Add pprof to relay server (Layr-Labs#1285)

inabox/deploy/config.go

@@ -344,7 +344,6 @@ func (env *Config) generateControllerVars(
 		CONTROLLER_AVAILABLE_RELAYS:                        "0,1,2,3",
 		CONTROLLER_DISPATCHER_PULL_INTERVAL:                "3s",
 		CONTROLLER_NODE_REQUEST_TIMEOUT:                    "5s",
-		CONTROLLER_NUM_CONNECTIONS_TO_NODES:                "10",
 		CONTROLLER_CHAIN_RPC:                               "",
 		CONTROLLER_PRIVATE_KEY:                             "123",
 		CONTROLLER_NUM_CONFIRMATIONS:                       "0",
@@ -378,6 +377,7 @@ func (env *Config) generateRelayVars(ind int, graphUrl, grpcPort string) RelayVa
 		RELAY_MAX_CONCURRENT_GET_CHUNK_OPS_CLIENT:   "10",
 		RELAY_MAX_GET_CHUNK_BYTES_PER_SECOND_CLIENT: "100000000",
 		RELAY_AUTHENTICATION_DISABLED:               "false",
+		RELAY_ENABLE_METRICS:                        "true",
 	}
 	env.applyDefaults(&v, "RELAY", "relay", ind)
 

relay/cmd/config.go

@@ -94,7 +94,10 @@ func NewConfig(ctx *cli.Context) (Config, error) {
 				InternalGetProofsTimeout:       ctx.Duration(flags.InternalGetProofsTimeoutFlag.Name),
 				InternalGetCoefficientsTimeout: ctx.Duration(flags.InternalGetCoefficientsTimeoutFlag.Name),
 			},
-			MetricsPort: ctx.Int(flags.MetricsPortFlag.Name),
+			MetricsPort:   ctx.Int(flags.MetricsPortFlag.Name),
+			EnableMetrics: ctx.Bool(flags.EnableMetricsFlag.Name),
+			EnablePprof:   ctx.Bool(flags.EnablePprofFlag.Name),
+			PprofHttpPort: ctx.Int(flags.PprofHttpPortFlag.Name),
 		},
 		EthClientConfig:               geth.ReadEthClientConfigRPCOnly(ctx),
 		BLSOperatorStateRetrieverAddr: ctx.String(flags.BlsOperatorStateRetrieverAddrFlag.Name),

relay/cmd/flags/flags.go

@@ -9,6 +9,7 @@ import (
 	"github.com/Layr-Labs/eigenda/common/aws"
 	"github.com/Layr-Labs/eigenda/common/geth"
 	"github.com/Layr-Labs/eigenda/core/thegraph"
+	"github.com/docker/go-units"
 	"github.com/urfave/cli"
 )
 
@@ -291,6 +292,25 @@ var (
 		EnvVar:   common.PrefixEnvVar(envVarPrefix, "METRICS_PORT"),
 		Value:    9101,
 	}
+	EnableMetricsFlag = cli.BoolFlag{
+		Name:     common.PrefixFlag(FlagPrefix, "enable-metrics"),
+		Usage:    "Enable prometheus metrics collection",
+		Required: true,
+		EnvVar:   common.PrefixEnvVar(envVarPrefix, "ENABLE_METRICS"),
+	}
+	EnablePprofFlag = cli.BoolFlag{
+		Name:     common.PrefixFlag(FlagPrefix, "enable-pprof"),
+		Usage:    "Enable pprof profiling",
+		Required: false,
+		EnvVar:   common.PrefixEnvVar(envVarPrefix, "ENABLE_PPROF"),
+	}
+	PprofHttpPortFlag = cli.IntFlag{
+		Name:     common.PrefixFlag(FlagPrefix, "pprof-port"),
+		Usage:    "Port to listen on for pprof",
+		Required: false,
+		EnvVar:   common.PrefixEnvVar(envVarPrefix, "PPROF_PORT"),
+		Value:    6060,
+	}
 )
 
 var requiredFlags = []cli.Flag{
@@ -300,6 +320,7 @@ var requiredFlags = []cli.Flag{
 	RelayKeysFlag,
 	BlsOperatorStateRetrieverAddrFlag,
 	EigenDAServiceManagerAddrFlag,
+	EnableMetricsFlag,
 }
 
 var optionalFlags = []cli.Flag{
@@ -337,6 +358,8 @@ var optionalFlags = []cli.Flag{
 	InternalGetCoefficientsTimeoutFlag,
 	OnchainStateRefreshIntervalFlag,
 	MetricsPortFlag,
+	EnablePprofFlag,
+	PprofHttpPortFlag,
 }
 
 var Flags []cli.Flag

relay/config.go

@@ -1,9 +1,10 @@
 package relay
 
 import (
+	"time"
+
 	v2 "github.com/Layr-Labs/eigenda/core/v2"
 	"github.com/Layr-Labs/eigenda/relay/limiter"
-	"time"
 )
 
 // Config is the configuration for the relay Server.
@@ -64,4 +65,13 @@ type Config struct {
 
 	// MetricsPort is the port that the relay metrics server listens on.
 	MetricsPort int
+
+	// EnableMetrics enables the metrics HTTP server for prometheus metrics collection
+	EnableMetrics bool
+
+	// EnablePprof enables the pprof HTTP server for profiling
+	EnablePprof bool
+
+	// PprofHttpPort is the port that the pprof HTTP server listens on
+	PprofHttpPort int
 }

relay/server.go

@@ -11,6 +11,7 @@ import (
 	"github.com/Layr-Labs/eigenda/api"
 	pb "github.com/Layr-Labs/eigenda/api/grpc/relay"
 	"github.com/Layr-Labs/eigenda/common/healthcheck"
+	"github.com/Layr-Labs/eigenda/common/pprof"
 	"github.com/Layr-Labs/eigenda/core"
 	v2 "github.com/Layr-Labs/eigenda/core/v2"
 	"github.com/Layr-Labs/eigenda/disperser/common/v2/blobstore"
@@ -489,7 +490,18 @@ func buildInsufficientGetChunksBandwidthError(
 
 // Start starts the server listening for requests. This method will block until the server is stopped.
 func (s *Server) Start(ctx context.Context) error {
-	s.metrics.Start()
+	// Start metrics server if enabled
+	if s.config.EnableMetrics {
+		s.metrics.Start()
+		s.logger.Info("Enabled metrics for relay server", "port", s.config.MetricsPort)
+	}
+
+	// Start pprof server if enabled
+	if s.config.EnablePprof {
+		pprofProfiler := pprof.NewPprofProfiler(fmt.Sprintf("%d", s.config.PprofHttpPort), s.logger)
+		go pprofProfiler.Start()
+		s.logger.Info("Enabled pprof for relay server", "port", s.config.PprofHttpPort)
+	}
 
 	if s.chainReader != nil && s.metadataProvider != nil {
 		go func() {
@@ -548,9 +560,11 @@ func (s *Server) Stop() error {
 		s.grpcServer.GracefulStop()
 	}
 
-	err := s.metrics.Stop()
-	if err != nil {
-		return fmt.Errorf("error stopping metrics server: %w", err)
+	if s.config.EnableMetrics {
+		err := s.metrics.Stop()
+		if err != nil {
+			return fmt.Errorf("error stopping metrics server: %w", err)
+		}
 	}
 
 	return nil