TUN-8456: Update quic-go to 0.45 and collect mtu and congestion control metrics

Author: chungthuang

component-tests/test_tail.py

@@ -47,7 +47,12 @@ async def test_streaming_logs(self, tmp_path, component_tests_config):
             url = cfd_cli.get_management_wsurl("logs", config, config_path)
             async with connect(url, open_timeout=5, close_timeout=5) as websocket:
                 # send start_streaming
-                await websocket.send('{"type": "start_streaming"}')
+                await websocket.send(json.dumps({
+                    "type": "start_streaming",
+                    "filters": {
+                        "events": ["http"]
+                    }
+                }))
                 # send some http requests to the tunnel to trigger some logs
                 await generate_and_validate_http_events(websocket, config.get_url(), 10)
                 # send stop_streaming
@@ -99,7 +104,8 @@ async def test_streaming_logs_sampling(self, tmp_path, component_tests_config):
                 await websocket.send(json.dumps({
                     "type": "start_streaming",
                     "filters": {
-                        "sampling": 0.5
+                        "sampling": 0.5,
+                        "events": ["http"]
                     }
                 }))
                 # don't expect any http logs

go.mod

@@ -23,9 +23,9 @@ require (
 	github.com/miekg/dns v1.1.50
 	github.com/mitchellh/go-homedir v1.1.0
 	github.com/pkg/errors v0.9.1
-	github.com/prometheus/client_golang v1.13.0
-	github.com/prometheus/client_model v0.2.0
-	github.com/quic-go/quic-go v0.42.0
+	github.com/prometheus/client_golang v1.19.1
+	github.com/prometheus/client_model v0.5.0
+	github.com/quic-go/quic-go v0.45.0
 	github.com/rs/zerolog v1.20.0
 	github.com/stretchr/testify v1.9.0
 	github.com/urfave/cli/v2 v2.3.0
@@ -38,7 +38,7 @@ require (
 	go.uber.org/automaxprocs v1.4.0
 	golang.org/x/crypto v0.23.0
 	golang.org/x/net v0.25.0
-	golang.org/x/sync v0.6.0
+	golang.org/x/sync v0.7.0
 	golang.org/x/sys v0.20.0
 	golang.org/x/term v0.20.0
 	google.golang.org/protobuf v1.34.1
@@ -80,16 +80,16 @@ require (
 	github.com/onsi/ginkgo/v2 v2.9.5 // indirect
 	github.com/opentracing/opentracing-go v1.2.0 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
-	github.com/prometheus/common v0.37.0 // indirect
-	github.com/prometheus/procfs v0.8.0 // indirect
+	github.com/prometheus/common v0.48.0 // indirect
+	github.com/prometheus/procfs v0.12.0 // indirect
 	github.com/russross/blackfriday/v2 v2.1.0 // indirect
 	go.opentelemetry.io/otel/metric v1.26.0 // indirect
 	go.uber.org/mock v0.4.0 // indirect
-	golang.org/x/exp v0.0.0-20221205204356-47842c84f3db // indirect
-	golang.org/x/mod v0.11.0 // indirect
+	golang.org/x/exp v0.0.0-20240506185415-9bf2ced13842 // indirect
+	golang.org/x/mod v0.17.0 // indirect
 	golang.org/x/oauth2 v0.17.0 // indirect
 	golang.org/x/text v0.15.0 // indirect
-	golang.org/x/tools v0.9.1 // indirect
+	golang.org/x/tools v0.21.0 // indirect
 	google.golang.org/appengine v1.6.8 // indirect
 	google.golang.org/genproto/googleapis/api v0.0.0-20240227224415-6ceb2ff114de // indirect
 	google.golang.org/genproto/googleapis/rpc v0.0.0-20240227224415-6ceb2ff114de // indirect

go.sum

@@ -30,6 +30,9 @@ var (
 		minRTT            *prometheus.GaugeVec
 		latestRTT         *prometheus.GaugeVec
 		smoothedRTT       *prometheus.GaugeVec
+		mtu               *prometheus.GaugeVec
+		congestionWindow  *prometheus.GaugeVec
+		congestionState   *prometheus.GaugeVec
 	}{
 		totalConnections: prometheus.NewCounter(
 			prometheus.CounterOpts{
@@ -146,6 +149,33 @@ var (
 			},
 			clientConnLabels,
 		),
+		mtu: prometheus.NewGaugeVec(
+			prometheus.GaugeOpts{
+				Namespace: namespace,
+				Subsystem: "client",
+				Name:      "mtu",
+				Help:      "Current maximum transmission unit (MTU) of a connection",
+			},
+			clientConnLabels,
+		),
+		congestionWindow: prometheus.NewGaugeVec(
+			prometheus.GaugeOpts{
+				Namespace: namespace,
+				Subsystem: "client",
+				Name:      "congestion_window",
+				Help:      "Current congestion window size",
+			},
+			clientConnLabels,
+		),
+		congestionState: prometheus.NewGaugeVec(
+			prometheus.GaugeOpts{
+				Namespace: namespace,
+				Subsystem: "client",
+				Name:      "congestion_state",
+				Help:      "Current congestion control state. See https://pkg.go.dev/github.com/quic-go/[email protected]/logging#CongestionState for what each value maps to",
+			},
+			clientConnLabels,
+		),
 	}
 
 	registerClient = sync.Once{}
@@ -179,6 +209,9 @@ func newClientCollector(index string, logger *zerolog.Logger) *clientCollector {
 			clientMetrics.minRTT,
 			clientMetrics.latestRTT,
 			clientMetrics.smoothedRTT,
+			clientMetrics.mtu,
+			clientMetrics.congestionWindow,
+			clientMetrics.congestionState,
 			packetTooBigDropped,
 		)
 	})
@@ -232,6 +265,19 @@ func (cc *clientCollector) updatedRTT(rtt *logging.RTTStats) {
 	clientMetrics.smoothedRTT.WithLabelValues(cc.index).Set(durationToPromGauge(rtt.SmoothedRTT()))
 }
 
+func (cc *clientCollector) updateCongestionWindow(size logging.ByteCount) {
+	clientMetrics.congestionWindow.WithLabelValues(cc.index).Set(float64(size))
+}
+
+func (cc *clientCollector) updatedCongestionState(state logging.CongestionState) {
+	clientMetrics.congestionState.WithLabelValues(cc.index).Set(float64(state))
+}
+
+func (cc *clientCollector) updateMTU(mtu logging.ByteCount) {
+	clientMetrics.mtu.WithLabelValues(cc.index).Set(float64(mtu))
+	cc.logger.Debug().Msgf("QUIC MTU updated to %d", mtu)
+}
+
 func (cc *clientCollector) collectPackets(size logging.ByteCount, frames []logging.Frame, counter, bandwidth *prometheus.CounterVec, direction direction) {
 	for _, frame := range frames {
 		switch f := frame.(type) {

quic/metrics.go

@@ -47,6 +47,8 @@ func newConnTracer(metricsCollector *clientCollector) *logging.ConnectionTracer
 		DroppedPacket:               tracer.DroppedPacket,
 		UpdatedMetrics:              tracer.UpdatedMetrics,
 		LostPacket:                  tracer.LostPacket,
+		UpdatedMTU:                  tracer.UpdatedMTU,
+		UpdatedCongestionState:      tracer.UpdatedCongestionState,
 	}
 }
 
@@ -76,6 +78,7 @@ func (ct *connTracer) LostPacket(level logging.EncryptionLevel, number logging.P
 
 func (ct *connTracer) UpdatedMetrics(rttStats *logging.RTTStats, cwnd, bytesInFlight logging.ByteCount, packetsInFlight int) {
 	ct.metricsCollector.updatedRTT(rttStats)
+	ct.metricsCollector.updateCongestionWindow(cwnd)
 }
 
 func (ct *connTracer) SentLongHeaderPacket(hdr *logging.ExtendedHeader, size logging.ByteCount, ecn logging.ECN, ack *logging.AckFrame, frames []logging.Frame) {
@@ -93,3 +96,11 @@ func (ct *connTracer) ReceivedLongHeaderPacket(hdr *logging.ExtendedHeader, size
 func (ct *connTracer) ReceivedShortHeaderPacket(hdr *logging.ShortHeader, size logging.ByteCount, ecn logging.ECN, frames []logging.Frame) {
 	ct.metricsCollector.receivedPackets(size, frames)
 }
+
+func (ct *connTracer) UpdatedMTU(mtu logging.ByteCount, done bool) {
+	ct.metricsCollector.updateMTU(mtu)
+}
+
+func (ct *connTracer) UpdatedCongestionState(state logging.CongestionState) {
+	ct.metricsCollector.updatedCongestionState(state)
+}

quic/tracing.go

@@ -569,6 +569,13 @@ func (e *EdgeTunnelServer) serveQUIC(
 
 	tlsConfig.CurvePreferences = curvePref
 
+	// quic-go 0.44 increases the initial packet size to 1280 by default. That breaks anyone running tunnel through WARP
+	// because WARP MTU is 1280.
+	var initialPacketSize uint16 = 1252
+	if edgeAddr.IP.To4() == nil {
+		initialPacketSize = 1232
+	}
+
 	quicConfig := &quic.Config{
 		HandshakeIdleTimeout:       quicpogs.HandshakeIdleTimeout,
 		MaxIdleTimeout:             quicpogs.MaxIdleTimeout,
@@ -580,6 +587,7 @@ func (e *EdgeTunnelServer) serveQUIC(
 		DisablePathMTUDiscovery:    e.config.DisableQUICPathMTUDiscovery,
 		MaxConnectionReceiveWindow: e.config.QUICConnectionLevelFlowControlLimit,
 		MaxStreamReceiveWindow:     e.config.QUICStreamLevelFlowControlLimit,
+		InitialPacketSize:          initialPacketSize,
 	}
 
 	quicConn, err := connection.NewQUICConnection(