tendermint/p2p/switch.go

package p2p

import (
	"fmt"
	"math"
	"sync"
	"time"

	"github.com/tendermint/tendermint/config"
	cmn "github.com/tendermint/tendermint/libs/common"
	"github.com/tendermint/tendermint/p2p/conn"
)

const (
	// wait a random amount of time from this interval
	// before dialing peers or reconnecting to help prevent DoS
	dialRandomizerIntervalMilliseconds = 3000

	// repeatedly try to reconnect for a few minutes
	// ie. 5 * 20 = 100s
	reconnectAttempts = 20
	reconnectInterval = 5 * time.Second

	// then move into exponential backoff mode for ~1day
	// ie. 3**10 = 16hrs
	reconnectBackOffAttempts    = 10
	reconnectBackOffBaseSeconds = 3
)

// MConnConfig returns an MConnConfig with fields updated
// from the P2PConfig.
func MConnConfig(cfg *config.P2PConfig) conn.MConnConfig {
	mConfig := conn.DefaultMConnConfig()
	mConfig.FlushThrottle = cfg.FlushThrottleTimeout
	mConfig.SendRate = cfg.SendRate
	mConfig.RecvRate = cfg.RecvRate
	mConfig.MaxPacketMsgPayloadSize = cfg.MaxPacketMsgPayloadSize
	return mConfig
}

//-----------------------------------------------------------------------------

// An AddrBook represents an address book from the pex package, which is used
// to store peer addresses.
type AddrBook interface {
	AddAddress(addr *NetAddress, src *NetAddress) error
	AddOurAddress(*NetAddress)
	OurAddress(*NetAddress) bool
	MarkGood(*NetAddress)
	RemoveAddress(*NetAddress)
	HasAddress(*NetAddress) bool
	Save()
}

// PeerFilterFunc to be implemented by filter hooks after a new Peer has been
// fully setup.
type PeerFilterFunc func(IPeerSet, Peer) error

//-----------------------------------------------------------------------------

// Switch handles peer connections and exposes an API to receive incoming messages
// on `Reactors`.  Each `Reactor` is responsible for handling incoming messages of one
// or more `Channels`.  So while sending outgoing messages is typically performed on the peer,
// incoming messages are received on the reactor.
type Switch struct {
	cmn.BaseService

	config       *config.P2PConfig
	reactors     map[string]Reactor
	chDescs      []*conn.ChannelDescriptor
	reactorsByCh map[byte]Reactor
	peers        *PeerSet
	dialing      *cmn.CMap
	reconnecting *cmn.CMap
	nodeInfo     NodeInfo // our node info
	nodeKey      *NodeKey // our node privkey
	addrBook     AddrBook

	transport Transport

	filterTimeout time.Duration
	peerFilters   []PeerFilterFunc

	rng *cmn.Rand // seed for randomizing dial times and orders

	metrics *Metrics
}

// NetAddress returns the address the switch is listening on.
func (sw *Switch) NetAddress() *NetAddress {
	addr := sw.transport.NetAddress()
	return &addr
}

// SwitchOption sets an optional parameter on the Switch.
type SwitchOption func(*Switch)

// NewSwitch creates a new Switch with the given config.
func NewSwitch(
	cfg *config.P2PConfig,
	transport Transport,
	options ...SwitchOption,
) *Switch {
	sw := &Switch{
		config:        cfg,
		reactors:      make(map[string]Reactor),
		chDescs:       make([]*conn.ChannelDescriptor, 0),
		reactorsByCh:  make(map[byte]Reactor),
		peers:         NewPeerSet(),
		dialing:       cmn.NewCMap(),
		reconnecting:  cmn.NewCMap(),
		metrics:       NopMetrics(),
		transport:     transport,
		filterTimeout: defaultFilterTimeout,
	}

	// Ensure we have a completely undeterministic PRNG.
	sw.rng = cmn.NewRand()

	sw.BaseService = *cmn.NewBaseService(nil, "P2P Switch", sw)

	for _, option := range options {
		option(sw)
	}

	return sw
}

// SwitchFilterTimeout sets the timeout used for peer filters.
func SwitchFilterTimeout(timeout time.Duration) SwitchOption {
	return func(sw *Switch) { sw.filterTimeout = timeout }
}

// SwitchPeerFilters sets the filters for rejection of new peers.
func SwitchPeerFilters(filters ...PeerFilterFunc) SwitchOption {
	return func(sw *Switch) { sw.peerFilters = filters }
}

// WithMetrics sets the metrics.
func WithMetrics(metrics *Metrics) SwitchOption {
	return func(sw *Switch) { sw.metrics = metrics }
}

//---------------------------------------------------------------------
// Switch setup

// AddReactor adds the given reactor to the switch.
// NOTE: Not goroutine safe.
func (sw *Switch) AddReactor(name string, reactor Reactor) Reactor {
	// Validate the reactor.
	// No two reactors can share the same channel.
	reactorChannels := reactor.GetChannels()
	for _, chDesc := range reactorChannels {
		chID := chDesc.ID
		if sw.reactorsByCh[chID] != nil {
			cmn.PanicSanity(fmt.Sprintf("Channel %X has multiple reactors %v & %v", chID, sw.reactorsByCh[chID], reactor))
		}
		sw.chDescs = append(sw.chDescs, chDesc)
		sw.reactorsByCh[chID] = reactor
	}
	sw.reactors[name] = reactor
	reactor.SetSwitch(sw)
	return reactor
}

// Reactors returns a map of reactors registered on the switch.
// NOTE: Not goroutine safe.
func (sw *Switch) Reactors() map[string]Reactor {
	return sw.reactors
}

// Reactor returns the reactor with the given name.
// NOTE: Not goroutine safe.
func (sw *Switch) Reactor(name string) Reactor {
	return sw.reactors[name]
}

// SetNodeInfo sets the switch's NodeInfo for checking compatibility and handshaking with other nodes.
// NOTE: Not goroutine safe.
func (sw *Switch) SetNodeInfo(nodeInfo NodeInfo) {
	sw.nodeInfo = nodeInfo
}

// NodeInfo returns the switch's NodeInfo.
// NOTE: Not goroutine safe.
func (sw *Switch) NodeInfo() NodeInfo {
	return sw.nodeInfo
}

// SetNodeKey sets the switch's private key for authenticated encryption.
// NOTE: Not goroutine safe.
func (sw *Switch) SetNodeKey(nodeKey *NodeKey) {
	sw.nodeKey = nodeKey
}

//---------------------------------------------------------------------
// Service start/stop

// OnStart implements BaseService. It starts all the reactors and peers.
func (sw *Switch) OnStart() error {
	// Start reactors
	for _, reactor := range sw.reactors {
		err := reactor.Start()
		if err != nil {
			return cmn.ErrorWrap(err, "failed to start %v", reactor)
		}
	}

	// Start accepting Peers.
	go sw.acceptRoutine()

	return nil
}

// OnStop implements BaseService. It stops all peers and reactors.
func (sw *Switch) OnStop() {
	// Stop peers
	for _, p := range sw.peers.List() {
		sw.transport.Cleanup(p)
		p.Stop()
		if sw.peers.Remove(p) {
			sw.metrics.Peers.Add(float64(-1))
		}
	}

	// Stop reactors
	sw.Logger.Debug("Switch: Stopping reactors")
	for _, reactor := range sw.reactors {
		reactor.Stop()
	}
}

//---------------------------------------------------------------------
// Peers

// Broadcast runs a go routine for each attempted send, which will block trying
// to send for defaultSendTimeoutSeconds. Returns a channel which receives
// success values for each attempted send (false if times out). Channel will be
// closed once msg bytes are sent to all peers (or time out).
//
// NOTE: Broadcast uses goroutines, so order of broadcast may not be preserved.
func (sw *Switch) Broadcast(chID byte, msgBytes []byte) chan bool {
	sw.Logger.Debug("Broadcast", "channel", chID, "msgBytes", fmt.Sprintf("%X", msgBytes))

	peers := sw.peers.List()
	var wg sync.WaitGroup
	wg.Add(len(peers))
	successChan := make(chan bool, len(peers))

	for _, peer := range peers {
		go func(p Peer) {
			defer wg.Done()
			success := p.Send(chID, msgBytes)
			successChan <- success
		}(peer)
	}

	go func() {
		wg.Wait()
		close(successChan)
	}()

	return successChan
}

// NumPeers returns the count of outbound/inbound and outbound-dialing peers.
func (sw *Switch) NumPeers() (outbound, inbound, dialing int) {
	peers := sw.peers.List()
	for _, peer := range peers {
		if peer.IsOutbound() {
			outbound++
		} else {
			inbound++
		}
	}
	dialing = sw.dialing.Size()
	return
}

// MaxNumOutboundPeers returns a maximum number of outbound peers.
func (sw *Switch) MaxNumOutboundPeers() int {
	return sw.config.MaxNumOutboundPeers
}

// Peers returns the set of peers that are connected to the switch.
func (sw *Switch) Peers() IPeerSet {
	return sw.peers
}

// StopPeerForError disconnects from a peer due to external error.
// If the peer is persistent, it will attempt to reconnect.
// TODO: make record depending on reason.
func (sw *Switch) StopPeerForError(peer Peer, reason interface{}) {
	sw.Logger.Error("Stopping peer for error", "peer", peer, "err", reason)
	sw.stopAndRemovePeer(peer, reason)

	if peer.IsPersistent() {
		go sw.reconnectToPeer(peer.SocketAddr())
	}
}

// StopPeerGracefully disconnects from a peer gracefully.
// TODO: handle graceful disconnects.
func (sw *Switch) StopPeerGracefully(peer Peer) {
	sw.Logger.Info("Stopping peer gracefully")
	sw.stopAndRemovePeer(peer, nil)
}

func (sw *Switch) stopAndRemovePeer(peer Peer, reason interface{}) {
	if sw.peers.Remove(peer) {
		sw.metrics.Peers.Add(float64(-1))
	}
	sw.transport.Cleanup(peer)
	peer.Stop()
	for _, reactor := range sw.reactors {
		reactor.RemovePeer(peer, reason)
	}
}

// reconnectToPeer tries to reconnect to the addr, first repeatedly
// with a fixed interval, then with exponential backoff.
// If no success after all that, it stops trying, and leaves it
// to the PEX/Addrbook to find the peer with the addr again
// NOTE: this will keep trying even if the handshake or auth fails.
// TODO: be more explicit with error types so we only retry on certain failures
//  - ie. if we're getting ErrDuplicatePeer we can stop
//  	because the addrbook got us the peer back already
func (sw *Switch) reconnectToPeer(addr *NetAddress) {
	if sw.reconnecting.Has(string(addr.ID)) {
		return
	}
	sw.reconnecting.Set(string(addr.ID), addr)
	defer sw.reconnecting.Delete(string(addr.ID))

	start := time.Now()
	sw.Logger.Info("Reconnecting to peer", "addr", addr)
	for i := 0; i < reconnectAttempts; i++ {
		if !sw.IsRunning() {
			return
		}

		err := sw.DialPeerWithAddress(addr, true)
		if err == nil {
			return // success
		} else if _, ok := err.(ErrCurrentlyDialingOrExistingAddress); ok {
			return
		}

		sw.Logger.Info("Error reconnecting to peer. Trying again", "tries", i, "err", err, "addr", addr)
		// sleep a set amount
		sw.randomSleep(reconnectInterval)
		continue
	}

	sw.Logger.Error("Failed to reconnect to peer. Beginning exponential backoff",
		"addr", addr, "elapsed", time.Since(start))
	for i := 0; i < reconnectBackOffAttempts; i++ {
		if !sw.IsRunning() {
			return
		}

		// sleep an exponentially increasing amount
		sleepIntervalSeconds := math.Pow(reconnectBackOffBaseSeconds, float64(i))
		sw.randomSleep(time.Duration(sleepIntervalSeconds) * time.Second)

		err := sw.DialPeerWithAddress(addr, true)
		if err == nil {
			return // success
		} else if _, ok := err.(ErrCurrentlyDialingOrExistingAddress); ok {
			return
		}
		sw.Logger.Info("Error reconnecting to peer. Trying again", "tries", i, "err", err, "addr", addr)
	}
	sw.Logger.Error("Failed to reconnect to peer. Giving up", "addr", addr, "elapsed", time.Since(start))
}

// SetAddrBook allows to set address book on Switch.
func (sw *Switch) SetAddrBook(addrBook AddrBook) {
	sw.addrBook = addrBook
}

// MarkPeerAsGood marks the given peer as good when it did something useful
// like contributed to consensus.
func (sw *Switch) MarkPeerAsGood(peer Peer) {
	if sw.addrBook != nil {
		sw.addrBook.MarkGood(peer.SocketAddr())
	}
}

//---------------------------------------------------------------------
// Dialing

// DialPeersAsync dials a list of peers asynchronously in random order (optionally, making them persistent).
// Used to dial peers from config on startup or from unsafe-RPC (trusted sources).
// TODO: remove addrBook arg since it's now set on the switch
func (sw *Switch) DialPeersAsync(addrBook AddrBook, peers []string, persistent bool) error {
	netAddrs, errs := NewNetAddressStrings(peers)
	// only log errors, dial correct addresses
	for _, err := range errs {
		sw.Logger.Error("Error in peer's address", "err", err)
	}

	ourAddr := sw.NetAddress()

	// TODO: this code feels like it's in the wrong place.
	// The integration tests depend on the addrBook being saved
	// right away but maybe we can change that. Recall that
	// the addrBook is only written to disk every 2min
	if addrBook != nil {
		// add peers to `addrBook`
		for _, netAddr := range netAddrs {
			// do not add our address or ID
			if !netAddr.Same(ourAddr) {
				if err := addrBook.AddAddress(netAddr, ourAddr); err != nil {
					sw.Logger.Error("Can't add peer's address to addrbook", "err", err)
				}
			}
		}
		// Persist some peers to disk right away.
		// NOTE: integration tests depend on this
		addrBook.Save()
	}

	// permute the list, dial them in random order.
	perm := sw.rng.Perm(len(netAddrs))
	for i := 0; i < len(perm); i++ {
		go func(i int) {
			j := perm[i]
			addr := netAddrs[j]

			if addr.Same(ourAddr) {
				sw.Logger.Debug("Ignore attempt to connect to ourselves", "addr", addr, "ourAddr", ourAddr)
				return
			}

			sw.randomSleep(0)

			err := sw.DialPeerWithAddress(addr, persistent)
			if err != nil {
				switch err.(type) {
				case ErrSwitchConnectToSelf, ErrSwitchDuplicatePeerID, ErrCurrentlyDialingOrExistingAddress:
					sw.Logger.Debug("Error dialing peer", "err", err)
				default:
					sw.Logger.Error("Error dialing peer", "err", err)
				}
			}
		}(i)
	}
	return nil
}

// DialPeerWithAddress dials the given peer and runs sw.addPeer if it connects
// and authenticates successfully.
// If `persistent == true`, the switch will always try to reconnect to this
// peer if the connection ever fails.
// If we're currently dialing this address or it belongs to an existing peer,
// ErrCurrentlyDialingOrExistingAddress is returned.
func (sw *Switch) DialPeerWithAddress(addr *NetAddress, persistent bool) error {
	if sw.IsDialingOrExistingAddress(addr) {
		return ErrCurrentlyDialingOrExistingAddress{addr.String()}
	}

	sw.dialing.Set(string(addr.ID), addr)
	defer sw.dialing.Delete(string(addr.ID))

	return sw.addOutboundPeerWithConfig(addr, sw.config, persistent)
}

// sleep for interval plus some random amount of ms on [0, dialRandomizerIntervalMilliseconds]
func (sw *Switch) randomSleep(interval time.Duration) {
	r := time.Duration(sw.rng.Int63n(dialRandomizerIntervalMilliseconds)) * time.Millisecond
	time.Sleep(r + interval)
}

// IsDialingOrExistingAddress returns true if switch has a peer with the given
// address or dialing it at the moment.
func (sw *Switch) IsDialingOrExistingAddress(addr *NetAddress) bool {
	return sw.dialing.Has(string(addr.ID)) ||
		sw.peers.Has(addr.ID) ||
		(!sw.config.AllowDuplicateIP && sw.peers.HasIP(addr.IP))
}

func (sw *Switch) acceptRoutine() {
	for {
		p, err := sw.transport.Accept(peerConfig{
			chDescs:      sw.chDescs,
			onPeerError:  sw.StopPeerForError,
			reactorsByCh: sw.reactorsByCh,
			metrics:      sw.metrics,
		})
		if err != nil {
			switch err := err.(type) {
			case ErrRejected:
				if err.IsSelf() {
					// Remove the given address from the address book and add to our addresses
					// to avoid dialing in the future.
					addr := err.Addr()
					sw.addrBook.RemoveAddress(&addr)
					sw.addrBook.AddOurAddress(&addr)
				}

				sw.Logger.Info(
					"Inbound Peer rejected",
					"err", err,
					"numPeers", sw.peers.Size(),
				)

				continue
			case ErrFilterTimeout:
				sw.Logger.Error(
					"Peer filter timed out",
					"err", err,
				)

				continue
			case ErrTransportClosed:
				sw.Logger.Error(
					"Stopped accept routine, as transport is closed",
					"numPeers", sw.peers.Size(),
				)
			default:
				sw.Logger.Error(
					"Accept on transport errored",
					"err", err,
					"numPeers", sw.peers.Size(),
				)
				// We could instead have a retry loop around the acceptRoutine,
				// but that would need to stop and let the node shutdown eventually.
				// So might as well panic and let process managers restart the node.
				// There's no point in letting the node run without the acceptRoutine,
				// since it won't be able to accept new connections.
				panic(fmt.Errorf("accept routine exited: %v", err))
			}

			break
		}

		// Ignore connection if we already have enough peers.
		_, in, _ := sw.NumPeers()
		if in >= sw.config.MaxNumInboundPeers {
			sw.Logger.Info(
				"Ignoring inbound connection: already have enough inbound peers",
				"address", p.SocketAddr(),
				"have", in,
				"max", sw.config.MaxNumInboundPeers,
			)

			sw.transport.Cleanup(p)

			continue
		}

		if err := sw.addPeer(p); err != nil {
			sw.transport.Cleanup(p)
			if p.IsRunning() {
				_ = p.Stop()
			}
			sw.Logger.Info(
				"Ignoring inbound connection: error while adding peer",
				"err", err,
				"id", p.ID(),
			)
		}
	}
}

// dial the peer; make secret connection; authenticate against the dialed ID;
// add the peer.
// if dialing fails, start the reconnect loop. If handhsake fails, its over.
// If peer is started succesffuly, reconnectLoop will start when
// StopPeerForError is called
func (sw *Switch) addOutboundPeerWithConfig(
	addr *NetAddress,
	cfg *config.P2PConfig,
	persistent bool,
) error {
	sw.Logger.Info("Dialing peer", "address", addr)

	// XXX(xla): Remove the leakage of test concerns in implementation.
	if cfg.TestDialFail {
		go sw.reconnectToPeer(addr)
		return fmt.Errorf("dial err (peerConfig.DialFail == true)")
	}

	p, err := sw.transport.Dial(*addr, peerConfig{
		chDescs:      sw.chDescs,
		onPeerError:  sw.StopPeerForError,
		persistent:   persistent,
		reactorsByCh: sw.reactorsByCh,
		metrics:      sw.metrics,
	})
	if err != nil {
		switch e := err.(type) {
		case ErrRejected:
			if e.IsSelf() {
				// Remove the given address from the address book and add to our addresses
				// to avoid dialing in the future.
				sw.addrBook.RemoveAddress(addr)
				sw.addrBook.AddOurAddress(addr)

				return err
			}
		}

		// retry persistent peers after
		// any dial error besides IsSelf()
		if persistent {
			go sw.reconnectToPeer(addr)
		}

		return err
	}

	if err := sw.addPeer(p); err != nil {
		sw.transport.Cleanup(p)
		if p.IsRunning() {
			_ = p.Stop()
		}
		return err
	}

	return nil
}

func (sw *Switch) filterPeer(p Peer) error {
	// Avoid duplicate
	if sw.peers.Has(p.ID()) {
		return ErrRejected{id: p.ID(), isDuplicate: true}
	}

	errc := make(chan error, len(sw.peerFilters))

	for _, f := range sw.peerFilters {
		go func(f PeerFilterFunc, p Peer, errc chan<- error) {
			errc <- f(sw.peers, p)
		}(f, p, errc)
	}

	for i := 0; i < cap(errc); i++ {
		select {
		case err := <-errc:
			if err != nil {
				return ErrRejected{id: p.ID(), err: err, isFiltered: true}
			}
		case <-time.After(sw.filterTimeout):
			return ErrFilterTimeout{}
		}
	}

	return nil
}

// addPeer starts up the Peer and adds it to the Switch. Error is returned if
// the peer is filtered out or failed to start or can't be added.
func (sw *Switch) addPeer(p Peer) error {
	if err := sw.filterPeer(p); err != nil {
		return err
	}

	p.SetLogger(sw.Logger.With("peer", p.SocketAddr()))

	// Handle the shut down case where the switch has stopped but we're
	// concurrently trying to add a peer.
	if !sw.IsRunning() {
		// XXX should this return an error or just log and terminate?
		sw.Logger.Error("Won't start a peer - switch is not running", "peer", p)
		return nil
	}

	// Start the peer's send/recv routines.
	// Must start it before adding it to the peer set
	// to prevent Start and Stop from being called concurrently.
	err := p.Start()
	if err != nil {
		// Should never happen
		sw.Logger.Error("Error starting peer", "err", err, "peer", p)
		return err
	}

	// Add the peer to PeerSet. Do this before starting the reactors
	// so that if Receive errors, we will find the peer and remove it.
	// Add should not err since we already checked peers.Has().
	if err := sw.peers.Add(p); err != nil {
		return err
	}
	sw.metrics.Peers.Add(float64(1))

	// Start all the reactor protocols on the peer.
	for _, reactor := range sw.reactors {
		reactor.AddPeer(p)
	}

	sw.Logger.Info("Added peer", "peer", p)

	return nil
}