Discovery configuration: replace regular expressions by "globs" (#31)

* Move regexp config to Globs

* Fix some integration test service selectors

* Fix the configuration of exe_path in integration tests

* Test Php-fpm integration tests

* fix some k8s tests

Author: mariomac

pkg/beyla/config.go

@@ -4,10 +4,10 @@ import (
 	"fmt"
 	"io"
 	"log/slog"
-	"regexp"
 	"time"
 
 	"github.com/caarlos0/env/v9"
+	"github.com/gobwas/glob"
 	"gopkg.in/yaml.v3"
 
 	"github.com/open-telemetry/opentelemetry-ebpf-instrumentation/pkg/config"
@@ -129,8 +129,8 @@ var DefaultConfig = Config{
 		ExcludeOTelInstrumentedServices: true,
 		DefaultExcludeServices: services.DefinitionCriteria{
 			services.Attributes{
-				// TODO: add final name for opentelemetry-ebpf-instrument executable name
-				Path: services.NewPathRegexp(regexp.MustCompile("(?:^|/)(beyla$|alloy$|otelcol[^/]*$)")),
+				// excluding telemetry components from being instrumented
+				Path: services.NewGlob(glob.MustCompile("{*beyla,*alloy,*ebpf-instrument,*otelcol,*otelcol-contrib,*otelcol-contrib[!/]*}")),
 			},
 		},
 	},
@@ -159,9 +159,11 @@ type Config struct {
 	TracePrinter debug.TracePrinter            `yaml:"trace_printer" env:"OTEL_EBPF_TRACE_PRINTER"`
 
 	// Exec allows selecting the instrumented executable whose complete path contains the Exec value.
-	Exec services.RegexpAttr `yaml:"executable_name" env:"OTEL_EBPF_EXECUTABLE_NAME"`
+	// TODO: setup an EXECUTABLE_NAME property that only cares about the executable name, ignoring the path
+	Exec services.GlobAttr `yaml:"executable_path" env:"OTEL_EBPF_EXECUTABLE_PATH"`
+
 	//nolint:undoc
-	ExecOtelGo services.RegexpAttr `env:"OTEL_GO_AUTO_TARGET_EXE"`
+	ExecOtelGo services.GlobAttr `env:"OTEL_GO_AUTO_TARGET_EXE"`
 	// Port allows selecting the instrumented executable that owns the Port value. If this value is set (and
 	// different to zero), the value of the Exec property won't take effect.
 	// It's important to emphasize that if your process opens multiple HTTP/GRPC ports, the auto-instrumenter
@@ -234,7 +236,7 @@ func (c *Config) Validate() error {
 		return ConfigError("missing to enable application discovery or network metrics. Check documentation")
 	}
 	if (c.Port.Len() > 0 || c.Exec.IsSet() || len(c.Discovery.Services) > 0) && c.Discovery.SystemWide {
-		return ConfigError("you can't use OTEL_EBPF_SYSTEM_WIDE if any of OTEL_EBPF_EXECUTABLE_NAME, OTEL_EBPF_OPEN_PORT or services (YAML) are set")
+		return ConfigError("you can't use OTEL_EBPF_SYSTEM_WIDE if any of OTEL_EBPF_EXECUTABLE_PATH, OTEL_EBPF_OPEN_PORT or services (YAML) are set")
 	}
 	if c.EBPF.BatchLength == 0 {
 		return ConfigError("OTEL_EBPF_BPF_BATCH_LENGTH must be at least 1")
@@ -364,7 +366,7 @@ func LoadConfig(file io.Reader) (*Config, error) {
 		return nil, fmt.Errorf("reading env vars: %w", err)
 	}
 
-	// We support OTEL_GO_AUTO_TARGET_EXE as an alias to OTEL_EBPF_EXECUTABLE_NAME
+	// We support OTEL_GO_AUTO_TARGET_EXE as an alias to OTEL_EBPF_EXECUTABLE_PATH
 	if !cfg.Exec.IsSet() && cfg.ExecOtelGo.IsSet() {
 		cfg.Exec = cfg.ExecOtelGo
 	}

pkg/beyla/config_test.go

@@ -11,6 +11,7 @@ import (
 	"testing"
 	"time"
 
+	"github.com/gobwas/glob"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 
@@ -71,8 +72,11 @@ network:
   enable: true
   cidrs:
     - 10.244.0.0/16
+discovery:
+  default_exclude_services:
+    - exe_path: "*foo*"
 `)
-	t.Setenv("OTEL_EBPF_EXECUTABLE_NAME", "tras")
+	t.Setenv("OTEL_EBPF_EXECUTABLE_PATH", "*tras*")
 	t.Setenv("OTEL_EBPF_NETWORK_AGENT_IP", "1.2.3.4")
 	t.Setenv("OTEL_EBPF_OPEN_PORT", "8080-8089")
 	t.Setenv("OTEL_SERVICE_NAME", "svc-name")
@@ -212,7 +216,7 @@ network:
 			ExcludeOTelInstrumentedServices: true,
 			DefaultExcludeServices: services.DefinitionCriteria{
 				services.Attributes{
-					Path: services.NewPathRegexp(regexp.MustCompile("(?:^|/)(beyla$|alloy$|otelcol[^/]*$)")),
+					Path: services.NewGlob(glob.MustCompile("*foo*")),
 				},
 			},
 		},
@@ -237,15 +241,15 @@ func TestConfig_ShutdownTimeout(t *testing.T) {
 
 func TestConfigValidate(t *testing.T) {
 	testCases := []envMap{
-		{"OTEL_EXPORTER_OTLP_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
-		{"OTEL_EXPORTER_OTLP_METRICS_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
-		{"OTEL_EXPORTER_OTLP_TRACES_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
-		{"OTEL_EBPF_TRACE_PRINTER": "text", "OTEL_EBPF_SHUTDOWN_TIMEOUT": "1m", "OTEL_EBPF_EXECUTABLE_NAME": "foo"},
-		{"OTEL_EBPF_TRACE_PRINTER": "json", "OTEL_EBPF_EXECUTABLE_NAME": "foo"},
-		{"OTEL_EBPF_TRACE_PRINTER": "json_indent", "OTEL_EBPF_EXECUTABLE_NAME": "foo"},
-		{"OTEL_EBPF_TRACE_PRINTER": "counter", "OTEL_EBPF_EXECUTABLE_NAME": "foo"},
-		{"OTEL_EBPF_PROMETHEUS_PORT": "8080", "OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
-		{"OTEL_EBPF_INTERNAL_OTEL_METRICS": "true", "OTEL_EXPORTER_OTLP_METRICS_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_NAME": "foo"},
+		{"OTEL_EXPORTER_OTLP_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
+		{"OTEL_EXPORTER_OTLP_METRICS_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
+		{"OTEL_EXPORTER_OTLP_TRACES_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
+		{"OTEL_EBPF_TRACE_PRINTER": "text", "OTEL_EBPF_SHUTDOWN_TIMEOUT": "1m", "OTEL_EBPF_EXECUTABLE_PATH": "foo"},
+		{"OTEL_EBPF_TRACE_PRINTER": "json", "OTEL_EBPF_EXECUTABLE_PATH": "foo"},
+		{"OTEL_EBPF_TRACE_PRINTER": "json_indent", "OTEL_EBPF_EXECUTABLE_PATH": "foo"},
+		{"OTEL_EBPF_TRACE_PRINTER": "counter", "OTEL_EBPF_EXECUTABLE_PATH": "foo"},
+		{"OTEL_EBPF_PROMETHEUS_PORT": "8080", "OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar"},
+		{"OTEL_EBPF_INTERNAL_OTEL_METRICS": "true", "OTEL_EXPORTER_OTLP_METRICS_ENDPOINT": "localhost:1234", "OTEL_EBPF_EXECUTABLE_PATH": "foo"},
 	}
 	for n, tc := range testCases {
 		t.Run(fmt.Sprint("case", n), func(t *testing.T) {
@@ -257,9 +261,9 @@ func TestConfigValidate(t *testing.T) {
 func TestConfigValidate_error(t *testing.T) {
 	testCases := []envMap{
 		{"OTEL_EXPORTER_OTLP_ENDPOINT": "localhost:1234", "INSTRUMENT_FUNC_NAME": "bar"},
-		{"OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar", "OTEL_EBPF_TRACE_PRINTER": "disabled"},
-		{"OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar", "OTEL_EBPF_TRACE_PRINTER": ""},
-		{"OTEL_EBPF_EXECUTABLE_NAME": "foo", "INSTRUMENT_FUNC_NAME": "bar", "OTEL_EBPF_TRACE_PRINTER": "invalid"},
+		{"OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar", "OTEL_EBPF_TRACE_PRINTER": "disabled"},
+		{"OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar", "OTEL_EBPF_TRACE_PRINTER": ""},
+		{"OTEL_EBPF_EXECUTABLE_PATH": "foo", "INSTRUMENT_FUNC_NAME": "bar", "OTEL_EBPF_TRACE_PRINTER": "invalid"},
 	}
 	for n, tc := range testCases {
 		t.Run(fmt.Sprint("case", n), func(t *testing.T) {
@@ -331,11 +335,11 @@ func TestConfigValidate_TracePrinter(t *testing.T) {
 
 	testCases := []test{
 		{
-			env:      envMap{"OTEL_EBPF_EXECUTABLE_NAME": "foo", "OTEL_EBPF_TRACE_PRINTER": "invalid_printer"},
+			env:      envMap{"OTEL_EBPF_EXECUTABLE_PATH": "foo", "OTEL_EBPF_TRACE_PRINTER": "invalid_printer"},
 			errorMsg: "invalid value for trace_printer: 'invalid_printer'",
 		},
 		{
-			env:      envMap{"OTEL_EBPF_EXECUTABLE_NAME": "foo"},
+			env:      envMap{"OTEL_EBPF_EXECUTABLE_PATH": "foo"},
 			errorMsg: "you need to define at least one exporter: trace_printer, grafana, otel_metrics_export, otel_traces_export or prometheus_export",
 		},
 	}
@@ -352,7 +356,7 @@ func TestConfigValidate_TracePrinter(t *testing.T) {
 }
 
 func TestConfigValidate_TracePrinterFallback(t *testing.T) {
-	env := envMap{"OTEL_EBPF_EXECUTABLE_NAME": "foo", "OTEL_EBPF_TRACE_PRINTER": "text"}
+	env := envMap{"OTEL_EBPF_EXECUTABLE_PATH": "foo", "OTEL_EBPF_TRACE_PRINTER": "text"}
 
 	cfg := loadConfig(t, env)
 	err := cfg.Validate()
@@ -361,7 +365,7 @@ func TestConfigValidate_TracePrinterFallback(t *testing.T) {
 }
 
 func TestConfigValidateRoutes(t *testing.T) {
-	userConfig := bytes.NewBufferString(`executable_name: foo
+	userConfig := bytes.NewBufferString(`executable_path: foo
 trace_printer: text
 routes:
   unmatched: heuristic
@@ -374,12 +378,12 @@ routes:
 
 func TestConfigValidateRoutes_Errors(t *testing.T) {
 	for _, tc := range []string{
-		`executable_name: foo
+		`executable_path: foo
 trace_printer: text
 routes:
   unmatched: heuristic
   wildcard_char: "##"
-`, `executable_name: foo
+`, `executable_path: foo
 trace_printer: text
 routes:
   unmatched: heuristic
@@ -397,16 +401,16 @@ routes:
 }
 
 func TestConfig_OtelGoAutoEnv(t *testing.T) {
-	// OTEL_GO_AUTO_TARGET_EXE is an alias to OTEL_EBPF_EXECUTABLE_NAME
+	// OTEL_GO_AUTO_TARGET_EXE is an alias to OTEL_EBPF_EXECUTABLE_PATH
 	// (Compatibility with OpenTelemetry)
 	t.Setenv("OTEL_GO_AUTO_TARGET_EXE", "testserver")
 	cfg, err := LoadConfig(bytes.NewReader(nil))
 	require.NoError(t, err)
-	assert.True(t, cfg.Exec.IsSet()) // Exec maps to OTEL_EBPF_EXECUTABLE_NAME
+	assert.True(t, cfg.Exec.IsSet()) // Exec maps to OTEL_EBPF_EXECUTABLE_PATH
 }
 
 func TestConfig_NetworkImplicit(t *testing.T) {
-	// OTEL_GO_AUTO_TARGET_EXE is an alias to OTEL_EBPF_EXECUTABLE_NAME
+	// OTEL_GO_AUTO_TARGET_EXE is an alias to OTEL_EBPF_EXECUTABLE_PATH
 	// (Compatibility with OpenTelemetry)
 	t.Setenv("OTEL_EXPORTER_OTLP_ENDPOINT", "http://localhost:4318")
 	t.Setenv("OTEL_EBPF_OTEL_METRIC_FEATURES", "network")
@@ -416,7 +420,7 @@ func TestConfig_NetworkImplicit(t *testing.T) {
 }
 
 func TestConfig_NetworkImplicitProm(t *testing.T) {
-	// OTEL_GO_AUTO_TARGET_EXE is an alias to OTEL_EBPF_EXECUTABLE_NAME
+	// OTEL_GO_AUTO_TARGET_EXE is an alias to OTEL_EBPF_EXECUTABLE_PATH
 	// (Compatibility with OpenTelemetry)
 	t.Setenv("OTEL_EBPF_PROMETHEUS_PORT", "9090")
 	t.Setenv("OTEL_EBPF_PROMETHEUS_FEATURES", "network")

pkg/internal/discover/matcher.go

@@ -6,9 +6,9 @@ import (
 	"fmt"
 	"log/slog"
 	"os"
-	"regexp"
 	"slices"
 
+	"github.com/gobwas/glob"
 	"github.com/shirou/gopsutil/v3/process"
 
 	"github.com/open-telemetry/opentelemetry-ebpf-instrumentation/pkg/beyla"
@@ -145,8 +145,8 @@ func (m *matcher) matchProcess(obj *processAttrs, p *services.ProcessInfo, a *se
 		log.Debug("no Kube metadata, no local selection criteria. Ignoring")
 		return false
 	}
-	if (a.Path.IsSet() || a.PathRegexp.IsSet()) && !m.matchByExecutable(p, a) {
-		log.Debug("executable path does not match", "path", a.Path, "pathregexp", a.PathRegexp)
+	if a.Path.IsSet() && !m.matchByExecutable(p, a) {
+		log.Debug("executable path does not match", "path", a.Path)
 		return false
 	}
 	if a.OpenPorts.Len() > 0 && !m.matchByPort(p, a) {
@@ -169,10 +169,7 @@ func (m *matcher) matchByPort(p *services.ProcessInfo, a *services.Attributes) b
 }
 
 func (m *matcher) matchByExecutable(p *services.ProcessInfo, a *services.Attributes) bool {
-	if a.Path.IsSet() {
-		return a.Path.MatchString(p.ExePath)
-	}
-	return a.PathRegexp.MatchString(p.ExePath)
+	return a.Path.MatchString(p.ExePath)
 }
 
 func (m *matcher) matchByAttributes(actual *processAttrs, required *services.Attributes) bool {
@@ -215,7 +212,7 @@ func FindingCriteria(cfg *beyla.Config) services.DefinitionCriteria {
 		return services.DefinitionCriteria{
 			services.Attributes{
 				Namespace: cfg.ServiceNamespace,
-				Path:      services.NewPathRegexp(regexp.MustCompile(".")),
+				Path:      services.NewGlob(glob.MustCompile("**")),
 			},
 		}
 	}
@@ -238,7 +235,7 @@ func FindingCriteria(cfg *beyla.Config) services.DefinitionCriteria {
 		fc := &finderCriteria[i]
 		if !fc.Path.IsSet() && fc.OpenPorts.Len() == 0 && (len(fc.Metadata) > 0 || len(fc.PodLabels) > 0 || len(fc.PodAnnotations) > 0) {
 			// match any executable path
-			if err := fc.Path.UnmarshalText([]byte(".")); err != nil {
+			if err := fc.Path.UnmarshalText([]byte("**")); err != nil {
 				panic("bug! " + err.Error())
 			}
 		}

pkg/internal/discover/matcher_test.go

@@ -21,10 +21,10 @@ func TestCriteriaMatcher(t *testing.T) {
     namespace: foo
     open_ports: 80,8080-8089
   - name: exec-only
-    exe_path: weird\d
+    exe_path: "*weird[0-9]*"
   - name: both
     open_ports: 443
-    exe_path_regexp: "server"
+    exe_path: "*server*"
 `), &pipeConfig))
 
 	discoveredProcesses := msg.NewQueue[[]Event[processAttrs]](msg.ChannelBufferLen(10))
@@ -84,12 +84,12 @@ func TestCriteriaMatcher_Exclude(t *testing.T) {
     namespace: foo
     open_ports: 80,8080-8089
   - name: exec-only
-    exe_path: weird\d
+    exe_path: "*weird*"
   - name: both
     open_ports: 443
-    exe_path_regexp: "server"
+    exe_path: "*server*"
   exclude_services:
-  - exe_path: s
+  - exe_path: "*s*"
 `), &pipeConfig))
 
 	discoveredProcesses := msg.NewQueue[[]Event[processAttrs]](msg.ChannelBufferLen(10))
@@ -135,7 +135,7 @@ func TestCriteriaMatcher_Exclude_Metadata(t *testing.T) {
 	pipeConfig := beyla.Config{}
 	require.NoError(t, yaml.Unmarshal([]byte(`discovery:
   services:
-  - k8s_node_name: .
+  - k8s_node_name: "*"
   exclude_services:
   - k8s_node_name: bar
 `), &pipeConfig))
@@ -184,11 +184,11 @@ func TestCriteriaMatcher_MustMatchAllAttributes(t *testing.T) {
   - name: all-attributes-must-match
     namespace: foons
     open_ports: 80,8080-8089
-    exe_path: foo
-    k8s_namespace: thens
-    k8s_pod_name: thepod
-    k8s_deployment_name: thedepl
-    k8s_replicaset_name: thers
+    exe_path: "*foo*"
+    k8s_namespace: "*thens*"
+    k8s_pod_name: "*thepod*"
+    k8s_deployment_name: "*thedepl*"
+    k8s_replicaset_name: "*thers*"
 `), &pipeConfig))
 
 	discoveredProcesses := msg.NewQueue[[]Event[processAttrs]](msg.ChannelBufferLen(10))

pkg/internal/discover/watcher_kube_test.go

@@ -132,21 +132,21 @@ func TestWatcherKubeEnricherWithMatcher(t *testing.T) {
     k8s_pod_name: chichi
   - name: both
     open_ports: 443
-    k8s_deployment_name: chacha
+    k8s_deployment_name: chacha*
   - name: pod-label-only
     k8s_pod_labels:
       instrument: "beyla"
   - name: pod-multi-label-only
     k8s_pod_labels:
       instrument: "ebpf"
-      lang: "go.*"
+      lang: "go*"
   - name: pod-annotation-only
     k8s_pod_annotations:
       deploy.type: "canary"
   - name: pod-multi-annotation-only
     k8s_pod_annotations:
       deploy.type: "prod"
-      version: "v\\d+"
+      version: "v[0-9]*"
 `), &pipeConfig))
 
 	swi.Add(CriteriaMatcherProvider(&pipeConfig, connectQueue, outputQueue))