go-ethereum/ethstats/ethstats.go
Bin 0bbd88bda0
all: use timer instead of time.After in loops, to avoid memleaks (#29241)
time.After is equivalent to NewTimer(d).C, and does not call Stop if the timer is no longer needed. This can cause memory leaks. This change changes many such occations to use NewTimer instead, and calling Stop once the timer is no longer needed.
2024-04-09 08:51:54 +02:00

820 lines
24 KiB
Go

// Copyright 2016 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
// Package ethstats implements the network stats reporting service.
package ethstats
import (
"context"
"encoding/json"
"errors"
"fmt"
"math/big"
"net/http"
"runtime"
"strconv"
"strings"
"sync"
"time"
"github.com/ethereum/go-ethereum"
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/common/mclock"
"github.com/ethereum/go-ethereum/consensus"
"github.com/ethereum/go-ethereum/core"
"github.com/ethereum/go-ethereum/core/types"
ethproto "github.com/ethereum/go-ethereum/eth/protocols/eth"
"github.com/ethereum/go-ethereum/event"
"github.com/ethereum/go-ethereum/log"
"github.com/ethereum/go-ethereum/node"
"github.com/ethereum/go-ethereum/p2p"
"github.com/ethereum/go-ethereum/rpc"
"github.com/gorilla/websocket"
)
const (
// historyUpdateRange is the number of blocks a node should report upon login or
// history request.
historyUpdateRange = 50
// txChanSize is the size of channel listening to NewTxsEvent.
// The number is referenced from the size of tx pool.
txChanSize = 4096
// chainHeadChanSize is the size of channel listening to ChainHeadEvent.
chainHeadChanSize = 10
messageSizeLimit = 15 * 1024 * 1024
)
// backend encompasses the bare-minimum functionality needed for ethstats reporting
type backend interface {
SubscribeChainHeadEvent(ch chan<- core.ChainHeadEvent) event.Subscription
SubscribeNewTxsEvent(ch chan<- core.NewTxsEvent) event.Subscription
CurrentHeader() *types.Header
HeaderByNumber(ctx context.Context, number rpc.BlockNumber) (*types.Header, error)
GetTd(ctx context.Context, hash common.Hash) *big.Int
Stats() (pending int, queued int)
SyncProgress() ethereum.SyncProgress
}
// fullNodeBackend encompasses the functionality necessary for a full node
// reporting to ethstats
type fullNodeBackend interface {
backend
BlockByNumber(ctx context.Context, number rpc.BlockNumber) (*types.Block, error)
CurrentBlock() *types.Header
SuggestGasTipCap(ctx context.Context) (*big.Int, error)
}
// Service implements an Ethereum netstats reporting daemon that pushes local
// chain statistics up to a monitoring server.
type Service struct {
server *p2p.Server // Peer-to-peer server to retrieve networking infos
backend backend
engine consensus.Engine // Consensus engine to retrieve variadic block fields
node string // Name of the node to display on the monitoring page
pass string // Password to authorize access to the monitoring page
host string // Remote address of the monitoring service
pongCh chan struct{} // Pong notifications are fed into this channel
histCh chan []uint64 // History request block numbers are fed into this channel
headSub event.Subscription
txSub event.Subscription
}
// connWrapper is a wrapper to prevent concurrent-write or concurrent-read on the
// websocket.
//
// From Gorilla websocket docs:
//
// Connections support one concurrent reader and one concurrent writer. Applications are
// responsible for ensuring that
// - no more than one goroutine calls the write methods
// NextWriter, SetWriteDeadline, WriteMessage, WriteJSON, EnableWriteCompression,
// SetCompressionLevel concurrently; and
// - that no more than one goroutine calls the
// read methods NextReader, SetReadDeadline, ReadMessage, ReadJSON, SetPongHandler,
// SetPingHandler concurrently.
//
// The Close and WriteControl methods can be called concurrently with all other methods.
type connWrapper struct {
conn *websocket.Conn
rlock sync.Mutex
wlock sync.Mutex
}
func newConnectionWrapper(conn *websocket.Conn) *connWrapper {
conn.SetReadLimit(messageSizeLimit)
return &connWrapper{conn: conn}
}
// WriteJSON wraps corresponding method on the websocket but is safe for concurrent calling
func (w *connWrapper) WriteJSON(v interface{}) error {
w.wlock.Lock()
defer w.wlock.Unlock()
return w.conn.WriteJSON(v)
}
// ReadJSON wraps corresponding method on the websocket but is safe for concurrent calling
func (w *connWrapper) ReadJSON(v interface{}) error {
w.rlock.Lock()
defer w.rlock.Unlock()
return w.conn.ReadJSON(v)
}
// Close wraps corresponding method on the websocket but is safe for concurrent calling
func (w *connWrapper) Close() error {
// The Close and WriteControl methods can be called concurrently with all other methods,
// so the mutex is not used here
return w.conn.Close()
}
// parseEthstatsURL parses the netstats connection url.
// URL argument should be of the form <nodename:secret@host:port>
// If non-erroring, the returned slice contains 3 elements: [nodename, pass, host]
func parseEthstatsURL(url string) (parts []string, err error) {
err = fmt.Errorf("invalid netstats url: \"%s\", should be nodename:secret@host:port", url)
hostIndex := strings.LastIndex(url, "@")
if hostIndex == -1 || hostIndex == len(url)-1 {
return nil, err
}
preHost, host := url[:hostIndex], url[hostIndex+1:]
passIndex := strings.LastIndex(preHost, ":")
if passIndex == -1 {
return []string{preHost, "", host}, nil
}
nodename, pass := preHost[:passIndex], ""
if passIndex != len(preHost)-1 {
pass = preHost[passIndex+1:]
}
return []string{nodename, pass, host}, nil
}
// New returns a monitoring service ready for stats reporting.
func New(node *node.Node, backend backend, engine consensus.Engine, url string) error {
parts, err := parseEthstatsURL(url)
if err != nil {
return err
}
ethstats := &Service{
backend: backend,
engine: engine,
server: node.Server(),
node: parts[0],
pass: parts[1],
host: parts[2],
pongCh: make(chan struct{}),
histCh: make(chan []uint64, 1),
}
node.RegisterLifecycle(ethstats)
return nil
}
// Start implements node.Lifecycle, starting up the monitoring and reporting daemon.
func (s *Service) Start() error {
// Subscribe to chain events to execute updates on
chainHeadCh := make(chan core.ChainHeadEvent, chainHeadChanSize)
s.headSub = s.backend.SubscribeChainHeadEvent(chainHeadCh)
txEventCh := make(chan core.NewTxsEvent, txChanSize)
s.txSub = s.backend.SubscribeNewTxsEvent(txEventCh)
go s.loop(chainHeadCh, txEventCh)
log.Info("Stats daemon started")
return nil
}
// Stop implements node.Lifecycle, terminating the monitoring and reporting daemon.
func (s *Service) Stop() error {
s.headSub.Unsubscribe()
s.txSub.Unsubscribe()
log.Info("Stats daemon stopped")
return nil
}
// loop keeps trying to connect to the netstats server, reporting chain events
// until termination.
func (s *Service) loop(chainHeadCh chan core.ChainHeadEvent, txEventCh chan core.NewTxsEvent) {
// Start a goroutine that exhausts the subscriptions to avoid events piling up
var (
quitCh = make(chan struct{})
headCh = make(chan *types.Block, 1)
txCh = make(chan struct{}, 1)
)
go func() {
var lastTx mclock.AbsTime
HandleLoop:
for {
select {
// Notify of chain head events, but drop if too frequent
case head := <-chainHeadCh:
select {
case headCh <- head.Block:
default:
}
// Notify of new transaction events, but drop if too frequent
case <-txEventCh:
if time.Duration(mclock.Now()-lastTx) < time.Second {
continue
}
lastTx = mclock.Now()
select {
case txCh <- struct{}{}:
default:
}
// node stopped
case <-s.txSub.Err():
break HandleLoop
case <-s.headSub.Err():
break HandleLoop
}
}
close(quitCh)
}()
// Resolve the URL, defaulting to TLS, but falling back to none too
path := fmt.Sprintf("%s/api", s.host)
urls := []string{path}
// url.Parse and url.IsAbs is unsuitable (https://github.com/golang/go/issues/19779)
if !strings.Contains(path, "://") {
urls = []string{"wss://" + path, "ws://" + path}
}
errTimer := time.NewTimer(0)
defer errTimer.Stop()
// Loop reporting until termination
for {
select {
case <-quitCh:
return
case <-errTimer.C:
// Establish a websocket connection to the server on any supported URL
var (
conn *connWrapper
err error
)
dialer := websocket.Dialer{HandshakeTimeout: 5 * time.Second}
header := make(http.Header)
header.Set("origin", "http://localhost")
for _, url := range urls {
c, _, e := dialer.Dial(url, header)
err = e
if err == nil {
conn = newConnectionWrapper(c)
break
}
}
if err != nil {
log.Warn("Stats server unreachable", "err", err)
errTimer.Reset(10 * time.Second)
continue
}
// Authenticate the client with the server
if err = s.login(conn); err != nil {
log.Warn("Stats login failed", "err", err)
conn.Close()
errTimer.Reset(10 * time.Second)
continue
}
go s.readLoop(conn)
// Send the initial stats so our node looks decent from the get go
if err = s.report(conn); err != nil {
log.Warn("Initial stats report failed", "err", err)
conn.Close()
errTimer.Reset(0)
continue
}
// Keep sending status updates until the connection breaks
fullReport := time.NewTicker(15 * time.Second)
for err == nil {
select {
case <-quitCh:
fullReport.Stop()
// Make sure the connection is closed
conn.Close()
return
case <-fullReport.C:
if err = s.report(conn); err != nil {
log.Warn("Full stats report failed", "err", err)
}
case list := <-s.histCh:
if err = s.reportHistory(conn, list); err != nil {
log.Warn("Requested history report failed", "err", err)
}
case head := <-headCh:
if err = s.reportBlock(conn, head); err != nil {
log.Warn("Block stats report failed", "err", err)
}
if err = s.reportPending(conn); err != nil {
log.Warn("Post-block transaction stats report failed", "err", err)
}
case <-txCh:
if err = s.reportPending(conn); err != nil {
log.Warn("Transaction stats report failed", "err", err)
}
}
}
fullReport.Stop()
// Close the current connection and establish a new one
conn.Close()
errTimer.Reset(0)
}
}
}
// readLoop loops as long as the connection is alive and retrieves data packets
// from the network socket. If any of them match an active request, it forwards
// it, if they themselves are requests it initiates a reply, and lastly it drops
// unknown packets.
func (s *Service) readLoop(conn *connWrapper) {
// If the read loop exits, close the connection
defer conn.Close()
for {
// Retrieve the next generic network packet and bail out on error
var blob json.RawMessage
if err := conn.ReadJSON(&blob); err != nil {
log.Warn("Failed to retrieve stats server message", "err", err)
return
}
// If the network packet is a system ping, respond to it directly
var ping string
if err := json.Unmarshal(blob, &ping); err == nil && strings.HasPrefix(ping, "primus::ping::") {
if err := conn.WriteJSON(strings.ReplaceAll(ping, "ping", "pong")); err != nil {
log.Warn("Failed to respond to system ping message", "err", err)
return
}
continue
}
// Not a system ping, try to decode an actual state message
var msg map[string][]interface{}
if err := json.Unmarshal(blob, &msg); err != nil {
log.Warn("Failed to decode stats server message", "err", err)
return
}
log.Trace("Received message from stats server", "msg", msg)
if len(msg["emit"]) == 0 {
log.Warn("Stats server sent non-broadcast", "msg", msg)
return
}
command, ok := msg["emit"][0].(string)
if !ok {
log.Warn("Invalid stats server message type", "type", msg["emit"][0])
return
}
// If the message is a ping reply, deliver (someone must be listening!)
if len(msg["emit"]) == 2 && command == "node-pong" {
select {
case s.pongCh <- struct{}{}:
// Pong delivered, continue listening
continue
default:
// Ping routine dead, abort
log.Warn("Stats server pinger seems to have died")
return
}
}
// If the message is a history request, forward to the event processor
if len(msg["emit"]) == 2 && command == "history" {
// Make sure the request is valid and doesn't crash us
request, ok := msg["emit"][1].(map[string]interface{})
if !ok {
log.Warn("Invalid stats history request", "msg", msg["emit"][1])
select {
case s.histCh <- nil: // Treat it as an no indexes request
default:
}
continue
}
list, ok := request["list"].([]interface{})
if !ok {
log.Warn("Invalid stats history block list", "list", request["list"])
return
}
// Convert the block number list to an integer list
numbers := make([]uint64, len(list))
for i, num := range list {
n, ok := num.(float64)
if !ok {
log.Warn("Invalid stats history block number", "number", num)
return
}
numbers[i] = uint64(n)
}
select {
case s.histCh <- numbers:
continue
default:
}
}
// Report anything else and continue
log.Info("Unknown stats message", "msg", msg)
}
}
// nodeInfo is the collection of meta information about a node that is displayed
// on the monitoring page.
type nodeInfo struct {
Name string `json:"name"`
Node string `json:"node"`
Port int `json:"port"`
Network string `json:"net"`
Protocol string `json:"protocol"`
API string `json:"api"`
Os string `json:"os"`
OsVer string `json:"os_v"`
Client string `json:"client"`
History bool `json:"canUpdateHistory"`
}
// authMsg is the authentication infos needed to login to a monitoring server.
type authMsg struct {
ID string `json:"id"`
Info nodeInfo `json:"info"`
Secret string `json:"secret"`
}
// login tries to authorize the client at the remote server.
func (s *Service) login(conn *connWrapper) error {
// Construct and send the login authentication
infos := s.server.NodeInfo()
var protocols []string
for _, proto := range s.server.Protocols {
protocols = append(protocols, fmt.Sprintf("%s/%d", proto.Name, proto.Version))
}
var network string
if info := infos.Protocols["eth"]; info != nil {
network = fmt.Sprintf("%d", info.(*ethproto.NodeInfo).Network)
} else {
return errors.New("no eth protocol available")
}
auth := &authMsg{
ID: s.node,
Info: nodeInfo{
Name: s.node,
Node: infos.Name,
Port: infos.Ports.Listener,
Network: network,
Protocol: strings.Join(protocols, ", "),
API: "No",
Os: runtime.GOOS,
OsVer: runtime.GOARCH,
Client: "0.1.1",
History: true,
},
Secret: s.pass,
}
login := map[string][]interface{}{
"emit": {"hello", auth},
}
if err := conn.WriteJSON(login); err != nil {
return err
}
// Retrieve the remote ack or connection termination
var ack map[string][]string
if err := conn.ReadJSON(&ack); err != nil || len(ack["emit"]) != 1 || ack["emit"][0] != "ready" {
return errors.New("unauthorized")
}
return nil
}
// report collects all possible data to report and send it to the stats server.
// This should only be used on reconnects or rarely to avoid overloading the
// server. Use the individual methods for reporting subscribed events.
func (s *Service) report(conn *connWrapper) error {
if err := s.reportLatency(conn); err != nil {
return err
}
if err := s.reportBlock(conn, nil); err != nil {
return err
}
if err := s.reportPending(conn); err != nil {
return err
}
if err := s.reportStats(conn); err != nil {
return err
}
return nil
}
// reportLatency sends a ping request to the server, measures the RTT time and
// finally sends a latency update.
func (s *Service) reportLatency(conn *connWrapper) error {
// Send the current time to the ethstats server
start := time.Now()
ping := map[string][]interface{}{
"emit": {"node-ping", map[string]string{
"id": s.node,
"clientTime": start.String(),
}},
}
if err := conn.WriteJSON(ping); err != nil {
return err
}
// Wait for the pong request to arrive back
timer := time.NewTimer(5 * time.Second)
defer timer.Stop()
select {
case <-s.pongCh:
// Pong delivered, report the latency
case <-timer.C:
// Ping timeout, abort
return errors.New("ping timed out")
}
latency := strconv.Itoa(int((time.Since(start) / time.Duration(2)).Nanoseconds() / 1000000))
// Send back the measured latency
log.Trace("Sending measured latency to ethstats", "latency", latency)
stats := map[string][]interface{}{
"emit": {"latency", map[string]string{
"id": s.node,
"latency": latency,
}},
}
return conn.WriteJSON(stats)
}
// blockStats is the information to report about individual blocks.
type blockStats struct {
Number *big.Int `json:"number"`
Hash common.Hash `json:"hash"`
ParentHash common.Hash `json:"parentHash"`
Timestamp *big.Int `json:"timestamp"`
Miner common.Address `json:"miner"`
GasUsed uint64 `json:"gasUsed"`
GasLimit uint64 `json:"gasLimit"`
Diff string `json:"difficulty"`
TotalDiff string `json:"totalDifficulty"`
Txs []txStats `json:"transactions"`
TxHash common.Hash `json:"transactionsRoot"`
Root common.Hash `json:"stateRoot"`
Uncles uncleStats `json:"uncles"`
}
// txStats is the information to report about individual transactions.
type txStats struct {
Hash common.Hash `json:"hash"`
}
// uncleStats is a custom wrapper around an uncle array to force serializing
// empty arrays instead of returning null for them.
type uncleStats []*types.Header
func (s uncleStats) MarshalJSON() ([]byte, error) {
if uncles := ([]*types.Header)(s); len(uncles) > 0 {
return json.Marshal(uncles)
}
return []byte("[]"), nil
}
// reportBlock retrieves the current chain head and reports it to the stats server.
func (s *Service) reportBlock(conn *connWrapper, block *types.Block) error {
// Gather the block details from the header or block chain
details := s.assembleBlockStats(block)
// Short circuit if the block detail is not available.
if details == nil {
return nil
}
// Assemble the block report and send it to the server
log.Trace("Sending new block to ethstats", "number", details.Number, "hash", details.Hash)
stats := map[string]interface{}{
"id": s.node,
"block": details,
}
report := map[string][]interface{}{
"emit": {"block", stats},
}
return conn.WriteJSON(report)
}
// assembleBlockStats retrieves any required metadata to report a single block
// and assembles the block stats. If block is nil, the current head is processed.
func (s *Service) assembleBlockStats(block *types.Block) *blockStats {
// Gather the block infos from the local blockchain
var (
header *types.Header
td *big.Int
txs []txStats
uncles []*types.Header
)
// check if backend is a full node
fullBackend, ok := s.backend.(fullNodeBackend)
if ok {
// Retrieve current chain head if no block is given.
if block == nil {
head := fullBackend.CurrentBlock()
block, _ = fullBackend.BlockByNumber(context.Background(), rpc.BlockNumber(head.Number.Uint64()))
}
// Short circuit if no block is available. It might happen when
// the blockchain is reorging.
if block == nil {
return nil
}
header = block.Header()
td = fullBackend.GetTd(context.Background(), header.Hash())
txs = make([]txStats, len(block.Transactions()))
for i, tx := range block.Transactions() {
txs[i].Hash = tx.Hash()
}
uncles = block.Uncles()
} else {
// Light nodes would need on-demand lookups for transactions/uncles, skip
if block != nil {
header = block.Header()
} else {
header = s.backend.CurrentHeader()
}
td = s.backend.GetTd(context.Background(), header.Hash())
txs = []txStats{}
}
// Assemble and return the block stats
author, _ := s.engine.Author(header)
return &blockStats{
Number: header.Number,
Hash: header.Hash(),
ParentHash: header.ParentHash,
Timestamp: new(big.Int).SetUint64(header.Time),
Miner: author,
GasUsed: header.GasUsed,
GasLimit: header.GasLimit,
Diff: header.Difficulty.String(),
TotalDiff: td.String(),
Txs: txs,
TxHash: header.TxHash,
Root: header.Root,
Uncles: uncles,
}
}
// reportHistory retrieves the most recent batch of blocks and reports it to the
// stats server.
func (s *Service) reportHistory(conn *connWrapper, list []uint64) error {
// Figure out the indexes that need reporting
indexes := make([]uint64, 0, historyUpdateRange)
if len(list) > 0 {
// Specific indexes requested, send them back in particular
indexes = append(indexes, list...)
} else {
// No indexes requested, send back the top ones
head := s.backend.CurrentHeader().Number.Int64()
start := head - historyUpdateRange + 1
if start < 0 {
start = 0
}
for i := uint64(start); i <= uint64(head); i++ {
indexes = append(indexes, i)
}
}
// Gather the batch of blocks to report
history := make([]*blockStats, len(indexes))
for i, number := range indexes {
fullBackend, ok := s.backend.(fullNodeBackend)
// Retrieve the next block if it's known to us
var block *types.Block
if ok {
block, _ = fullBackend.BlockByNumber(context.Background(), rpc.BlockNumber(number)) // TODO ignore error here ?
} else {
if header, _ := s.backend.HeaderByNumber(context.Background(), rpc.BlockNumber(number)); header != nil {
block = types.NewBlockWithHeader(header)
}
}
// If we do have the block, add to the history and continue
if block != nil {
history[len(history)-1-i] = s.assembleBlockStats(block)
continue
}
// Ran out of blocks, cut the report short and send
history = history[len(history)-i:]
break
}
// Assemble the history report and send it to the server
if len(history) > 0 {
log.Trace("Sending historical blocks to ethstats", "first", history[0].Number, "last", history[len(history)-1].Number)
} else {
log.Trace("No history to send to stats server")
}
stats := map[string]interface{}{
"id": s.node,
"history": history,
}
report := map[string][]interface{}{
"emit": {"history", stats},
}
return conn.WriteJSON(report)
}
// pendStats is the information to report about pending transactions.
type pendStats struct {
Pending int `json:"pending"`
}
// reportPending retrieves the current number of pending transactions and reports
// it to the stats server.
func (s *Service) reportPending(conn *connWrapper) error {
// Retrieve the pending count from the local blockchain
pending, _ := s.backend.Stats()
// Assemble the transaction stats and send it to the server
log.Trace("Sending pending transactions to ethstats", "count", pending)
stats := map[string]interface{}{
"id": s.node,
"stats": &pendStats{
Pending: pending,
},
}
report := map[string][]interface{}{
"emit": {"pending", stats},
}
return conn.WriteJSON(report)
}
// nodeStats is the information to report about the local node.
type nodeStats struct {
Active bool `json:"active"`
Syncing bool `json:"syncing"`
Peers int `json:"peers"`
GasPrice int `json:"gasPrice"`
Uptime int `json:"uptime"`
}
// reportStats retrieves various stats about the node at the networking layer
// and reports it to the stats server.
func (s *Service) reportStats(conn *connWrapper) error {
// Gather the syncing infos from the local miner instance
var (
syncing bool
gasprice int
)
// check if backend is a full node
if fullBackend, ok := s.backend.(fullNodeBackend); ok {
sync := fullBackend.SyncProgress()
syncing = !sync.Done()
price, _ := fullBackend.SuggestGasTipCap(context.Background())
gasprice = int(price.Uint64())
if basefee := fullBackend.CurrentHeader().BaseFee; basefee != nil {
gasprice += int(basefee.Uint64())
}
} else {
sync := s.backend.SyncProgress()
syncing = !sync.Done()
}
// Assemble the node stats and send it to the server
log.Trace("Sending node details to ethstats")
stats := map[string]interface{}{
"id": s.node,
"stats": &nodeStats{
Active: true,
Peers: s.server.PeerCount(),
GasPrice: gasprice,
Syncing: syncing,
Uptime: 100,
},
}
report := map[string][]interface{}{
"emit": {"stats", stats},
}
return conn.WriteJSON(report)
}