metrics.go 25.9 KB
Newer Older
1
// Package metrics provides a set of metrics for the op-node.
2 3 4 5
package metrics

import (
	"context"
6
	"encoding/binary"
7 8 9 10
	"errors"
	"fmt"
	"net"
	"strconv"
11
	"time"
12

13
	ophttp "github.com/ethereum-optimism/optimism/op-node/http"
14 15
	"github.com/ethereum-optimism/optimism/op-service/metrics"

16
	pb "github.com/libp2p/go-libp2p-pubsub/pb"
17
	libp2pmetrics "github.com/libp2p/go-libp2p/core/metrics"
18 19 20
	"github.com/prometheus/client_golang/prometheus"
	"github.com/prometheus/client_golang/prometheus/collectors"
	"github.com/prometheus/client_golang/prometheus/promhttp"
21

22
	"github.com/ethereum/go-ethereum"
23
	"github.com/ethereum/go-ethereum/common"
24
	"github.com/ethereum/go-ethereum/rpc"
25 26

	"github.com/ethereum-optimism/optimism/op-node/eth"
27 28 29 30 31 32 33 34 35 36 37
)

const (
	Namespace = "op_node"

	RPCServerSubsystem = "rpc_server"
	RPCClientSubsystem = "rpc_client"

	BatchMethod = "<batch>"
)

38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55
type Metricer interface {
	RecordInfo(version string)
	RecordUp()
	RecordRPCServerRequest(method string) func()
	RecordRPCClientRequest(method string) func(err error)
	RecordRPCClientResponse(method string, err error)
	SetDerivationIdle(status bool)
	RecordPipelineReset()
	RecordSequencingError()
	RecordPublishingError()
	RecordDerivationError()
	RecordReceivedUnsafePayload(payload *eth.ExecutionPayload)
	recordRef(layer string, name string, num uint64, timestamp uint64, h common.Hash)
	RecordL1Ref(name string, ref eth.L1BlockRef)
	RecordL2Ref(name string, ref eth.L2BlockRef)
	RecordUnsafePayloadsBuffer(length uint64, memSize uint64, next eth.BlockID)
	CountSequencedTxs(count int)
	RecordL1ReorgDepth(d uint64)
56 57
	RecordSequencerInconsistentL1Origin(from eth.BlockID, to eth.BlockID)
	RecordSequencerReset()
58 59 60 61 62 63
	RecordGossipEvent(evType int32)
	IncPeerCount()
	DecPeerCount()
	IncStreamCount()
	DecStreamCount()
	RecordBandwidth(ctx context.Context, bwc *libp2pmetrics.BandwidthCounter)
64 65
	RecordSequencerBuildingDiffTime(duration time.Duration)
	RecordSequencerSealingTime(duration time.Duration)
66
	Document() []metrics.DocumentedMetric
67
	RecordChannelInputBytes(num int)
68
	// P2P Metrics
69
	SetPeerScores(scores map[string]float64)
70 71 72
	ClientPayloadByNumberEvent(num uint64, resultCode byte, duration time.Duration)
	ServerPayloadByNumberEvent(num uint64, resultCode byte, duration time.Duration)
	PayloadsQuarantineSize(n int)
73 74 75 76
	RecordPeerUnban()
	RecordIPUnban()
	RecordDial(allow bool)
	RecordAccept(allow bool)
77 78
}

79
// Metrics tracks all the metrics for the op-node.
80
type Metrics struct {
81 82 83
	Info *prometheus.GaugeVec
	Up   prometheus.Gauge

84 85 86 87 88 89
	RPCServerRequestsTotal          *prometheus.CounterVec
	RPCServerRequestDurationSeconds *prometheus.HistogramVec
	RPCClientRequestsTotal          *prometheus.CounterVec
	RPCClientRequestDurationSeconds *prometheus.HistogramVec
	RPCClientResponsesTotal         *prometheus.CounterVec

90
	L1SourceCache *CacheMetrics
91
	L2SourceCache *CacheMetrics
92

93 94 95 96 97 98 99 100
	DerivationIdle prometheus.Gauge

	PipelineResets   *EventMetrics
	UnsafePayloads   *EventMetrics
	DerivationErrors *EventMetrics
	SequencingErrors *EventMetrics
	PublishingErrors *EventMetrics

101 102 103 104 105 106
	P2PReqDurationSeconds *prometheus.HistogramVec
	P2PReqTotal           *prometheus.CounterVec
	P2PPayloadByNumber    *prometheus.GaugeVec

	PayloadsQuarantineTotal prometheus.Gauge

107 108 109
	SequencerInconsistentL1Origin *EventMetrics
	SequencerResets               *EventMetrics

110 111
	L1RequestDurationSeconds *prometheus.HistogramVec

112 113 114 115 116 117
	SequencerBuildingDiffDurationSeconds prometheus.Histogram
	SequencerBuildingDiffTotal           prometheus.Counter

	SequencerSealingDurationSeconds prometheus.Histogram
	SequencerSealingTotal           prometheus.Counter

118 119 120
	UnsafePayloadsBufferLen     prometheus.Gauge
	UnsafePayloadsBufferMemSize prometheus.Gauge

121 122 123 124 125 126 127 128 129 130
	RefsNumber  *prometheus.GaugeVec
	RefsTime    *prometheus.GaugeVec
	RefsHash    *prometheus.GaugeVec
	RefsSeqNr   *prometheus.GaugeVec
	RefsLatency *prometheus.GaugeVec
	// hash of the last seen block per name, so we don't reduce/increase latency on updates of the same data,
	// and only count the first occurrence
	LatencySeen map[string]common.Hash

	L1ReorgDepth prometheus.Histogram
131

132 133
	TransactionsSequencedTotal prometheus.Counter

134 135 136
	// P2P Metrics
	PeerCount         prometheus.Gauge
	StreamCount       prometheus.Gauge
137
	PeerScores        *prometheus.GaugeVec
138 139
	GossipEventsTotal *prometheus.CounterVec
	BandwidthTotal    *prometheus.GaugeVec
140 141 142 143
	PeerUnbans        prometheus.Counter
	IPUnbans          prometheus.Counter
	Dials             *prometheus.CounterVec
	Accepts           *prometheus.CounterVec
144

145
	ChannelInputBytes prometheus.Counter
146

147
	registry *prometheus.Registry
148
	factory  metrics.Factory
149 150
}

151 152
var _ Metricer = (*Metrics)(nil)

Andreas Bigger's avatar
Andreas Bigger committed
153
// NewMetrics creates a new [Metrics] instance with the given process name.
154 155 156 157 158 159 160 161 162
func NewMetrics(procName string) *Metrics {
	if procName == "" {
		procName = "default"
	}
	ns := Namespace + "_" + procName

	registry := prometheus.NewRegistry()
	registry.MustRegister(collectors.NewProcessCollector(collectors.ProcessCollectorOpts{}))
	registry.MustRegister(collectors.NewGoCollector())
163
	factory := metrics.With(registry)
164
	return &Metrics{
165
		Info: factory.NewGaugeVec(prometheus.GaugeOpts{
166 167 168 169 170 171
			Namespace: ns,
			Name:      "info",
			Help:      "Pseudo-metric tracking version and config info",
		}, []string{
			"version",
		}),
172
		Up: factory.NewGauge(prometheus.GaugeOpts{
173 174 175 176
			Namespace: ns,
			Name:      "up",
			Help:      "1 if the op node has finished starting up",
		}),
177

178
		RPCServerRequestsTotal: factory.NewCounterVec(prometheus.CounterOpts{
179 180 181 182 183 184 185
			Namespace: ns,
			Subsystem: RPCServerSubsystem,
			Name:      "requests_total",
			Help:      "Total requests to the RPC server",
		}, []string{
			"method",
		}),
186
		RPCServerRequestDurationSeconds: factory.NewHistogramVec(prometheus.HistogramOpts{
187 188 189 190 191 192 193 194
			Namespace: ns,
			Subsystem: RPCServerSubsystem,
			Name:      "request_duration_seconds",
			Buckets:   []float64{.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10},
			Help:      "Histogram of RPC server request durations",
		}, []string{
			"method",
		}),
195
		RPCClientRequestsTotal: factory.NewCounterVec(prometheus.CounterOpts{
196 197 198 199 200 201 202
			Namespace: ns,
			Subsystem: RPCClientSubsystem,
			Name:      "requests_total",
			Help:      "Total RPC requests initiated by the opnode's RPC client",
		}, []string{
			"method",
		}),
203
		RPCClientRequestDurationSeconds: factory.NewHistogramVec(prometheus.HistogramOpts{
204 205 206 207 208 209 210 211
			Namespace: ns,
			Subsystem: RPCClientSubsystem,
			Name:      "request_duration_seconds",
			Buckets:   []float64{.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10},
			Help:      "Histogram of RPC client request durations",
		}, []string{
			"method",
		}),
212
		RPCClientResponsesTotal: factory.NewCounterVec(prometheus.CounterOpts{
213 214 215 216 217 218 219 220
			Namespace: ns,
			Subsystem: RPCClientSubsystem,
			Name:      "responses_total",
			Help:      "Total RPC request responses received by the opnode's RPC client",
		}, []string{
			"method",
			"error",
		}),
221

222 223
		L1SourceCache: NewCacheMetrics(factory, ns, "l1_source_cache", "L1 Source cache"),
		L2SourceCache: NewCacheMetrics(factory, ns, "l2_source_cache", "L2 Source cache"),
224

225
		DerivationIdle: factory.NewGauge(prometheus.GaugeOpts{
226 227 228 229
			Namespace: ns,
			Name:      "derivation_idle",
			Help:      "1 if the derivation pipeline is idle",
		}),
230

231 232 233 234 235
		PipelineResets:   NewEventMetrics(factory, ns, "pipeline_resets", "derivation pipeline resets"),
		UnsafePayloads:   NewEventMetrics(factory, ns, "unsafe_payloads", "unsafe payloads"),
		DerivationErrors: NewEventMetrics(factory, ns, "derivation_errors", "derivation errors"),
		SequencingErrors: NewEventMetrics(factory, ns, "sequencing_errors", "sequencing errors"),
		PublishingErrors: NewEventMetrics(factory, ns, "publishing_errors", "p2p publishing errors"),
236

237 238 239
		SequencerInconsistentL1Origin: NewEventMetrics(factory, ns, "sequencer_inconsistent_l1_origin", "events when the sequencer selects an inconsistent L1 origin"),
		SequencerResets:               NewEventMetrics(factory, ns, "sequencer_resets", "sequencer resets"),

240
		UnsafePayloadsBufferLen: factory.NewGauge(prometheus.GaugeOpts{
241 242 243 244
			Namespace: ns,
			Name:      "unsafe_payloads_buffer_len",
			Help:      "Number of buffered L2 unsafe payloads",
		}),
245
		UnsafePayloadsBufferMemSize: factory.NewGauge(prometheus.GaugeOpts{
246 247 248 249 250
			Namespace: ns,
			Name:      "unsafe_payloads_buffer_mem_size",
			Help:      "Total estimated memory size of buffered L2 unsafe payloads",
		}),

251
		RefsNumber: factory.NewGaugeVec(prometheus.GaugeOpts{
252
			Namespace: ns,
253 254 255 256 257
			Name:      "refs_number",
			Help:      "Gauge representing the different L1/L2 reference block numbers",
		}, []string{
			"layer",
			"type",
258
		}),
259
		RefsTime: factory.NewGaugeVec(prometheus.GaugeOpts{
260
			Namespace: ns,
261 262 263 264 265
			Name:      "refs_time",
			Help:      "Gauge representing the different L1/L2 reference block timestamps",
		}, []string{
			"layer",
			"type",
266
		}),
267
		RefsHash: factory.NewGaugeVec(prometheus.GaugeOpts{
268
			Namespace: ns,
269 270 271 272 273
			Name:      "refs_hash",
			Help:      "Gauge representing the different L1/L2 reference block hashes truncated to float values",
		}, []string{
			"layer",
			"type",
274
		}),
275
		RefsSeqNr: factory.NewGaugeVec(prometheus.GaugeOpts{
276
			Namespace: ns,
277 278 279 280
			Name:      "refs_seqnr",
			Help:      "Gauge representing the different L2 reference sequence numbers",
		}, []string{
			"type",
281
		}),
282
		RefsLatency: factory.NewGaugeVec(prometheus.GaugeOpts{
283
			Namespace: ns,
284 285
			Name:      "refs_latency",
			Help:      "Gauge representing the different L1/L2 reference block timestamps minus current time, in seconds",
286
		}, []string{
287
			"layer",
288 289
			"type",
		}),
290 291
		LatencySeen: make(map[string]common.Hash),

292
		L1ReorgDepth: factory.NewHistogram(prometheus.HistogramOpts{
293 294 295 296 297
			Namespace: ns,
			Name:      "l1_reorg_depth",
			Buckets:   []float64{0.5, 1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5, 9.5, 10.5, 20.5, 50.5, 100.5},
			Help:      "Histogram of L1 Reorg Depths",
		}),
298

299
		TransactionsSequencedTotal: factory.NewGauge(prometheus.GaugeOpts{
300 301 302 303 304
			Namespace: ns,
			Name:      "transactions_sequenced_total",
			Help:      "Count of total transactions sequenced",
		}),

305
		PeerCount: factory.NewGauge(prometheus.GaugeOpts{
306 307 308 309 310
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "peer_count",
			Help:      "Count of currently connected p2p peers",
		}),
311 312 313 314 315
		// Notice: We cannot use peer ids as [Labels] in the GaugeVec
		// since peer ids would open a service attack vector.
		// Each peer id would be a separate metric, flooding prometheus.
		//
		// [Labels]: https://prometheus.io/docs/practices/naming/#labels
316
		PeerScores: factory.NewGaugeVec(prometheus.GaugeOpts{
asnared's avatar
asnared committed
317 318 319
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "peer_scores",
320
			Help:      "Count of peer scores grouped by score",
asnared's avatar
asnared committed
321
		}, []string{
322 323 324 325 326 327 328
			"band",
		}),
		StreamCount: factory.NewGauge(prometheus.GaugeOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "stream_count",
			Help:      "Count of currently connected p2p streams",
asnared's avatar
asnared committed
329
		}),
330
		GossipEventsTotal: factory.NewCounterVec(prometheus.CounterOpts{
331 332 333 334 335 336 337
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "gossip_events_total",
			Help:      "Count of gossip events by type",
		}, []string{
			"type",
		}),
338
		BandwidthTotal: factory.NewGaugeVec(prometheus.GaugeOpts{
339 340 341 342 343 344 345
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "bandwidth_bytes_total",
			Help:      "P2P bandwidth by direction",
		}, []string{
			"direction",
		}),
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
		PeerUnbans: factory.NewCounter(prometheus.CounterOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "peer_unbans",
			Help:      "Count of peer unbans",
		}),
		IPUnbans: factory.NewCounter(prometheus.CounterOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "ip_unbans",
			Help:      "Count of IP unbans",
		}),
		Dials: factory.NewCounterVec(prometheus.CounterOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "dials",
			Help:      "Count of outgoing dial attempts, with label to filter to allowed attempts",
		}, []string{"allow"}),
		Accepts: factory.NewCounterVec(prometheus.CounterOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "accepts",
			Help:      "Count of incoming dial attempts to accept, with label to filter to allowed attempts",
		}, []string{"allow"}),
370

371
		ChannelInputBytes: factory.NewCounter(prometheus.CounterOpts{
372 373 374 375 376
			Namespace: ns,
			Name:      "channel_input_bytes",
			Help:      "Number of compressed bytes added to the channel",
		}),

377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414
		P2PReqDurationSeconds: factory.NewHistogramVec(prometheus.HistogramOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "req_duration_seconds",
			Buckets:   []float64{},
			Help:      "Duration of P2P requests",
		}, []string{
			"p2p_role", // "client" or "server"
			"p2p_method",
			"result_code",
		}),

		P2PReqTotal: factory.NewCounterVec(prometheus.CounterOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "req_total",
			Help:      "Number of P2P requests",
		}, []string{
			"p2p_role", // "client" or "server"
			"p2p_method",
			"result_code",
		}),

		P2PPayloadByNumber: factory.NewGaugeVec(prometheus.GaugeOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "payload_by_number",
			Help:      "Payload by number requests",
		}, []string{
			"p2p_role", // "client" or "server"
		}),
		PayloadsQuarantineTotal: factory.NewGauge(prometheus.GaugeOpts{
			Namespace: ns,
			Subsystem: "p2p",
			Name:      "payloads_quarantine_total",
			Help:      "number of unverified execution payloads buffered in quarantine",
		}),

415 416 417 418 419 420 421 422
		L1RequestDurationSeconds: factory.NewHistogramVec(prometheus.HistogramOpts{
			Namespace: ns,
			Name:      "l1_request_seconds",
			Buckets: []float64{
				.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10},
			Help: "Histogram of L1 request time",
		}, []string{"request"}),

423
		SequencerBuildingDiffDurationSeconds: factory.NewHistogram(prometheus.HistogramOpts{
424 425 426 427 428 429 430
			Namespace: ns,
			Name:      "sequencer_building_diff_seconds",
			Buckets: []float64{
				-10, -5, -2.5, -1, -.5, -.25, -.1, -0.05, -0.025, -0.01, -0.005,
				.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10},
			Help: "Histogram of Sequencer building time, minus block time",
		}),
431
		SequencerBuildingDiffTotal: factory.NewCounter(prometheus.CounterOpts{
432 433 434 435
			Namespace: ns,
			Name:      "sequencer_building_diff_total",
			Help:      "Number of sequencer block building jobs",
		}),
436
		SequencerSealingDurationSeconds: factory.NewHistogram(prometheus.HistogramOpts{
437 438 439 440 441
			Namespace: ns,
			Name:      "sequencer_sealing_seconds",
			Buckets:   []float64{.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10},
			Help:      "Histogram of Sequencer block sealing time",
		}),
442
		SequencerSealingTotal: factory.NewCounter(prometheus.CounterOpts{
443 444 445 446 447
			Namespace: ns,
			Name:      "sequencer_sealing_total",
			Help:      "Number of sequencer block sealing jobs",
		}),

448
		registry: registry,
449
		factory:  factory,
450 451 452
	}
}

453 454 455 456 457 458 459 460
// SetPeerScores updates the peer score [prometheus.GaugeVec].
// This takes a map of labels to scores.
func (m *Metrics) SetPeerScores(scores map[string]float64) {
	for label, score := range scores {
		m.PeerScores.WithLabelValues(label).Set(score)
	}
}

461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519
// RecordInfo sets a pseudo-metric that contains versioning and
// config info for the opnode.
func (m *Metrics) RecordInfo(version string) {
	m.Info.WithLabelValues(version).Set(1)
}

// RecordUp sets the up metric to 1.
func (m *Metrics) RecordUp() {
	prometheus.MustRegister()
	m.Up.Set(1)
}

// RecordRPCServerRequest is a helper method to record an incoming RPC
// call to the opnode's RPC server. It bumps the requests metric,
// and tracks how long it takes to serve a response.
func (m *Metrics) RecordRPCServerRequest(method string) func() {
	m.RPCServerRequestsTotal.WithLabelValues(method).Inc()
	timer := prometheus.NewTimer(m.RPCServerRequestDurationSeconds.WithLabelValues(method))
	return func() {
		timer.ObserveDuration()
	}
}

// RecordRPCClientRequest is a helper method to record an RPC client
// request. It bumps the requests metric, tracks the response
// duration, and records the response's error code.
func (m *Metrics) RecordRPCClientRequest(method string) func(err error) {
	m.RPCClientRequestsTotal.WithLabelValues(method).Inc()
	timer := prometheus.NewTimer(m.RPCClientRequestDurationSeconds.WithLabelValues(method))
	return func(err error) {
		m.RecordRPCClientResponse(method, err)
		timer.ObserveDuration()
	}
}

// RecordRPCClientResponse records an RPC response. It will
// convert the passed-in error into something metrics friendly.
// Nil errors get converted into <nil>, RPC errors are converted
// into rpc_<error code>, HTTP errors are converted into
// http_<status code>, and everything else is converted into
// <unknown>.
func (m *Metrics) RecordRPCClientResponse(method string, err error) {
	var errStr string
	var rpcErr rpc.Error
	var httpErr rpc.HTTPError
	if err == nil {
		errStr = "<nil>"
	} else if errors.As(err, &rpcErr) {
		errStr = fmt.Sprintf("rpc_%d", rpcErr.ErrorCode())
	} else if errors.As(err, &httpErr) {
		errStr = fmt.Sprintf("http_%d", httpErr.StatusCode)
	} else if errors.Is(err, ethereum.NotFound) {
		errStr = "<not found>"
	} else {
		errStr = "<unknown>"
	}
	m.RPCClientResponsesTotal.WithLabelValues(method, errStr).Inc()
}

520 521 522 523 524 525 526 527
func (m *Metrics) SetDerivationIdle(status bool) {
	var val float64
	if status {
		val = 1
	}
	m.DerivationIdle.Set(val)
}

528 529
func (m *Metrics) RecordPipelineReset() {
	m.PipelineResets.RecordEvent()
530 531
}

532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573
func (m *Metrics) RecordSequencingError() {
	m.SequencingErrors.RecordEvent()
}

func (m *Metrics) RecordPublishingError() {
	m.PublishingErrors.RecordEvent()
}

func (m *Metrics) RecordDerivationError() {
	m.DerivationErrors.RecordEvent()
}

func (m *Metrics) RecordReceivedUnsafePayload(payload *eth.ExecutionPayload) {
	m.UnsafePayloads.RecordEvent()
	m.recordRef("l2", "received_payload", uint64(payload.BlockNumber), uint64(payload.Timestamp), payload.BlockHash)
}

func (m *Metrics) recordRef(layer string, name string, num uint64, timestamp uint64, h common.Hash) {
	m.RefsNumber.WithLabelValues(layer, name).Set(float64(num))
	if timestamp != 0 {
		m.RefsTime.WithLabelValues(layer, name).Set(float64(timestamp))
		// only meter the latency when we first see this hash for the given label name
		if m.LatencySeen[name] != h {
			m.LatencySeen[name] = h
			m.RefsLatency.WithLabelValues(layer, name).Set(float64(timestamp) - (float64(time.Now().UnixNano()) / 1e9))
		}
	}
	// we map the first 8 bytes to a float64, so we can graph changes of the hash to find divergences visually.
	// We don't do math.Float64frombits, just a regular conversion, to keep the value within a manageable range.
	m.RefsHash.WithLabelValues(layer, name).Set(float64(binary.LittleEndian.Uint64(h[:])))
}

func (m *Metrics) RecordL1Ref(name string, ref eth.L1BlockRef) {
	m.recordRef("l1", name, ref.Number, ref.Time, ref.Hash)
}

func (m *Metrics) RecordL2Ref(name string, ref eth.L2BlockRef) {
	m.recordRef("l2", name, ref.Number, ref.Time, ref.Hash)
	m.recordRef("l1_origin", name, ref.L1Origin.Number, 0, ref.L1Origin.Hash)
	m.RefsSeqNr.WithLabelValues(name).Set(float64(ref.SequenceNumber))
}

574 575 576 577 578 579
func (m *Metrics) RecordUnsafePayloadsBuffer(length uint64, memSize uint64, next eth.BlockID) {
	m.recordRef("l2", "l2_buffer_unsafe", next.Number, 0, next.Hash)
	m.UnsafePayloadsBufferLen.Set(float64(length))
	m.UnsafePayloadsBufferMemSize.Set(float64(memSize))
}

580 581
func (m *Metrics) CountSequencedTxs(count int) {
	m.TransactionsSequencedTotal.Add(float64(count))
582 583
}

584 585 586 587
func (m *Metrics) RecordL1ReorgDepth(d uint64) {
	m.L1ReorgDepth.Observe(float64(d))
}

588 589 590 591 592 593 594 595 596 597
func (m *Metrics) RecordSequencerInconsistentL1Origin(from eth.BlockID, to eth.BlockID) {
	m.SequencerInconsistentL1Origin.RecordEvent()
	m.recordRef("l1_origin", "inconsistent_from", from.Number, 0, from.Hash)
	m.recordRef("l1_origin", "inconsistent_to", to.Number, 0, to.Hash)
}

func (m *Metrics) RecordSequencerReset() {
	m.SequencerResets.RecordEvent()
}

598 599 600 601
func (m *Metrics) RecordGossipEvent(evType int32) {
	m.GossipEventsTotal.WithLabelValues(pb.TraceEvent_Type_name[evType]).Inc()
}

Matthew Slipper's avatar
Matthew Slipper committed
602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618
func (m *Metrics) IncPeerCount() {
	m.PeerCount.Inc()
}

func (m *Metrics) DecPeerCount() {
	m.PeerCount.Dec()
}

func (m *Metrics) IncStreamCount() {
	m.StreamCount.Inc()
}

func (m *Metrics) DecStreamCount() {
	m.StreamCount.Dec()
}

func (m *Metrics) RecordBandwidth(ctx context.Context, bwc *libp2pmetrics.BandwidthCounter) {
619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
	tick := time.NewTicker(10 * time.Second)
	defer tick.Stop()

	for {
		select {
		case <-tick.C:
			bwTotals := bwc.GetBandwidthTotals()
			m.BandwidthTotal.WithLabelValues("in").Set(float64(bwTotals.TotalIn))
			m.BandwidthTotal.WithLabelValues("out").Set(float64(bwTotals.TotalOut))
		case <-ctx.Done():
			return
		}
	}
}

634 635 636 637 638
// RecordL1RequestTime tracks the amount of time the derivation pipeline spent waiting for L1 data requests.
func (m *Metrics) RecordL1RequestTime(method string, duration time.Duration) {
	m.L1RequestDurationSeconds.WithLabelValues(method).Observe(float64(duration) / float64(time.Second))
}

639 640 641 642 643 644 645 646 647 648 649 650 651 652 653
// RecordSequencerBuildingDiffTime tracks the amount of time the sequencer was allowed between
// start to finish, incl. sealing, minus the block time.
// Ideally this is 0, realistically the sequencer scheduler may be busy with other jobs like syncing sometimes.
func (m *Metrics) RecordSequencerBuildingDiffTime(duration time.Duration) {
	m.SequencerBuildingDiffTotal.Inc()
	m.SequencerBuildingDiffDurationSeconds.Observe(float64(duration) / float64(time.Second))
}

// RecordSequencerSealingTime tracks the amount of time the sequencer took to finish sealing the block.
// Ideally this is 0, realistically it may take some time.
func (m *Metrics) RecordSequencerSealingTime(duration time.Duration) {
	m.SequencerSealingTotal.Inc()
	m.SequencerSealingDurationSeconds.Observe(float64(duration) / float64(time.Second))
}

654 655 656 657
// Serve starts the metrics server on the given hostname and port.
// The server will be closed when the passed-in context is cancelled.
func (m *Metrics) Serve(ctx context.Context, hostname string, port int) error {
	addr := net.JoinHostPort(hostname, strconv.Itoa(port))
658
	server := ophttp.NewHttpServer(promhttp.InstrumentMetricHandler(
659 660 661
		m.registry, promhttp.HandlerFor(m.registry, promhttp.HandlerOpts{}),
	))
	server.Addr = addr
662 663 664 665 666 667
	go func() {
		<-ctx.Done()
		server.Close()
	}()
	return server.ListenAndServe()
}
668

669 670 671 672
func (m *Metrics) Document() []metrics.DocumentedMetric {
	return m.factory.Document()
}

673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
func (m *Metrics) ClientPayloadByNumberEvent(num uint64, resultCode byte, duration time.Duration) {
	if resultCode > 4 { // summarize all high codes to reduce metrics overhead
		resultCode = 5
	}
	code := strconv.FormatUint(uint64(resultCode), 10)
	m.P2PReqTotal.WithLabelValues("client", "payload_by_number", code).Inc()
	m.P2PReqDurationSeconds.WithLabelValues("client", "payload_by_number", code).Observe(float64(duration) / float64(time.Second))
	m.P2PPayloadByNumber.WithLabelValues("client").Set(float64(num))
}

func (m *Metrics) ServerPayloadByNumberEvent(num uint64, resultCode byte, duration time.Duration) {
	code := strconv.FormatUint(uint64(resultCode), 10)
	m.P2PReqTotal.WithLabelValues("server", "payload_by_number", code).Inc()
	m.P2PReqDurationSeconds.WithLabelValues("server", "payload_by_number", code).Observe(float64(duration) / float64(time.Second))
	m.P2PPayloadByNumber.WithLabelValues("server").Set(float64(num))
}

func (m *Metrics) PayloadsQuarantineSize(n int) {
	m.PayloadsQuarantineTotal.Set(float64(n))
}

694
func (m *Metrics) RecordChannelInputBytes(inputCompressedBytes int) {
695
	m.ChannelInputBytes.Add(float64(inputCompressedBytes))
696 697
}

698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
func (m *Metrics) RecordPeerUnban() {
	m.PeerUnbans.Inc()
}

func (m *Metrics) RecordIPUnban() {
	m.IPUnbans.Inc()
}

func (m *Metrics) RecordDial(allow bool) {
	if allow {
		m.Dials.WithLabelValues("true").Inc()
	} else {
		m.Dials.WithLabelValues("false").Inc()
	}
}

func (m *Metrics) RecordAccept(allow bool) {
	if allow {
		m.Accepts.WithLabelValues("true").Inc()
	} else {
		m.Accepts.WithLabelValues("false").Inc()
	}
}

722 723
type noopMetricer struct{}

724
var NoopMetrics Metricer = new(noopMetricer)
725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778

func (n *noopMetricer) RecordInfo(version string) {
}

func (n *noopMetricer) RecordUp() {
}

func (n *noopMetricer) RecordRPCServerRequest(method string) func() {
	return func() {}
}

func (n *noopMetricer) RecordRPCClientRequest(method string) func(err error) {
	return func(err error) {}
}

func (n *noopMetricer) RecordRPCClientResponse(method string, err error) {
}

func (n *noopMetricer) SetDerivationIdle(status bool) {
}

func (n *noopMetricer) RecordPipelineReset() {
}

func (n *noopMetricer) RecordSequencingError() {
}

func (n *noopMetricer) RecordPublishingError() {
}

func (n *noopMetricer) RecordDerivationError() {
}

func (n *noopMetricer) RecordReceivedUnsafePayload(payload *eth.ExecutionPayload) {
}

func (n *noopMetricer) recordRef(layer string, name string, num uint64, timestamp uint64, h common.Hash) {
}

func (n *noopMetricer) RecordL1Ref(name string, ref eth.L1BlockRef) {
}

func (n *noopMetricer) RecordL2Ref(name string, ref eth.L2BlockRef) {
}

func (n *noopMetricer) RecordUnsafePayloadsBuffer(length uint64, memSize uint64, next eth.BlockID) {
}

func (n *noopMetricer) CountSequencedTxs(count int) {
}

func (n *noopMetricer) RecordL1ReorgDepth(d uint64) {
}

779 780 781 782 783 784
func (n *noopMetricer) RecordSequencerInconsistentL1Origin(from eth.BlockID, to eth.BlockID) {
}

func (n *noopMetricer) RecordSequencerReset() {
}

785 786 787
func (n *noopMetricer) RecordGossipEvent(evType int32) {
}

788
func (n *noopMetricer) SetPeerScores(scores map[string]float64) {
789 790
}

791 792 793 794 795 796 797 798 799 800 801 802 803 804
func (n *noopMetricer) IncPeerCount() {
}

func (n *noopMetricer) DecPeerCount() {
}

func (n *noopMetricer) IncStreamCount() {
}

func (n *noopMetricer) DecStreamCount() {
}

func (n *noopMetricer) RecordBandwidth(ctx context.Context, bwc *libp2pmetrics.BandwidthCounter) {
}
805 806 807 808 809 810

func (n *noopMetricer) RecordSequencerBuildingDiffTime(duration time.Duration) {
}

func (n *noopMetricer) RecordSequencerSealingTime(duration time.Duration) {
}
811 812 813 814

func (n *noopMetricer) Document() []metrics.DocumentedMetric {
	return nil
}
815 816 817 818 819 820 821 822 823

func (n *noopMetricer) ClientPayloadByNumberEvent(num uint64, resultCode byte, duration time.Duration) {
}

func (n *noopMetricer) ServerPayloadByNumberEvent(num uint64, resultCode byte, duration time.Duration) {
}

func (n *noopMetricer) PayloadsQuarantineSize(int) {
}
824 825 826

func (n *noopMetricer) RecordChannelInputBytes(int) {
}
827 828 829 830 831 832 833 834 835 836 837 838

func (n *noopMetricer) RecordPeerUnban() {
}

func (n *noopMetricer) RecordIPUnban() {
}

func (n *noopMetricer) RecordDial(allow bool) {
}

func (n *noopMetricer) RecordAccept(allow bool) {
}