feat: add diagnostic to grpc server

go.mod

@@ -173,3 +173,5 @@ require (
 )
 
 replace github.com/kubernetes/cri-api => k8s.io/cri-api v0.23.5-rc.0
+
+replace github.com/aquasecurity/tracee/types v0.0.0-20230816112254-7c722620761f => ./types

pkg/cmd/flags/logger.go

@@ -291,6 +291,7 @@ func PrepareLogger(logOptions []string, newBinary bool) (logger.LoggingConfig, e
 	llogger := logger.NewLogger(loggerCfg)
 	return logger.LoggingConfig{
 		Logger:        llogger,
+		LoggerConfig:  loggerCfg,
 		Filter:        filter,
 		Aggregate:     agg,
 		FlushInterval: interval,

pkg/cmd/tracee.go

@@ -47,7 +47,7 @@ func (r Runner) Run(ctx context.Context) error {
 
 			// start server if one is configured
 			if r.GRPCServer != nil {
-				go r.GRPCServer.Start(ctx)
+				go r.GRPCServer.Start(ctx, t)
 			}
 		},
 	)

pkg/logger/logger.go

@@ -88,7 +88,7 @@ const (
 // Tracee offers aggregation and filtering support on top of any logger implementation complying to it's interface.
 type LoggingConfig struct {
 	Logger        LoggerInterface
-	loggerConfig  LoggerConfig
+	LoggerConfig  LoggerConfig
 	Filter        LoggerFilter
 	Aggregate     bool
 	FlushInterval time.Duration
@@ -105,7 +105,7 @@ type LoggerConfig struct {
 // SetLevel sets the logging level of the package level logger.
 // It is thread safe.
 func (lc LoggingConfig) SetLevel(lvl Level) {
-	lc.loggerConfig.Level.SetLevel(lvl)
+	lc.LoggerConfig.Level.SetLevel(lvl)
 }
 
 func defaultEncoder() Encoder {
@@ -124,7 +124,7 @@ func NewDefaultLoggingConfig() LoggingConfig {
 	loggerConfig := NewDefaultLoggerConfig()
 	return LoggingConfig{
 		Logger:        NewLogger(loggerConfig),
-		loggerConfig:  loggerConfig,
+		LoggerConfig:  loggerConfig,
 		Filter:        NewLoggerFilter(),
 		Aggregate:     false,
 		FlushInterval: DefaultFlushInterval,

pkg/server/grpc/diagnostic.go

@@ -0,0 +1,75 @@
+package grpc
+
+import (
+	"context"
+	"runtime"
+
+	tracee "github.com/aquasecurity/tracee/pkg/ebpf"
+	"github.com/aquasecurity/tracee/pkg/logger"
+	pb "github.com/aquasecurity/tracee/types/api/v1beta1"
+)
+
+type DiagnosticService struct {
+	pb.UnimplementedDiagnosticServiceServer
+	tracee *tracee.Tracee
+}
+
+func (s *DiagnosticService) GetMetrics(ctx context.Context, in *pb.GetMetricsRequest) (*pb.GetMetricsResponse, error) {
+	stats := s.tracee.Stats()
+	metrics := &pb.GetMetricsResponse{
+		EventCount:       stats.EventCount.Get(),
+		EventsFiltered:   stats.EventsFiltered.Get(),
+		NetCapCount:      stats.NetCapCount.Get(),
+		BPFLogsCount:     stats.BPFLogsCount.Get(),
+		ErrorCount:       stats.ErrorCount.Get(),
+		LostEvCount:      stats.LostEvCount.Get(),
+		LostWrCount:      stats.LostWrCount.Get(),
+		LostNtCapCount:   stats.LostNtCapCount.Get(),
+		LostBPFLogsCount: stats.LostBPFLogsCount.Get(),
+	}
+
+	return metrics, nil
+}
+
+func (s *DiagnosticService) ChangeLogLevel(ctx context.Context, in *pb.ChangeLogLevelRequest) (*pb.ChangeLogLevelResponse, error) {
+	// default level
+	level := logger.InfoLevel
+
+	switch in.Level {
+	case pb.LogLevel_Debug:
+		level = logger.DebugLevel
+	case pb.LogLevel_Warn:
+		level = logger.WarnLevel
+	case pb.LogLevel_Error:
+		level = logger.ErrorLevel
+	case pb.LogLevel_DPanic:
+		level = logger.DPanicLevel
+	case pb.LogLevel_Panic:
+		level = logger.PanicLevel
+	case pb.LogLevel_Fatal:
+		level = logger.FatalLevel
+	}
+
+	logger.SetLevel(level)
+
+	return &pb.ChangeLogLevelResponse{}, nil
+}
+
+func (s *DiagnosticService) GetStacktrace(ctx context.Context, in *pb.GetStacktraceRequest) (*pb.GetStacktraceResponse, error) {
+	return &pb.GetStacktraceResponse{
+		Stacktrace: stack(),
+	}, nil
+}
+
+// This func is based on runtime.Stack(),
+// but instead if pass true to runtime.Stack in order to include all goroutines.
+func stack() []byte {
+	buf := make([]byte, 1024)
+	for {
+		n := runtime.Stack(buf, true)
+		if n < len(buf) {
+			return buf[:n]
+		}
+		buf = make([]byte, 2*len(buf))
+	}
+}

pkg/server/grpc/server.go

@@ -8,6 +8,7 @@ import (
 	"google.golang.org/grpc"
 	"google.golang.org/grpc/keepalive"
 
+	tracee "github.com/aquasecurity/tracee/pkg/ebpf"
 	"github.com/aquasecurity/tracee/pkg/logger"
 	pb "github.com/aquasecurity/tracee/types/api/v1beta1"
 )
@@ -31,7 +32,7 @@ func New(protocol, listenAddr string) (*Server, error) {
 	return &Server{listener: lis, protocol: protocol, listenAddr: listenAddr}, nil
 }
 
-func (s *Server) Start(ctx context.Context) {
+func (s *Server) Start(ctx context.Context, t *tracee.Tracee) {
 	srvCtx, srvCancel := context.WithCancel(ctx)
 	defer srvCancel()
 
@@ -43,6 +44,7 @@ func (s *Server) Start(ctx context.Context) {
 
 	grpcServer := grpc.NewServer(grpc.KeepaliveParams(keepaliveParams))
 	pb.RegisterTraceeServiceServer(grpcServer, &TraceeService{})
+	pb.RegisterDiagnosticServiceServer(grpcServer, &DiagnosticService{tracee: t})
 
 	go func() {
 		logger.Debugw("Starting grpc server", "protocol", s.protocol, "address", s.listenAddr)

pkg/server/grpc/server_test.go

@@ -31,7 +31,7 @@ func TestServer(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	go grpcServer.Start(ctx)
+	go grpcServer.Start(ctx, nil)
 
 	c := grpcClient("unix", unixSock)