -0,0 +1,13 @@
+routes:
+  ignored_patterns:
+    - /metrics
+  unmatched: low-cardinality
+  max_path_segment_cardinality: 3
+otel_metrics_export:
+  endpoint: http://otelcol:4318
+otel_traces_export:
+  endpoint: http://jaeger:4318
+attributes:
+  select:
+    "*":
+      include: ["*"]

internal/test/integration/red_test.go

@@ -881,6 +881,42 @@ func testREDMetricsForHTTPLibraryNoRoute(t *testing.T, url, svcName string) {
 	require.Empty(t, results)
 }
 
+func testREDMetricsForHTTPLibraryNoRouteLowCardinality(t *testing.T, url, svcName string) {
+	validNames := []string{"user", "customer", "test", "option", "metric"}
+
+	// Call 3 times the instrumented service, forcing it to:
+	// - take at least 30ms to respond
+	// - returning a 404 code
+	for i := 0; i < 3; i++ {
+		for _, s := range validNames {
+			ti.DoHTTPGet(t, url+"/api/"+s+"?delay=30ms&status=404", 404)
+		}
+	}
+
+	// Eventually, Prometheus would make this query visible
+	pq := prom.Client{HostPort: prometheusHostPort}
+	var results []prom.Result
+	test.Eventually(t, testTimeout, func(t require.TestingT) {
+		var err error
+		results, err = pq.Query(`http_server_request_duration_seconds_count{` +
+			`http_request_method="GET",` +
+			`http_response_status_code="404",` +
+			`service_namespace="integration-test",` +
+			`service_name="` + svcName + `",` +
+			`http_route="/api/*"}`)
+		require.NoError(t, err)
+		// check duration_count has 3 calls and all the arguments
+		enoughPromResults(t, results)
+		val := totalPromCount(t, results)
+		assert.LessOrEqual(t, 3, val)
+		if len(results) > 0 {
+			res := results[0]
+			addr := res.Metric["client_address"]
+			assert.NotNil(t, addr)
+		}
+	})
+}
+
 func testREDMetricsHTTPNoRoute(t *testing.T) {
 	for _, testCaseURL := range []string{
 		instrumentedServiceGorillaURL,
@@ -892,6 +928,17 @@ func testREDMetricsHTTPNoRoute(t *testing.T) {
 	}
 }
 
+func testREDMetricsHTTPNoRouteLowCardinality(t *testing.T) {
+	for _, testCaseURL := range []string{
+		instrumentedServiceStdURL,
+	} {
+		t.Run(testCaseURL, func(t *testing.T) {
+			waitForTestComponents(t, testCaseURL)
+			testREDMetricsForHTTPLibraryNoRouteLowCardinality(t, testCaseURL, "testserver")
+		})
+	}
+}
+
 func testREDMetricsUnsupportedHTTP(t *testing.T) {
 	for _, testCaseURL := range []string{
 		instrumentedServiceStdURL,

internal/test/integration/suites_test.go

@@ -625,6 +625,16 @@ func TestSuiteNoRoutes(t *testing.T) {
 	require.NoError(t, compose.Close())
 }
 
+func TestSuiteNoRoutesLowCardinality(t *testing.T) {
+	compose, err := docker.ComposeSuite("docker-compose.yml", path.Join(pathOutput, "test-suite-no-routes-low-cardinality.log"))
+	require.NoError(t, err)
+
+	compose.Env = append(compose.Env, "INSTRUMENTER_CONFIG_SUFFIX=-no-route-lc")
+	require.NoError(t, compose.Up())
+	t.Run("RED metrics", testREDMetricsHTTPNoRouteLowCardinality)
+	require.NoError(t, compose.Close())
+}
+
 func TestSuite_Elixir(t *testing.T) {
 	compose, err := docker.ComposeSuite("docker-compose-elixir.yml", path.Join(pathOutput, "test-suite-elixir.log"))
 	require.NoError(t, err)

internal/tools/tools.go

@@ -11,8 +11,9 @@ import (
 	_ "github.com/google/go-licenses/v2"
 	_ "github.com/grafana/go-offsets-tracker/cmd/go-offsets-tracker"
 	_ "github.com/onsi/ginkgo/v2/ginkgo"
-	_ "go.opentelemetry.io/build-tools/multimod"
 	_ "gotest.tools/gotestsum"
 	_ "sigs.k8s.io/controller-runtime/tools/setup-envtest"
 	_ "sigs.k8s.io/kind"
+
+	_ "go.opentelemetry.io/build-tools/multimod"
 )

pkg/appolly/app/svc/svc.go

@@ -10,6 +10,7 @@ import (
 	"go.opentelemetry.io/obi/pkg/appolly/services"
 	attr "go.opentelemetry.io/obi/pkg/export/attributes/names"
 	"go.opentelemetry.io/obi/pkg/internal/transform/route"
+	"go.opentelemetry.io/obi/pkg/internal/transform/route/clusterurl"
 )
 
 type InstrumentableType int
@@ -117,6 +118,7 @@ type Attrs struct {
 	CustomInRouteMatcher  route.Matcher
 	CustomOutRouteMatcher route.Matcher
 	HarvestedRouteMatcher route.Matcher
+	PathTrie              *clusterurl.PathTrie
 }
 
 func (i *Attrs) GetUID() UID {

pkg/appolly/discover/matcher_test.go

@@ -14,6 +14,7 @@ import (
 	"go.opentelemetry.io/obi/pkg/internal/testutil"
 	"go.opentelemetry.io/obi/pkg/obi"
 	"go.opentelemetry.io/obi/pkg/pipe/msg"
+	"go.opentelemetry.io/obi/pkg/transform"
 )
 
 func testMatch(t *testing.T, m Event[ProcessMatch], name string,
@@ -551,7 +552,7 @@ func TestCriteriaMatcher_Granular(t *testing.T) {
 
 	require.Len(t, planetMatch.Criteria, 2)
 
-	planetAttrs := makeServiceAttrs(&planetMatch)
+	planetAttrs := makeServiceAttrs(&planetMatch, &transform.RoutesConfig{})
 
 	assert.True(t, planetAttrs.ExportModes.CanExportTraces())
 	assert.False(t, planetAttrs.ExportModes.CanExportMetrics())
@@ -562,7 +563,7 @@ func TestCriteriaMatcher_Granular(t *testing.T) {
 
 	require.Len(t, satelliteMatch.Criteria, 2)
 
-	satelliteAttrs := makeServiceAttrs(&satelliteMatch)
+	satelliteAttrs := makeServiceAttrs(&satelliteMatch, &transform.RoutesConfig{})
 
 	assert.False(t, satelliteAttrs.ExportModes.CanExportTraces())
 	assert.False(t, satelliteAttrs.ExportModes.CanExportMetrics())
@@ -572,7 +573,7 @@ func TestCriteriaMatcher_Granular(t *testing.T) {
 
 	require.Len(t, starMatch.Criteria, 2)
 
-	starAttrs := makeServiceAttrs(&starMatch)
+	starAttrs := makeServiceAttrs(&starMatch, &transform.RoutesConfig{})
 
 	assert.False(t, starAttrs.ExportModes.CanExportTraces())
 	assert.True(t, starAttrs.ExportModes.CanExportMetrics())
@@ -582,7 +583,7 @@ func TestCriteriaMatcher_Granular(t *testing.T) {
 
 	require.Len(t, asteroidMatch.Criteria, 2)
 
-	asteroidAttrs := makeServiceAttrs(&asteroidMatch)
+	asteroidAttrs := makeServiceAttrs(&asteroidMatch, &transform.RoutesConfig{})
 
 	assert.True(t, asteroidAttrs.ExportModes.CanExportTraces())
 	assert.True(t, asteroidAttrs.ExportModes.CanExportMetrics())

pkg/appolly/discover/typer.go

@@ -21,11 +21,13 @@ import (
 	"go.opentelemetry.io/obi/pkg/export/imetrics"
 	"go.opentelemetry.io/obi/pkg/internal/goexec"
 	"go.opentelemetry.io/obi/pkg/internal/procs"
+	"go.opentelemetry.io/obi/pkg/internal/transform/route/clusterurl"
 	"go.opentelemetry.io/obi/pkg/kube"
 	"go.opentelemetry.io/obi/pkg/obi"
 	"go.opentelemetry.io/obi/pkg/pipe/msg"
 	"go.opentelemetry.io/obi/pkg/pipe/swarm"
 	"go.opentelemetry.io/obi/pkg/pipe/swarm/swarms"
+	"go.opentelemetry.io/obi/pkg/transform"
 )
 
 type instrumentedExecutable struct {
@@ -87,7 +89,7 @@ func samplerFromConfig(s *services.SamplerConfig) trace.Sampler {
 	return nil
 }
 
-func makeServiceAttrs(processMatch *ProcessMatch) svc.Attrs {
+func makeServiceAttrs(processMatch *ProcessMatch, routesCfg *transform.RoutesConfig) svc.Attrs {
 	var name string
 	var namespace string
 	exportModes := services.ExportModeUnset
@@ -116,6 +118,11 @@ func makeServiceAttrs(processMatch *ProcessMatch) svc.Attrs {
 		}
 	}
 
+	wildcard := byte('*')
+	if routesCfg.WildcardChar != "" {
+		wildcard = routesCfg.WildcardChar[0]
+	}
+
 	s := svc.Attrs{
 		UID: svc.UID{
 			Name:      name,
@@ -124,6 +131,7 @@ func makeServiceAttrs(processMatch *ProcessMatch) svc.Attrs {
 		ProcPID:     processMatch.Process.Pid,
 		ExportModes: exportModes,
 		Sampler:     samplerFromConfig(samplerConfig),
+		PathTrie:    clusterurl.NewPathTrie(routesCfg.MaxPathSegmentCardinality, wildcard),
 	}
 
 	if routesConfig != nil {
@@ -146,7 +154,7 @@ func (t *typer) FilterClassify(evs []Event[ProcessMatch]) []Event[ebpf.Instrumen
 		ev := &evs[i]
 		switch evs[i].Type {
 		case EventCreated:
-			svcID := makeServiceAttrs(&ev.Obj)
+			svcID := makeServiceAttrs(&ev.Obj, t.cfg.Routes)
 
 			if elfFile, err := findExecElf(ev.Obj.Process, svcID, t.k8sInformer.IsKubeEnabled()); err != nil {
 				t.log.Debug("error finding process ELF. Ignoring", "error", err)

pkg/appolly/discover/typer_test.go

@@ -10,6 +10,7 @@ import (
 	"github.com/stretchr/testify/assert"
 
 	"go.opentelemetry.io/obi/pkg/appolly/services"
+	"go.opentelemetry.io/obi/pkg/transform"
 )
 
 type dummyCriterion struct {
@@ -41,7 +42,7 @@ func TestMakeServiceAttrs(t *testing.T) {
 			dummyCriterion{name: "svc1", namespace: "ns1", export: services.ExportModeUnset},
 		},
 	}
-	attrs := makeServiceAttrs(proc)
+	attrs := makeServiceAttrs(proc, &transform.RoutesConfig{})
 	assert.Equal(t, "svc1", attrs.UID.Name)
 	assert.Equal(t, "ns1", attrs.UID.Namespace)
 	assert.Equal(t, int32(1234), attrs.ProcPID)
@@ -60,7 +61,7 @@ func TestMakeServiceAttrs(t *testing.T) {
 			dummyCriterion{sampler: sampler, routes: routes},
 		},
 	}
-	attrs2 := makeServiceAttrs(proc2)
+	attrs2 := makeServiceAttrs(proc2, &transform.RoutesConfig{})
 	assert.NotNil(t, attrs2.Sampler)
 	assert.NotNil(t, attrs2.CustomInRouteMatcher)
 	assert.NotNil(t, attrs2.CustomOutRouteMatcher)

pkg/internal/transform/route/clusterurl/cluster_test.go

@@ -66,6 +66,11 @@ func TestClusterURL(t *testing.T) {
 	assert.Equal(t, "/*", csf.ClusterURL("/1#"))
 	assert.Equal(t, "a", csf.ClusterURL("a#"))
 	assert.Equal(t, "/a/b/c/d/e/f/g/h/i", csf.ClusterURL("/a/b/c/d/e/f/g/h/i/j"))
+	assert.Equal(t, "/api/user", csf.ClusterURL("/api/user"))
+	assert.Equal(t, "/api/customer", csf.ClusterURL("/api/customer"))
+	assert.Equal(t, "/api/test", csf.ClusterURL("/api/test"))
+	assert.Equal(t, "/api/option", csf.ClusterURL("/api/option"))
+	assert.Equal(t, "/api/metric", csf.ClusterURL("/api/metric"))
 }
 
 func BenchmarkClusterURLWithCache(b *testing.B) {