Compare commits

...

13 Commits

Author SHA1 Message Date
Zoltan Papp
753925032a [client] Move metrics saver into its own metricsstages package
Relocate MetricsStages (per-connection metric stage timestamps) out of
the peer package into client/internal/peer/metricsstages.
2026-07-11 20:15:30 +02:00
Zoltan Papp
9be5f238da [client] Trim WGIface to the methods peer uses and inline it into conn.go
Drop the unused Address() method from WGIface (only ice.Candidate.Address
was ever called, never the wg interface), and move the interface next to
its sole user WgConfig in conn.go, removing iface.go.
2026-07-11 20:09:59 +02:00
Zoltan Papp
791e8b33ae [client] Move ICE and relay workers into the worker package
Relocate WorkerICE (renamed worker.ICE) and WorkerRelay plus ConnPriority
into client/internal/peer/worker. To break the peer<->worker cycle the
workers no longer take *Conn or ConnConfig: callbacks are passed as plain
functions (Conn's unexported methods as method values), and each worker
receives only the fields it needs (key, ICE config, isController) plus a
small services struct. Context is passed to OnNewOffer instead of stored.

Move the worker connection-status helper the other way, out of the worker
package into peer as worker_status.go (WorkerStatus / AtomicWorkerStatus),
since only Conn uses it.
2026-07-11 20:06:46 +02:00
Zoltan Papp
917e85f648 [client] Extract Handshaker and Signaler into a signaling package
Move the signaling protocol out of the peer package into
client/internal/peer/signaling: OfferAnswer, IceCredentials, Handshaker
and Signaler. Break the peer<->signaling cycle by giving the Handshaker
a plain Config and an ICEWorker interface instead of *Conn/*WorkerICE,
and by passing the relay manager directly rather than the relay worker.
Combine the ICE worker's local-credentials and session-id accessors into
a single Credentials() returning a Credentials struct.

Move the ICE session id to the ice package as ice.SessionID, since it
identifies an ICE agent session and is minted there alongside
GenerateICECredentials; signaling only carries it.
2026-07-11 19:09:07 +02:00
Zoltan Papp
9e75d5c732 [client] Decouple ICE and relay workers from Conn via interfaces
WorkerICE and WorkerRelay took a concrete *Conn back-pointer. Replace it
with small callback interfaces (iceCallbacks, relayCallbacks) covering
only the methods each worker invokes on the connection. WorkerICE also
receives its portForwardManager as an explicit parameter instead of
reaching into conn.portForwardManager.
2026-07-11 18:29:26 +02:00
Zoltan Papp
1afc9bcac7 [client] Merge conntype package into peer
The conntype package held only ConnPriority and its constants and was
imported solely by peer. Move it into the peer package as priority.go
and drop the conntype. qualifier from conn.go, event.go and worker_ice.go.
2026-07-11 15:24:54 +02:00
Zoltan Papp
78f3165e85 [client] Extract WGWatcher into its own wg_watcher package
Move the WireGuard handshake watcher out of the peer package into
client/internal/peer/wg_watcher. The test stays an internal test since
it drives the unexported checkPeriod. Update conn.go to reference the
watcher through the package.
2026-07-11 15:17:16 +02:00
Zoltan Papp
4d76fd3c80 [client] Extract stateDump into its own state_dump package
Move the debug state dumper out of the peer package into
client/internal/peer/state_dump with an exported StateDump type and
NewStateDump constructor. Update conn.go, wg_watcher.go and the watcher
test to reference it through the package.
2026-07-11 15:14:22 +02:00
Zoltan Papp
07ffbc9424 [client] Extract peer status recorder into its own package
Move the Status recorder and its state types out of the peer package
into client/internal/peer/status, split by struct across recorder.go,
peer_state.go, full_status.go, events.go, notifier.go and route.go
instead of one 1600-line file. Rename the type Status -> Recorder
(NewRecorder already implied it; avoids status.Status stutter). Split
conn_status.go: the ConnStatus type and its constants move to the status
package, connStatusInputs stays with the peer event loop.

The peer package references the status package directly; a transitional
status_alias.go re-exports the moved symbols for the ~50 external callers
still using peer.Status/State/ConnStatus, to be removed once they are
migrated.
2026-07-11 15:09:45 +02:00
Zoltan Papp
467b2a1712 [client] Deduplicate peer state update methods in Status
The five UpdatePeer* methods repeated the same lock/copy/snapshot/notify
boilerplate. Extract a shared updatePeer helper taking a router-notify
predicate and a mutate closure; each method now only declares which
fields it copies. Replace the repeated inline peer-not-found error with
an errPeerNotExists sentinel.
2026-07-11 14:30:37 +02:00
Zoltan Papp
cb4088484e [client] Suppress ICE events from replaced agents
Track connection state per agent generation instead of the shared
lastKnownState field, which was written from concurrent agent callbacks
without a lock. The connect goroutine now drops the connection if the
agent was replaced during dialing, and a replaced agent's late
disconnected callback no longer reaches the conn after its successor
already reported ready. Only the agent whose connection was last
reported ready may report it down.
2026-07-11 14:22:58 +02:00
Zoltan Papp
7620599961 [client] Fix ICE dialer cancel race in WorkerICE.connect
The connect goroutine read the agentDialerCancel field without holding
muxAgent, racing with OnNewOffer replacing it for a new session. On
failure paths the stale read could cancel the new session's dialer
instead of its own. Pass the cancel func of the owning session as a
parameter, like the dialer context.
2026-07-11 13:37:59 +02:00
Zoltan Papp
5f98524e02 [client] Refactor peer Conn to a single-owner event loop
Replace the mutex-guarded callback model of peer.Conn with a per-peer
event loop that exclusively owns all mutable connection state. External
callers and transport workers post typed events into a non-blocking,
coalescing mailbox instead of contending on conn.mu:

- offers/answers coalesce to the newest message, a new offer flushes
  queued candidates of the superseded session
- candidates are applied in arrival order from a bounded FIFO
- transport state changes are never dropped
- the blocking relay dial runs on a helper goroutine with a single dial
  in flight; signaling I/O (offer/answer sends) runs off the loop

conn.mu now only guards the open/close lifecycle. Close posts a close
event and waits for the loop teardown; the loop also tears down on
engine context cancellation and releases resources of unprocessed
events.

Delete the Handshaker listener machinery (Listen loop, unbuffered
drop-on-busy channels, AsyncOfferListener with its double-processing of
the first offer), the never-wired dispatcher package and the unused
ICEMonitor.ReconnectCh. Fix a goroutine leak in the WG watcher test
that raced with tests mutating the package-level check timing vars.
2026-07-11 13:30:06 +02:00
37 changed files with 1719 additions and 1444 deletions

View File

@@ -48,6 +48,7 @@ import (
"github.com/netbirdio/netbird/client/internal/peer"
"github.com/netbirdio/netbird/client/internal/peer/guard"
icemaker "github.com/netbirdio/netbird/client/internal/peer/ice"
"github.com/netbirdio/netbird/client/internal/peer/signaling"
"github.com/netbirdio/netbird/client/internal/peerstore"
"github.com/netbirdio/netbird/client/internal/portforward"
"github.com/netbirdio/netbird/client/internal/profilemanager"
@@ -182,7 +183,7 @@ type EngineServices struct {
type Engine struct {
// signal is a Signal Service client
signal signal.Client
signaler *peer.Signaler
signaler *signaling.Signaler
// mgmClient is a Management Service client
mgmClient mgm.Client
// peerConns is a map that holds all the peers that are known to this peer
@@ -318,7 +319,7 @@ func NewEngine(
ctx: ctx,
cancel: cancel,
signal: services.SignalClient,
signaler: peer.NewSignaler(services.SignalClient, config.WgPrivateKey),
signaler: signaling.NewSignaler(services.SignalClient, config.WgPrivateKey),
mgmClient: services.MgmClient,
relayManager: services.RelayManager,
peerStore: peerstore.NewConnStore(),
@@ -2747,7 +2748,7 @@ func createFile(path string) error {
return file.Close()
}
func convertToOfferAnswer(msg *sProto.Message) (*peer.OfferAnswer, error) {
func convertToOfferAnswer(msg *sProto.Message) (*signaling.OfferAnswer, error) {
remoteCred, err := signal.UnMarshalCredential(msg)
if err != nil {
return nil, err
@@ -2763,9 +2764,9 @@ func convertToOfferAnswer(msg *sProto.Message) (*peer.OfferAnswer, error) {
}
// Handle optional SessionID
var sessionID *peer.ICESessionID
var sessionID *icemaker.SessionID
if sessionBytes := msg.GetBody().GetSessionId(); sessionBytes != nil {
if id, err := peer.ICESessionIDFromBytes(sessionBytes); err != nil {
if id, err := icemaker.SessionIDFromBytes(sessionBytes); err != nil {
log.Warnf("Invalid session ID in message: %v", err)
sessionID = nil // Set to nil if conversion fails
} else {
@@ -2775,8 +2776,8 @@ func convertToOfferAnswer(msg *sProto.Message) (*peer.OfferAnswer, error) {
relayIP := decodeRelayIP(msg.GetBody().GetRelayServerIP())
offerAnswer := peer.OfferAnswer{
IceCredentials: peer.IceCredentials{
offerAnswer := signaling.OfferAnswer{
IceCredentials: signaling.IceCredentials{
UFrag: remoteCred.UFrag,
Pwd: remoteCred.Pwd,
},

File diff suppressed because it is too large Load Diff

View File

@@ -1,18 +1,5 @@
package peer
import (
log "github.com/sirupsen/logrus"
)
const (
// StatusIdle indicate the peer is in disconnected state
StatusIdle ConnStatus = iota
// StatusConnecting indicate the peer is in connecting state
StatusConnecting
// StatusConnected indicate the peer is in connected state
StatusConnected
)
// connStatusInputs is the primitive-valued snapshot of the state that drives the
// tri-state connection classification. Extracted so the decision logic can be unit-tested
// without constructing full Worker/Handshaker objects.
@@ -21,24 +8,7 @@ type connStatusInputs struct {
peerUsesRelay bool // remote peer advertises relay support AND local has relay
relayConnected bool // statusRelay reports Connected (independent of whether peer uses relay)
remoteSupportsICE bool // remote peer sent ICE credentials
iceWorkerCreated bool // local WorkerICE exists (false in force-relay mode)
iceWorkerCreated bool // local ICE worker exists (false in force-relay mode)
iceStatusConnecting bool // statusICE is anything other than Disconnected
iceInProgress bool // a negotiation is currently in flight
}
// ConnStatus describe the status of a peer's connection
type ConnStatus int32
func (s ConnStatus) String() string {
switch s {
case StatusConnecting:
return "Connecting"
case StatusConnected:
return "Connected"
case StatusIdle:
return "Idle"
default:
log.Errorf("unknown status: %d", s)
return "INVALID_PEER_CONNECTION_STATUS"
}
}

View File

@@ -3,28 +3,33 @@ package peer
import (
"context"
"fmt"
"net/netip"
"os"
"testing"
"time"
log "github.com/sirupsen/logrus"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/client/iface"
"github.com/netbirdio/netbird/client/internal/peer/dispatcher"
"github.com/netbirdio/netbird/client/internal/peer/guard"
"github.com/netbirdio/netbird/client/internal/peer/ice"
"github.com/netbirdio/netbird/client/internal/peer/metricsstages"
"github.com/netbirdio/netbird/client/internal/peer/signaling"
"github.com/netbirdio/netbird/client/internal/peer/status"
"github.com/netbirdio/netbird/client/internal/stdnet"
"github.com/netbirdio/netbird/util"
)
var testDispatcher = dispatcher.NewConnectionDispatcher()
var connConf = ConnConfig{
Key: "LLHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
LocalKey: "RRHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
Timeout: time.Second,
LocalWgPort: 51820,
WgConfig: WgConfig{
AllowedIps: []netip.Prefix{netip.MustParsePrefix("100.64.0.1/32")},
},
ICEConfig: ice.Config{
InterfaceBlackList: nil,
},
@@ -52,92 +57,37 @@ func TestConn_GetKey(t *testing.T) {
swWatcher := guard.NewSRWatcher(nil, nil, nil, connConf.ICEConfig)
sd := ServiceDependencies{
SrWatcher: swWatcher,
PeerConnDispatcher: testDispatcher,
SrWatcher: swWatcher,
}
conn, err := NewConn(connConf, sd)
if err != nil {
return
}
require.NoError(t, err)
got := conn.GetKey()
assert.Equal(t, got, connConf.Key, "they should be equal")
}
func TestConn_OnRemoteOffer(t *testing.T) {
// TestConn_DiscardMessagesWhenNotOpened: signal messages posted to a not yet
// opened connection must be discarded without blocking or panicking.
func TestConn_DiscardMessagesWhenNotOpened(t *testing.T) {
swWatcher := guard.NewSRWatcher(nil, nil, nil, connConf.ICEConfig)
sd := ServiceDependencies{
StatusRecorder: NewRecorder("https://mgm"),
SrWatcher: swWatcher,
PeerConnDispatcher: testDispatcher,
StatusRecorder: status.NewRecorder("https://mgm"),
SrWatcher: swWatcher,
}
conn, err := NewConn(connConf, sd)
if err != nil {
return
}
require.NoError(t, err)
onNewOfferChan := make(chan struct{})
conn.handshaker.AddRelayListener(func(remoteOfferAnswer *OfferAnswer) {
onNewOfferChan <- struct{}{}
})
conn.OnRemoteOffer(OfferAnswer{
IceCredentials: IceCredentials{
offerAnswer := signaling.OfferAnswer{
IceCredentials: signaling.IceCredentials{
UFrag: "test",
Pwd: "test",
},
WgListenPort: 0,
Version: "",
})
ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
defer cancel()
select {
case <-onNewOfferChan:
// success
case <-ctx.Done():
t.Error("expected to receive a new offer notification, but timed out")
}
}
func TestConn_OnRemoteAnswer(t *testing.T) {
swWatcher := guard.NewSRWatcher(nil, nil, nil, connConf.ICEConfig)
sd := ServiceDependencies{
StatusRecorder: NewRecorder("https://mgm"),
SrWatcher: swWatcher,
PeerConnDispatcher: testDispatcher,
}
conn, err := NewConn(connConf, sd)
if err != nil {
return
}
onNewOfferChan := make(chan struct{})
conn.handshaker.AddRelayListener(func(remoteOfferAnswer *OfferAnswer) {
onNewOfferChan <- struct{}{}
})
conn.OnRemoteAnswer(OfferAnswer{
IceCredentials: IceCredentials{
UFrag: "test",
Pwd: "test",
},
WgListenPort: 0,
Version: "",
})
ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
defer cancel()
select {
case <-onNewOfferChan:
// success
case <-ctx.Done():
t.Error("expected to receive a new offer notification, but timed out")
}
conn.OnRemoteOffer(offerAnswer)
conn.OnRemoteAnswer(offerAnswer)
conn.OnRemoteCandidate(nil, nil)
conn.Close(false)
}
func TestConn_presharedKey(t *testing.T) {
@@ -320,7 +270,7 @@ func newWGTimeoutTestConn(rosenpassEnabled bool, disconnected *[]string) *Conn {
ctx: context.Background(),
config: cfg,
Log: log.WithField("peer", cfg.Key),
metricsStages: &MetricsStages{},
metricsStages: &metricsstages.MetricsStages{},
}
conn.SetOnDisconnected(func(remotePeer string) {
*disconnected = append(*disconnected, remotePeer)
@@ -339,20 +289,20 @@ func TestConn_onWGDisconnected_EscalatesToRosenpassReset(t *testing.T) {
conn := newWGTimeoutTestConn(true, &disconnected)
for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
conn.onWGDisconnected()
conn.handleWGTimeout()
}
assert.Empty(t, disconnected, "escalation must not fire below the threshold")
conn.onWGDisconnected()
conn.handleWGTimeout()
assert.Equal(t, []string{conn.config.WgConfig.RemoteKey}, disconnected,
"reaching the threshold must report the peer disconnected once")
for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
conn.onWGDisconnected()
conn.handleWGTimeout()
}
assert.Len(t, disconnected, 1, "escalation must restart counting after firing")
conn.onWGDisconnected()
conn.handleWGTimeout()
assert.Len(t, disconnected, 2, "continued timeouts must escalate again")
}
@@ -364,12 +314,12 @@ func TestConn_onWGDisconnected_CheckSuccessResetsEscalation(t *testing.T) {
conn := newWGTimeoutTestConn(true, &disconnected)
for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
conn.onWGDisconnected()
conn.handleWGTimeout()
}
conn.onWGCheckSuccess()
conn.handleWGCheckSuccess()
for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
conn.onWGDisconnected()
conn.handleWGTimeout()
}
assert.Empty(t, disconnected, "handshake success must reset the timeout count")
}
@@ -382,7 +332,7 @@ func TestConn_onWGDisconnected_NoEscalationWithoutRosenpass(t *testing.T) {
conn := newWGTimeoutTestConn(false, &disconnected)
for i := 0; i < wgTimeoutEscalationThreshold*3; i++ {
conn.onWGDisconnected()
conn.handleWGTimeout()
}
assert.Empty(t, disconnected, "escalation must be limited to rosenpass connections")
}

View File

@@ -1,52 +0,0 @@
package dispatcher
import (
"sync"
"github.com/netbirdio/netbird/client/internal/peer/id"
)
type ConnectionListener struct {
OnConnected func(peerID id.ConnID)
OnDisconnected func(peerID id.ConnID)
}
type ConnectionDispatcher struct {
listeners map[*ConnectionListener]struct{}
mu sync.Mutex
}
func NewConnectionDispatcher() *ConnectionDispatcher {
return &ConnectionDispatcher{
listeners: make(map[*ConnectionListener]struct{}),
}
}
func (e *ConnectionDispatcher) AddListener(listener *ConnectionListener) {
e.mu.Lock()
defer e.mu.Unlock()
e.listeners[listener] = struct{}{}
}
func (e *ConnectionDispatcher) RemoveListener(listener *ConnectionListener) {
e.mu.Lock()
defer e.mu.Unlock()
delete(e.listeners, listener)
}
func (e *ConnectionDispatcher) NotifyConnected(peerConnID id.ConnID) {
e.mu.Lock()
defer e.mu.Unlock()
for listener := range e.listeners {
listener.OnConnected(peerConnID)
}
}
func (e *ConnectionDispatcher) NotifyDisconnected(peerConnID id.ConnID) {
e.mu.Lock()
defer e.mu.Unlock()
for listener := range e.listeners {
listener.OnDisconnected(peerConnID)
}
}

View File

@@ -0,0 +1,69 @@
package peer
import (
"time"
"github.com/pion/ice/v4"
"github.com/netbirdio/netbird/client/internal/peer/signaling"
"github.com/netbirdio/netbird/client/internal/peer/worker"
"github.com/netbirdio/netbird/route"
)
// event is a message processed by the Conn event loop. All mutable Conn state
// is owned by that loop; producers deliver events through the mailbox and
// never mutate Conn state directly.
type event any
// evClose asks the event loop to tear down the connection. done is closed
// once the teardown finished.
type evClose struct {
signalToRemote bool
done chan struct{}
}
type evRemoteOffer struct {
offer signaling.OfferAnswer
}
type evRemoteAnswer struct {
answer signaling.OfferAnswer
}
type evRemoteCandidate struct {
candidate ice.Candidate
haRoutes route.HAMap
}
type evICEReady struct {
priority worker.ConnPriority
info worker.ICEConnInfo
}
type evICEDown struct {
sessionChanged bool
}
type evRelayReady struct {
info worker.RelayConnInfo
}
type evRelayDown struct{}
// evRelayDialDone reports that the relay dial helper goroutine finished,
// successfully or not, so the loop may dispatch a pending offer.
type evRelayDialDone struct{}
type evWGTimeout struct{}
// evWGHandshake reports the first WireGuard handshake of the current watcher run.
type evWGHandshake struct {
when time.Time
}
// evWGCheckOK reports a watcher check that observed a fresh handshake,
// including handshakes of connections that were already up.
type evWGCheckOK struct{}
// evGuardTick asks the loop to send a new offer to restore connectivity.
type evGuardTick struct{}

View File

@@ -21,8 +21,6 @@ const (
)
type ICEMonitor struct {
ReconnectCh chan struct{}
iFaceDiscover stdnet.ExternalIFaceDiscover
iceConfig icemaker.Config
tickerPeriod time.Duration
@@ -34,7 +32,6 @@ type ICEMonitor struct {
func NewICEMonitor(iFaceDiscover stdnet.ExternalIFaceDiscover, config icemaker.Config, period time.Duration) *ICEMonitor {
log.Debugf("prepare ICE monitor with period: %s", period)
cm := &ICEMonitor{
ReconnectCh: make(chan struct{}, 1),
iFaceDiscover: iFaceDiscover,
iceConfig: config,
tickerPeriod: period,

View File

@@ -1,246 +0,0 @@
package peer
import (
"context"
"errors"
"net/netip"
"sync"
"sync/atomic"
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/version"
)
var (
ErrSignalIsNotReady = errors.New("signal is not ready")
)
// IceCredentials ICE protocol credentials struct
type IceCredentials struct {
UFrag string
Pwd string
}
// OfferAnswer represents a session establishment offer or answer
type OfferAnswer struct {
IceCredentials IceCredentials
// WgListenPort is a remote WireGuard listen port.
// This field is used when establishing a direct WireGuard connection without any proxy.
// We can set the remote peer's endpoint with this port.
WgListenPort int
// Version of NetBird Agent
Version string
// RosenpassPubKey is the Rosenpass public key of the remote peer when receiving this message
// This value is the local Rosenpass server public key when sending the message
RosenpassPubKey []byte
// RosenpassAddr is the Rosenpass server address (IP:port) of the remote peer when receiving this message
// This value is the local Rosenpass server address when sending the message
RosenpassAddr string
// relay server address
RelaySrvAddress string
// RelaySrvIP is the IP the remote peer is connected to on its
// relay server. Used as a dial target if DNS for RelaySrvAddress
// fails. Zero value if the peer did not advertise an IP.
RelaySrvIP netip.Addr
// SessionID is the unique identifier of the session, used to discard old messages
SessionID *ICESessionID
}
func (o *OfferAnswer) hasICECredentials() bool {
return o.IceCredentials.UFrag != "" && o.IceCredentials.Pwd != ""
}
type Handshaker struct {
mu sync.Mutex
log *log.Entry
config ConnConfig
signaler *Signaler
ice *WorkerICE
relay *WorkerRelay
metricsStages *MetricsStages
// relayListener is not blocking because the listener is using a goroutine to process the messages
// and it will only keep the latest message if multiple offers are received in a short time
// this is to avoid blocking the handshaker if the listener is doing some heavy processing
// and also to avoid processing old offers if multiple offers are received in a short time
// the listener will always process the latest offer
relayListener *AsyncOfferListener
iceListener func(remoteOfferAnswer *OfferAnswer)
// remoteICESupported tracks whether the remote peer includes ICE credentials in its offers/answers.
// When false, the local side skips ICE listener dispatch and suppresses ICE credentials in responses.
remoteICESupported atomic.Bool
// remoteOffersCh is a channel used to wait for remote credentials to proceed with the connection
remoteOffersCh chan OfferAnswer
// remoteAnswerCh is a channel used to wait for remote credentials answer (confirmation of our offer) to proceed with the connection
remoteAnswerCh chan OfferAnswer
}
func NewHandshaker(log *log.Entry, config ConnConfig, signaler *Signaler, ice *WorkerICE, relay *WorkerRelay, metricsStages *MetricsStages) *Handshaker {
h := &Handshaker{
log: log,
config: config,
signaler: signaler,
ice: ice,
relay: relay,
metricsStages: metricsStages,
remoteOffersCh: make(chan OfferAnswer),
remoteAnswerCh: make(chan OfferAnswer),
}
// assume remote supports ICE until we learn otherwise from received offers
h.remoteICESupported.Store(ice != nil)
return h
}
func (h *Handshaker) RemoteICESupported() bool {
return h.remoteICESupported.Load()
}
func (h *Handshaker) AddRelayListener(offer func(remoteOfferAnswer *OfferAnswer)) {
h.relayListener = NewAsyncOfferListener(offer)
}
func (h *Handshaker) AddICEListener(offer func(remoteOfferAnswer *OfferAnswer)) {
h.iceListener = offer
}
func (h *Handshaker) Listen(ctx context.Context) {
for {
select {
case remoteOfferAnswer := <-h.remoteOffersCh:
h.log.Infof("received offer, running version %s, remote WireGuard listen port %d, session id: %s, remote ICE supported: %t", remoteOfferAnswer.Version, remoteOfferAnswer.WgListenPort, remoteOfferAnswer.SessionIDString(), remoteOfferAnswer.hasICECredentials())
// Record signaling received for reconnection attempts
if h.metricsStages != nil {
h.metricsStages.RecordSignalingReceived()
}
h.updateRemoteICEState(&remoteOfferAnswer)
if h.relayListener != nil {
h.relayListener.Notify(&remoteOfferAnswer)
}
if h.iceListener != nil && h.RemoteICESupported() {
h.iceListener(&remoteOfferAnswer)
}
if err := h.sendAnswer(); err != nil {
h.log.Errorf("failed to send remote offer confirmation: %s", err)
continue
}
case remoteOfferAnswer := <-h.remoteAnswerCh:
h.log.Infof("received answer, running version %s, remote WireGuard listen port %d, session id: %s, remote ICE supported: %t", remoteOfferAnswer.Version, remoteOfferAnswer.WgListenPort, remoteOfferAnswer.SessionIDString(), remoteOfferAnswer.hasICECredentials())
// Record signaling received for reconnection attempts
if h.metricsStages != nil {
h.metricsStages.RecordSignalingReceived()
}
h.updateRemoteICEState(&remoteOfferAnswer)
if h.relayListener != nil {
h.relayListener.Notify(&remoteOfferAnswer)
}
if h.iceListener != nil && h.RemoteICESupported() {
h.iceListener(&remoteOfferAnswer)
}
case <-ctx.Done():
h.log.Infof("stop listening for remote offers and answers")
return
}
}
}
func (h *Handshaker) SendOffer() error {
h.mu.Lock()
defer h.mu.Unlock()
return h.sendOffer()
}
// OnRemoteOffer handles an offer from the remote peer and returns true if the message was accepted, false otherwise
// doesn't block, discards the message if connection wasn't ready
func (h *Handshaker) OnRemoteOffer(offer OfferAnswer) {
select {
case h.remoteOffersCh <- offer:
return
default:
h.log.Warnf("skipping remote offer message because receiver not ready")
// connection might not be ready yet to receive so we ignore the message
return
}
}
// OnRemoteAnswer handles an offer from the remote peer and returns true if the message was accepted, false otherwise
// doesn't block, discards the message if connection wasn't ready
func (h *Handshaker) OnRemoteAnswer(answer OfferAnswer) {
select {
case h.remoteAnswerCh <- answer:
return
default:
// connection might not be ready yet to receive so we ignore the message
h.log.Warnf("skipping remote answer message because receiver not ready")
return
}
}
// sendOffer prepares local user credentials and signals them to the remote peer
func (h *Handshaker) sendOffer() error {
if !h.signaler.Ready() {
return ErrSignalIsNotReady
}
offer := h.buildOfferAnswer()
h.log.Debugf("sending offer with serial: %s", offer.SessionIDString())
return h.signaler.SignalOffer(offer, h.config.Key)
}
func (h *Handshaker) sendAnswer() error {
answer := h.buildOfferAnswer()
h.log.Debugf("sending answer with serial: %s", answer.SessionIDString())
return h.signaler.SignalAnswer(answer, h.config.Key)
}
func (h *Handshaker) buildOfferAnswer() OfferAnswer {
answer := OfferAnswer{
WgListenPort: h.config.LocalWgPort,
Version: version.NetbirdVersion(),
RosenpassPubKey: h.config.RosenpassConfig.PubKey,
RosenpassAddr: h.config.RosenpassConfig.Addr,
}
if h.ice != nil && h.RemoteICESupported() {
uFrag, pwd := h.ice.GetLocalUserCredentials()
sid := h.ice.SessionID()
answer.IceCredentials = IceCredentials{uFrag, pwd}
answer.SessionID = &sid
}
if addr, ip, err := h.relay.RelayInstanceAddress(); err == nil {
answer.RelaySrvAddress = addr
answer.RelaySrvIP = ip
}
return answer
}
func (h *Handshaker) updateRemoteICEState(offer *OfferAnswer) {
hasICE := offer.hasICECredentials()
prev := h.remoteICESupported.Swap(hasICE)
if prev != hasICE {
if hasICE {
h.log.Infof("remote peer started sending ICE credentials")
} else {
h.log.Infof("remote peer stopped sending ICE credentials")
if h.ice != nil {
h.ice.Close()
}
}
}
}

View File

@@ -1,62 +0,0 @@
package peer
import (
"sync"
)
type callbackFunc func(remoteOfferAnswer *OfferAnswer)
func (oa *OfferAnswer) SessionIDString() string {
if oa.SessionID == nil {
return "unknown"
}
return oa.SessionID.String()
}
type AsyncOfferListener struct {
fn callbackFunc
running bool
latest *OfferAnswer
mu sync.Mutex
}
func NewAsyncOfferListener(fn callbackFunc) *AsyncOfferListener {
return &AsyncOfferListener{
fn: fn,
}
}
func (o *AsyncOfferListener) Notify(remoteOfferAnswer *OfferAnswer) {
o.mu.Lock()
defer o.mu.Unlock()
// Store the latest offer
o.latest = remoteOfferAnswer
// If already running, the running goroutine will pick up this latest value
if o.running {
return
}
// Start processing
o.running = true
// Process in a goroutine to avoid blocking the caller
go func(remoteOfferAnswer *OfferAnswer) {
for {
o.fn(remoteOfferAnswer)
o.mu.Lock()
if o.latest == nil {
// No more work to do
o.running = false
o.mu.Unlock()
return
}
remoteOfferAnswer = o.latest
// Clear the latest to mark it as being processed
o.latest = nil
o.mu.Unlock()
}
}(remoteOfferAnswer)
}

View File

@@ -1,39 +0,0 @@
package peer
import (
"testing"
"time"
)
func Test_newOfferListener(t *testing.T) {
dummyOfferAnswer := &OfferAnswer{}
runChan := make(chan struct{}, 10)
longRunningFn := func(remoteOfferAnswer *OfferAnswer) {
time.Sleep(1 * time.Second)
runChan <- struct{}{}
}
hl := NewAsyncOfferListener(longRunningFn)
hl.Notify(dummyOfferAnswer)
hl.Notify(dummyOfferAnswer)
hl.Notify(dummyOfferAnswer)
// Wait for exactly 2 callbacks
for i := 0; i < 2; i++ {
select {
case <-runChan:
case <-time.After(3 * time.Second):
t.Fatal("Timeout waiting for callback")
}
}
// Verify no additional callbacks happen
select {
case <-runChan:
t.Fatal("Unexpected additional callback")
case <-time.After(100 * time.Millisecond):
t.Log("Correctly received exactly 2 callbacks")
}
}

View File

@@ -1,4 +1,4 @@
package peer
package ice
import (
"crypto/rand"
@@ -9,26 +9,26 @@ import (
const sessionIDSize = 5
type ICESessionID string
type SessionID string
// NewICESessionID generates a new session ID for distinguishing sessions
func NewICESessionID() (ICESessionID, error) {
// NewSessionID generates a new session ID for distinguishing sessions
func NewSessionID() (SessionID, error) {
b := make([]byte, sessionIDSize)
if _, err := io.ReadFull(rand.Reader, b); err != nil {
return "", fmt.Errorf("failed to generate session ID: %w", err)
}
return ICESessionID(hex.EncodeToString(b)), nil
return SessionID(hex.EncodeToString(b)), nil
}
func ICESessionIDFromBytes(b []byte) (ICESessionID, error) {
func SessionIDFromBytes(b []byte) (SessionID, error) {
if len(b) != sessionIDSize {
return "", fmt.Errorf("invalid session ID length: %d", len(b))
}
return ICESessionID(hex.EncodeToString(b)), nil
return SessionID(hex.EncodeToString(b)), nil
}
// Bytes returns the raw bytes of the session ID for protobuf serialization
func (id ICESessionID) Bytes() ([]byte, error) {
func (id SessionID) Bytes() ([]byte, error) {
if len(id) == 0 {
return nil, fmt.Errorf("ICE session ID is empty")
}
@@ -42,6 +42,6 @@ func (id ICESessionID) Bytes() ([]byte, error) {
return b, nil
}
func (id ICESessionID) String() string {
func (id SessionID) String() string {
return string(id)
}

View File

@@ -1,22 +0,0 @@
package peer
import (
"net"
"net/netip"
"time"
"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
"github.com/netbirdio/netbird/client/iface/configurer"
"github.com/netbirdio/netbird/client/iface/wgaddr"
"github.com/netbirdio/netbird/client/iface/wgproxy"
)
type WGIface interface {
UpdatePeer(peerKey string, allowedIps []netip.Prefix, keepAlive time.Duration, endpoint *net.UDPAddr, preSharedKey *wgtypes.Key) error
RemovePeer(peerKey string) error
GetStats() (map[string]configurer.WGStats, error)
GetProxy() wgproxy.Proxy
Address() wgaddr.Address
RemoveEndpointAddress(key string) error
}

View File

@@ -1,11 +0,0 @@
package peer
// Listener is a callback type about the NetBird network connection state
type Listener interface {
OnConnected()
OnDisconnected()
OnConnecting()
OnDisconnecting()
OnAddressChanged(string, string)
OnPeersListChanged(int)
}

View File

@@ -0,0 +1,116 @@
package peer
import (
"sync"
)
// maxQueuedCandidates bounds the remote candidate queue; on overflow the
// oldest candidate is dropped. Lost candidates are recovered by the next
// offer exchange triggered by the guard.
const maxQueuedCandidates = 128
// mailbox is the coalescing inbox of the Conn event loop. Posting never
// blocks. Per message kind either the latest value wins (offer, answer,
// guard tick), the values queue in bounded FIFO order (candidates) or in
// unbounded FIFO order (lifecycle and transport state changes, which are
// low-volume and must not be lost). A new offer flushes the queued
// candidates because they belong to the superseded session.
type mailbox struct {
mu sync.Mutex
closed bool
lifecycle []event
transport []event
offer *evRemoteOffer
answer *evRemoteAnswer
candidates []evRemoteCandidate
guardTick bool
wake chan struct{}
}
func newMailbox() *mailbox {
return &mailbox{
wake: make(chan struct{}, 1),
}
}
// post stores the event and wakes the loop. It reports false if the mailbox
// is already closed and the event was not accepted.
func (m *mailbox) post(ev event) bool {
m.mu.Lock()
if m.closed {
m.mu.Unlock()
return false
}
switch e := ev.(type) {
case evClose:
m.lifecycle = append(m.lifecycle, e)
case evRemoteOffer:
m.offer = &e
m.candidates = nil
case evRemoteAnswer:
m.answer = &e
case evRemoteCandidate:
if len(m.candidates) >= maxQueuedCandidates {
m.candidates = m.candidates[1:]
}
m.candidates = append(m.candidates, e)
case evGuardTick:
m.guardTick = true
default:
m.transport = append(m.transport, ev)
}
m.mu.Unlock()
select {
case m.wake <- struct{}{}:
default:
}
return true
}
// drain returns the pending events in processing order: lifecycle first,
// then transport state changes, the coalesced offer and answer, the queued
// candidates and finally the guard tick.
func (m *mailbox) drain() []event {
m.mu.Lock()
defer m.mu.Unlock()
return m.drainLocked()
}
// closeAndDrain marks the mailbox closed so further posts are rejected and
// returns the events that were still pending.
func (m *mailbox) closeAndDrain() []event {
m.mu.Lock()
defer m.mu.Unlock()
m.closed = true
return m.drainLocked()
}
func (m *mailbox) drainLocked() []event {
evs := make([]event, 0, len(m.lifecycle)+len(m.transport)+len(m.candidates)+3)
evs = append(evs, m.lifecycle...)
evs = append(evs, m.transport...)
if m.offer != nil {
evs = append(evs, *m.offer)
}
if m.answer != nil {
evs = append(evs, *m.answer)
}
for _, c := range m.candidates {
evs = append(evs, c)
}
if m.guardTick {
evs = append(evs, evGuardTick{})
}
m.lifecycle = nil
m.transport = nil
m.offer = nil
m.answer = nil
m.candidates = nil
m.guardTick = false
return evs
}

View File

@@ -0,0 +1,128 @@
package peer
import (
"testing"
"github.com/netbirdio/netbird/client/internal/peer/signaling"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func TestMailbox_OfferCoalescing(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evRemoteOffer{offer: signaling.OfferAnswer{WgListenPort: 1}}))
require.True(t, mb.post(evRemoteOffer{offer: signaling.OfferAnswer{WgListenPort: 2}}))
require.True(t, mb.post(evRemoteOffer{offer: signaling.OfferAnswer{WgListenPort: 3}}))
evs := mb.drain()
require.Len(t, evs, 1, "consecutive offers must coalesce to a single event")
offer, ok := evs[0].(evRemoteOffer)
require.True(t, ok, "coalesced event must be an offer")
assert.Equal(t, 3, offer.offer.WgListenPort, "the newest offer must win")
}
func TestMailbox_OfferFlushesCandidates(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evRemoteCandidate{}))
require.True(t, mb.post(evRemoteCandidate{}))
require.True(t, mb.post(evRemoteOffer{offer: signaling.OfferAnswer{}}))
evs := mb.drain()
require.Len(t, evs, 1, "candidates of the superseded session must be flushed")
_, ok := evs[0].(evRemoteOffer)
assert.True(t, ok, "only the offer must remain after the flush")
}
func TestMailbox_CandidatesKeepOrderAfterOffer(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evRemoteOffer{offer: signaling.OfferAnswer{}}))
require.True(t, mb.post(evRemoteCandidate{haRoutes: nil}))
require.True(t, mb.post(evRemoteCandidate{haRoutes: nil}))
evs := mb.drain()
require.Len(t, evs, 3)
_, ok := evs[0].(evRemoteOffer)
assert.True(t, ok, "offer must be processed before the candidates")
for _, ev := range evs[1:] {
_, ok := ev.(evRemoteCandidate)
assert.True(t, ok, "candidates posted after the offer must survive")
}
}
func TestMailbox_CandidateQueueBounded(t *testing.T) {
mb := newMailbox()
for i := 0; i < maxQueuedCandidates+10; i++ {
require.True(t, mb.post(evRemoteCandidate{}))
}
evs := mb.drain()
assert.Len(t, evs, maxQueuedCandidates, "candidate queue must stay bounded")
}
func TestMailbox_DrainOrder(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evGuardTick{}))
require.True(t, mb.post(evRemoteAnswer{answer: signaling.OfferAnswer{}}))
require.True(t, mb.post(evRemoteOffer{offer: signaling.OfferAnswer{}}))
require.True(t, mb.post(evRelayDown{}))
require.True(t, mb.post(evICEDown{sessionChanged: true}))
require.True(t, mb.post(evClose{}))
evs := mb.drain()
require.Len(t, evs, 6)
_, ok := evs[0].(evClose)
assert.True(t, ok, "lifecycle events must come first")
_, ok = evs[1].(evRelayDown)
assert.True(t, ok, "transport events must keep FIFO order")
_, ok = evs[2].(evICEDown)
assert.True(t, ok, "transport events must keep FIFO order")
_, ok = evs[3].(evRemoteOffer)
assert.True(t, ok, "offer must come after transport events")
_, ok = evs[4].(evRemoteAnswer)
assert.True(t, ok, "answer must come after the offer")
_, ok = evs[5].(evGuardTick)
assert.True(t, ok, "guard tick must come last")
}
func TestMailbox_GuardTickCoalesced(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evGuardTick{}))
require.True(t, mb.post(evGuardTick{}))
require.True(t, mb.post(evGuardTick{}))
evs := mb.drain()
assert.Len(t, evs, 1, "guard ticks must coalesce to a single event")
}
func TestMailbox_PostAfterCloseRejected(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evRelayDown{}))
leftovers := mb.closeAndDrain()
assert.Len(t, leftovers, 1, "pending events must be returned on close")
assert.False(t, mb.post(evRelayDown{}), "posts must be rejected after close")
assert.Empty(t, mb.drain(), "no events must remain after close")
}
func TestMailbox_WakeSignal(t *testing.T) {
mb := newMailbox()
require.True(t, mb.post(evRelayDown{}))
require.True(t, mb.post(evGuardTick{}))
select {
case <-mb.wake:
default:
t.Fatal("wake signal must be pending after posts")
}
assert.Len(t, mb.drain(), 2, "a single wake must deliver all pending events")
}

View File

@@ -1,4 +1,4 @@
package peer
package metricsstages
import (
"sync"

View File

@@ -1,4 +1,4 @@
package peer
package metricsstages
import (
"testing"

View File

@@ -0,0 +1,189 @@
package signaling
import (
"errors"
"net/netip"
"sync"
"sync/atomic"
log "github.com/sirupsen/logrus"
icemaker "github.com/netbirdio/netbird/client/internal/peer/ice"
relayClient "github.com/netbirdio/netbird/shared/relay/client"
"github.com/netbirdio/netbird/version"
)
var (
ErrSignalIsNotReady = errors.New("signal is not ready")
)
// IceCredentials ICE protocol credentials struct
type IceCredentials struct {
UFrag string
Pwd string
}
// OfferAnswer represents a session establishment offer or answer
type OfferAnswer struct {
IceCredentials IceCredentials
// WgListenPort is a remote WireGuard listen port.
// This field is used when establishing a direct WireGuard connection without any proxy.
// We can set the remote peer's endpoint with this port.
WgListenPort int
// Version of NetBird Agent
Version string
// RosenpassPubKey is the Rosenpass public key of the remote peer when receiving this message
// This value is the local Rosenpass server public key when sending the message
RosenpassPubKey []byte
// RosenpassAddr is the Rosenpass server address (IP:port) of the remote peer when receiving this message
// This value is the local Rosenpass server address when sending the message
RosenpassAddr string
// relay server address
RelaySrvAddress string
// RelaySrvIP is the IP the remote peer is connected to on its
// relay server. Used as a dial target if DNS for RelaySrvAddress
// fails. Zero value if the peer did not advertise an IP.
RelaySrvIP netip.Addr
// SessionID is the unique identifier of the session, used to discard old messages
SessionID *icemaker.SessionID
}
func (o *OfferAnswer) HasICECredentials() bool {
return o.IceCredentials.UFrag != "" && o.IceCredentials.Pwd != ""
}
func (o *OfferAnswer) SessionIDString() string {
if o.SessionID == nil {
return "unknown"
}
return o.SessionID.String()
}
// Config carries the peer-specific values the Handshaker embeds into offers
// and answers.
type Config struct {
Key string
LocalWgPort int
RosenpassPubKey []byte
RosenpassAddr string
}
// Credentials are the local ICE credentials and session id the Handshaker embeds in offers.
type Credentials struct {
UFrag string
Pwd string
SessionID icemaker.SessionID
}
// ICEWorker is the subset of the ICE worker the Handshaker needs to build offers.
type ICEWorker interface {
Credentials() Credentials
Close()
}
// Handshaker keeps the signaling protocol logic: building and sending offers
// and answers and tracking whether the remote peer supports ICE. Incoming
// message processing is driven by the Conn event loop.
type Handshaker struct {
mu sync.Mutex
log *log.Entry
config Config
signaler *Signaler
ice ICEWorker
relayManager *relayClient.Manager
// remoteICESupported tracks whether the remote peer includes ICE credentials in its offers/answers.
// When false, the local side skips ICE dispatch and suppresses ICE credentials in responses.
remoteICESupported atomic.Bool
}
func NewHandshaker(log *log.Entry, config Config, signaler *Signaler, ice ICEWorker, relayManager *relayClient.Manager) *Handshaker {
h := &Handshaker{
log: log,
config: config,
signaler: signaler,
ice: ice,
relayManager: relayManager,
}
// assume remote supports ICE until we learn otherwise from received offers
h.remoteICESupported.Store(ice != nil)
return h
}
func (h *Handshaker) RemoteICESupported() bool {
return h.remoteICESupported.Load()
}
func (h *Handshaker) SendOffer() error {
h.mu.Lock()
defer h.mu.Unlock()
return h.sendOffer()
}
func (h *Handshaker) SendAnswer() error {
h.mu.Lock()
defer h.mu.Unlock()
return h.sendAnswer()
}
// sendOffer prepares local user credentials and signals them to the remote peer
func (h *Handshaker) sendOffer() error {
if !h.signaler.Ready() {
return ErrSignalIsNotReady
}
offer := h.buildOfferAnswer()
h.log.Debugf("sending offer with serial: %s", offer.SessionIDString())
return h.signaler.SignalOffer(offer, h.config.Key)
}
func (h *Handshaker) sendAnswer() error {
answer := h.buildOfferAnswer()
h.log.Debugf("sending answer with serial: %s", answer.SessionIDString())
return h.signaler.SignalAnswer(answer, h.config.Key)
}
func (h *Handshaker) buildOfferAnswer() OfferAnswer {
answer := OfferAnswer{
WgListenPort: h.config.LocalWgPort,
Version: version.NetbirdVersion(),
RosenpassPubKey: h.config.RosenpassPubKey,
RosenpassAddr: h.config.RosenpassAddr,
}
if h.ice != nil && h.RemoteICESupported() {
creds := h.ice.Credentials()
answer.IceCredentials = IceCredentials{creds.UFrag, creds.Pwd}
sid := creds.SessionID
answer.SessionID = &sid
}
if addr, ip, err := h.relayManager.RelayInstanceAddress(); err == nil {
answer.RelaySrvAddress = addr
answer.RelaySrvIP = ip
}
return answer
}
// UpdateRemoteICEState refreshes the remote ICE support flag from a received
// offer or answer and closes the ICE worker when the remote peer stopped
// sending ICE credentials. Runs on the Conn event loop.
func (h *Handshaker) UpdateRemoteICEState(offer *OfferAnswer) {
hasICE := offer.HasICECredentials()
prev := h.remoteICESupported.Swap(hasICE)
if prev != hasICE {
if hasICE {
h.log.Infof("remote peer started sending ICE credentials")
} else {
h.log.Infof("remote peer stopped sending ICE credentials")
if h.ice != nil {
h.ice.Close()
}
}
}
}

View File

@@ -1,4 +1,4 @@
package peer
package signaling
import (
"github.com/pion/ice/v4"

View File

@@ -1,4 +1,4 @@
package peer
package state_dump
import (
"context"
@@ -6,11 +6,13 @@ import (
"time"
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/client/internal/peer/status"
)
type stateDump struct {
type StateDump struct {
log *log.Entry
status *Status
status *status.Recorder
key string
sentOffer int
@@ -26,15 +28,15 @@ type stateDump struct {
mu sync.Mutex
}
func newStateDump(key string, log *log.Entry, statusRecorder *Status) *stateDump {
return &stateDump{
func NewStateDump(key string, log *log.Entry, statusRecorder *status.Recorder) *StateDump {
return &StateDump{
log: log,
status: statusRecorder,
key: key,
}
}
func (s *stateDump) Start(ctx context.Context) {
func (s *StateDump) Start(ctx context.Context) {
ticker := time.NewTicker(10 * time.Minute)
defer ticker.Stop()
@@ -48,25 +50,25 @@ func (s *stateDump) Start(ctx context.Context) {
}
}
func (s *stateDump) RemoteOffer() {
func (s *StateDump) RemoteOffer() {
s.mu.Lock()
defer s.mu.Unlock()
s.remoteOffer++
}
func (s *stateDump) RemoteCandidate() {
func (s *StateDump) RemoteCandidate() {
s.mu.Lock()
defer s.mu.Unlock()
s.remoteCandidate++
}
func (s *stateDump) SendOffer() {
func (s *StateDump) SendOffer() {
s.mu.Lock()
defer s.mu.Unlock()
s.sentOffer++
}
func (s *stateDump) dumpState() {
func (s *StateDump) dumpState() {
s.mu.Lock()
defer s.mu.Unlock()
@@ -80,41 +82,41 @@ func (s *stateDump) dumpState() {
status, s.sentOffer, s.remoteOffer, s.remoteAnswer, s.remoteCandidate, s.p2pConnected, s.switchToRelay, s.wgCheckSuccess, s.relayConnected, s.localProxies)
}
func (s *stateDump) RemoteAnswer() {
func (s *StateDump) RemoteAnswer() {
s.mu.Lock()
defer s.mu.Unlock()
s.remoteAnswer++
}
func (s *stateDump) P2PConnected() {
func (s *StateDump) P2PConnected() {
s.mu.Lock()
defer s.mu.Unlock()
s.p2pConnected++
}
func (s *stateDump) SwitchToRelay() {
func (s *StateDump) SwitchToRelay() {
s.mu.Lock()
defer s.mu.Unlock()
s.switchToRelay++
}
func (s *stateDump) WGcheckSuccess() {
func (s *StateDump) WGcheckSuccess() {
s.mu.Lock()
defer s.mu.Unlock()
s.wgCheckSuccess++
}
func (s *stateDump) RelayConnected() {
func (s *StateDump) RelayConnected() {
s.mu.Lock()
defer s.mu.Unlock()
s.relayConnected++
}
func (s *stateDump) NewLocalProxy() {
func (s *StateDump) NewLocalProxy() {
s.mu.Lock()
defer s.mu.Unlock()

View File

@@ -0,0 +1,31 @@
package status
import (
log "github.com/sirupsen/logrus"
)
const (
// StatusIdle indicate the peer is in disconnected state
StatusIdle ConnStatus = iota
// StatusConnecting indicate the peer is in connecting state
StatusConnecting
// StatusConnected indicate the peer is in connected state
StatusConnected
)
// ConnStatus describe the status of a peer's connection
type ConnStatus int32
func (s ConnStatus) String() string {
switch s {
case StatusConnecting:
return "Connecting"
case StatusConnected:
return "Connected"
case StatusIdle:
return "Idle"
default:
log.Errorf("unknown status: %d", s)
return "INVALID_PEER_CONNECTION_STATUS"
}
}

View File

@@ -1,4 +1,4 @@
package peer
package status
import (
"testing"

View File

@@ -0,0 +1,48 @@
package status
import (
"slices"
"sync"
"github.com/netbirdio/netbird/client/proto"
)
type EventQueue struct {
maxSize int
events []*proto.SystemEvent
mutex sync.RWMutex
}
func NewEventQueue(size int) *EventQueue {
return &EventQueue{
maxSize: size,
events: make([]*proto.SystemEvent, 0, size),
}
}
func (q *EventQueue) Add(event *proto.SystemEvent) {
q.mutex.Lock()
defer q.mutex.Unlock()
q.events = append(q.events, event)
if len(q.events) > q.maxSize {
q.events = q.events[len(q.events)-q.maxSize:]
}
}
func (q *EventQueue) GetAll() []*proto.SystemEvent {
q.mutex.RLock()
defer q.mutex.RUnlock()
return slices.Clone(q.events)
}
type EventSubscription struct {
id string
events chan *proto.SystemEvent
}
func (s *EventSubscription) Events() <-chan *proto.SystemEvent {
return s.events
}

View File

@@ -0,0 +1,122 @@
package status
import (
"golang.org/x/exp/maps"
"google.golang.org/protobuf/types/known/durationpb"
"google.golang.org/protobuf/types/known/timestamppb"
"github.com/netbirdio/netbird/client/internal/relay"
"github.com/netbirdio/netbird/client/proto"
)
// FullStatus contains the full state held by the Recorder instance
type FullStatus struct {
Peers []State
ManagementState ManagementState
SignalState SignalState
LocalPeerState LocalPeerState
RosenpassState RosenpassState
Relays []relay.ProbeResult
NSGroupStates []NSGroupState
NumOfForwardingRules int
LazyConnectionEnabled bool
Events []*proto.SystemEvent
}
// ToProto converts FullStatus to proto.FullStatus.
func (fs FullStatus) ToProto() *proto.FullStatus {
pbFullStatus := proto.FullStatus{
ManagementState: &proto.ManagementState{},
SignalState: &proto.SignalState{},
LocalPeerState: &proto.LocalPeerState{},
Peers: []*proto.PeerState{},
}
pbFullStatus.ManagementState.URL = fs.ManagementState.URL
pbFullStatus.ManagementState.Connected = fs.ManagementState.Connected
if err := fs.ManagementState.Error; err != nil {
pbFullStatus.ManagementState.Error = err.Error()
}
pbFullStatus.SignalState.URL = fs.SignalState.URL
pbFullStatus.SignalState.Connected = fs.SignalState.Connected
if err := fs.SignalState.Error; err != nil {
pbFullStatus.SignalState.Error = err.Error()
}
pbFullStatus.LocalPeerState.IP = fs.LocalPeerState.IP
pbFullStatus.LocalPeerState.Ipv6 = fs.LocalPeerState.IPv6
pbFullStatus.LocalPeerState.PubKey = fs.LocalPeerState.PubKey
pbFullStatus.LocalPeerState.KernelInterface = fs.LocalPeerState.KernelInterface
pbFullStatus.LocalPeerState.Fqdn = fs.LocalPeerState.FQDN
pbFullStatus.LocalPeerState.WgPort = int32(fs.LocalPeerState.WgPort)
pbFullStatus.LocalPeerState.RosenpassPermissive = fs.RosenpassState.Permissive
pbFullStatus.LocalPeerState.RosenpassEnabled = fs.RosenpassState.Enabled
pbFullStatus.NumberOfForwardingRules = int32(fs.NumOfForwardingRules)
pbFullStatus.LazyConnectionEnabled = fs.LazyConnectionEnabled
pbFullStatus.LocalPeerState.Networks = maps.Keys(fs.LocalPeerState.Routes)
for _, peerState := range fs.Peers {
networks := maps.Keys(peerState.GetRoutes())
pbPeerState := &proto.PeerState{
IP: peerState.IP,
Ipv6: peerState.IPv6,
PubKey: peerState.PubKey,
ConnStatus: peerState.ConnStatus.String(),
ConnStatusUpdate: timestamppb.New(peerState.ConnStatusUpdate),
Relayed: peerState.Relayed,
LocalIceCandidateType: peerState.LocalIceCandidateType,
RemoteIceCandidateType: peerState.RemoteIceCandidateType,
LocalIceCandidateEndpoint: peerState.LocalIceCandidateEndpoint,
RemoteIceCandidateEndpoint: peerState.RemoteIceCandidateEndpoint,
RelayAddress: peerState.RelayServerAddress,
Fqdn: peerState.FQDN,
LastWireguardHandshake: timestamppb.New(peerState.LastWireguardHandshake),
BytesRx: peerState.BytesRx,
BytesTx: peerState.BytesTx,
RosenpassEnabled: peerState.RosenpassEnabled,
Networks: networks,
Latency: durationpb.New(peerState.Latency),
SshHostKey: peerState.SSHHostKey,
}
pbFullStatus.Peers = append(pbFullStatus.Peers, pbPeerState)
}
for _, relayState := range fs.Relays {
pbRelayState := &proto.RelayState{
URI: relayState.URI,
Available: relayState.Err == nil,
Transport: relayState.Transport,
}
if err := relayState.Err; err != nil {
pbRelayState.Error = err.Error()
}
pbFullStatus.Relays = append(pbFullStatus.Relays, pbRelayState)
}
for _, dnsState := range fs.NSGroupStates {
var err string
if dnsState.Error != nil {
err = dnsState.Error.Error()
}
var servers []string
for _, server := range dnsState.Servers {
servers = append(servers, server.String())
}
pbDnsState := &proto.NSGroupState{
Servers: servers,
Domains: dnsState.Domains,
Enabled: dnsState.Enabled,
Error: err,
}
pbFullStatus.DnsServers = append(pbFullStatus.DnsServers, pbDnsState)
}
pbFullStatus.Events = fs.Events
return &pbFullStatus
}

View File

@@ -1,4 +1,4 @@
package peer
package status
import (
"sync"
@@ -11,6 +11,16 @@ const (
stateDisconnecting
)
// Listener is a callback type about the NetBird network connection state
type Listener interface {
OnConnected()
OnDisconnected()
OnConnecting()
OnDisconnecting()
OnAddressChanged(string, string)
OnPeersListChanged(int)
}
type notifier struct {
serverStateLock sync.Mutex
listenersLock sync.Mutex

View File

@@ -1,4 +1,4 @@
package peer
package status
import (
"sync"

View File

@@ -0,0 +1,63 @@
package status
import (
"sync"
"time"
"golang.org/x/exp/maps"
)
// State contains the latest state of a peer
type State struct {
Mux *sync.RWMutex
IP string
IPv6 string
PubKey string
FQDN string
ConnStatus ConnStatus
ConnStatusUpdate time.Time
Relayed bool
LocalIceCandidateType string
RemoteIceCandidateType string
LocalIceCandidateEndpoint string
RemoteIceCandidateEndpoint string
RelayServerAddress string
LastWireguardHandshake time.Time
BytesTx int64
BytesRx int64
Latency time.Duration
RosenpassEnabled bool
SSHHostKey []byte
routes map[string]struct{}
}
// AddRoute add a single route to routes map
func (s *State) AddRoute(network string) {
s.Mux.Lock()
defer s.Mux.Unlock()
if s.routes == nil {
s.routes = make(map[string]struct{})
}
s.routes[network] = struct{}{}
}
// SetRoutes set state routes
func (s *State) SetRoutes(routes map[string]struct{}) {
s.Mux.Lock()
defer s.Mux.Unlock()
s.routes = routes
}
// DeleteRoute removes a route from the network amp
func (s *State) DeleteRoute(network string) {
s.Mux.Lock()
defer s.Mux.Unlock()
delete(s.routes, network)
}
// GetRoutes return routes map
func (s *State) GetRoutes() map[string]struct{} {
s.Mux.RLock()
defer s.Mux.RUnlock()
return maps.Clone(s.routes)
}

View File

@@ -1,4 +1,4 @@
package peer
package status
import (
"context"

View File

@@ -1,4 +1,4 @@
package peer
package status
import (
"net/netip"

View File

@@ -0,0 +1,36 @@
package peer
import "github.com/netbirdio/netbird/client/internal/peer/status"
// Transitional aliases re-exporting the peer status recorder from its own
// package. Callers are being migrated to reference the status package
// directly; these aliases will be removed once the migration completes.
type (
Status = status.Recorder
State = status.State
ConnStatus = status.ConnStatus
FullStatus = status.FullStatus
RouterState = status.RouterState
LocalPeerState = status.LocalPeerState
SignalState = status.SignalState
ManagementState = status.ManagementState
RosenpassState = status.RosenpassState
NSGroupState = status.NSGroupState
ResolvedDomainInfo = status.ResolvedDomainInfo
StatusChangeSubscription = status.StatusChangeSubscription
EventQueue = status.EventQueue
EventSubscription = status.EventSubscription
WGIfaceStatus = status.WGIfaceStatus
Listener = status.Listener
EventListener = status.EventListener
)
const (
StatusIdle = status.StatusIdle
StatusConnecting = status.StatusConnecting
StatusConnected = status.StatusConnected
)
var (
NewRecorder = status.NewRecorder
)

View File

@@ -1,4 +1,4 @@
package peer
package wg_watcher
import (
"context"
@@ -9,6 +9,7 @@ import (
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/client/iface/configurer"
"github.com/netbirdio/netbird/client/internal/peer/state_dump"
)
const (
@@ -28,7 +29,7 @@ type WGWatcher struct {
log *log.Entry
wgIfaceStater WGInterfaceStater
peerKey string
stateDump *stateDump
stateDump *state_dump.StateDump
enabled bool
muEnabled sync.Mutex
@@ -38,7 +39,7 @@ type WGWatcher struct {
resetCh chan struct{}
}
func NewWGWatcher(log *log.Entry, wgIfaceStater WGInterfaceStater, peerKey string, stateDump *stateDump) *WGWatcher {
func NewWGWatcher(log *log.Entry, wgIfaceStater WGInterfaceStater, peerKey string, stateDump *state_dump.StateDump) *WGWatcher {
return &WGWatcher{
log: log,
wgIfaceStater: wgIfaceStater,

View File

@@ -1,4 +1,4 @@
package peer
package wg_watcher
import (
"context"
@@ -10,6 +10,8 @@ import (
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/client/iface/configurer"
"github.com/netbirdio/netbird/client/internal/peer/state_dump"
"github.com/netbirdio/netbird/client/internal/peer/status"
)
type MocWgIface struct {
@@ -57,7 +59,7 @@ func TestWGWatcher_CheckSuccessCallback(t *testing.T) {
// platforms with coarse clock resolution (Windows), where two time.Now() calls
// microseconds apart can return the same instant and read as a timed-out handshake.
stats := &mockHandshakeStats{handshake: time.Now().Add(-time.Hour)}
watcher := NewWGWatcher(mlog, stats, "", newStateDump("peer", mlog, &Status{}))
watcher := NewWGWatcher(mlog, stats, "", state_dump.NewStateDump("peer", mlog, &status.Recorder{}))
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
@@ -66,14 +68,18 @@ func TestWGWatcher_CheckSuccessCallback(t *testing.T) {
firstHandshake := make(chan struct{}, 1)
checkSuccess := make(chan struct{}, 1)
go watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {
firstHandshake <- struct{}{}
}, func() {
select {
case checkSuccess <- struct{}{}:
default:
}
})
watcherDone := make(chan struct{})
go func() {
defer close(watcherDone)
watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {
firstHandshake <- struct{}{}
}, func() {
select {
case checkSuccess <- struct{}{}:
default:
}
})
}()
stats.advance()
@@ -88,6 +94,11 @@ func TestWGWatcher_CheckSuccessCallback(t *testing.T) {
t.Errorf("first-handshake callback must not fire for a non-zero baseline")
default:
}
// Wait for the watcher goroutine to exit so it cannot race with other
// tests mutating the package-level check timing variables.
cancel()
<-watcherDone
}
func TestWGWatcher_EnableWgWatcher(t *testing.T) {
@@ -96,7 +107,7 @@ func TestWGWatcher_EnableWgWatcher(t *testing.T) {
mlog := log.WithField("peer", "tet")
mocWgIface := &MocWgIface{}
watcher := NewWGWatcher(mlog, mocWgIface, "", newStateDump("peer", mlog, &Status{}))
watcher := NewWGWatcher(mlog, mocWgIface, "", state_dump.NewStateDump("peer", mlog, &status.Recorder{}))
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
@@ -129,7 +140,7 @@ func TestWGWatcher_ReEnable(t *testing.T) {
mlog := log.WithField("peer", "tet")
mocWgIface := &MocWgIface{}
watcher := NewWGWatcher(mlog, mocWgIface, "", newStateDump("peer", mlog, &Status{}))
watcher := NewWGWatcher(mlog, mocWgIface, "", state_dump.NewStateDump("peer", mlog, &status.Recorder{}))
ctx, cancel := context.WithCancel(context.Background())
ok := watcher.PrepareInitialHandshake()

View File

@@ -1,4 +1,4 @@
package conntype
package worker
import (
"fmt"

View File

@@ -1,4 +1,4 @@
package peer
package worker
import (
"context"
@@ -13,8 +13,9 @@ import (
"github.com/netbirdio/netbird/client/iface"
"github.com/netbirdio/netbird/client/iface/udpmux"
"github.com/netbirdio/netbird/client/internal/peer/conntype"
icemaker "github.com/netbirdio/netbird/client/internal/peer/ice"
"github.com/netbirdio/netbird/client/internal/peer/signaling"
"github.com/netbirdio/netbird/client/internal/peer/status"
"github.com/netbirdio/netbird/client/internal/portforward"
"github.com/netbirdio/netbird/client/internal/stdnet"
"github.com/netbirdio/netbird/route"
@@ -32,57 +33,68 @@ type ICEConnInfo struct {
RelayedOnLocal bool
}
type WorkerICE struct {
ctx context.Context
log *log.Entry
config ConnConfig
conn *Conn
signaler *Signaler
iFaceDiscover stdnet.ExternalIFaceDiscover
statusRecorder *Status
hasRelayOnLocally bool
type ICEDependencies struct {
Signaler *signaling.Signaler
IFaceDiscover stdnet.ExternalIFaceDiscover
StatusRecorder *status.Recorder
PortForwardManager *portforward.Manager
}
type ICE struct {
log *log.Entry
key string
iceConfig icemaker.Config
isController bool
onConnReady func(priority ConnPriority, iceConnInfo ICEConnInfo)
onStatusDisconnect func(sessionChanged bool)
signaler *signaling.Signaler
iFaceDiscover stdnet.ExternalIFaceDiscover
statusRecorder *status.Recorder
portForwardManager *portforward.Manager
hasRelayOnLocally bool
agent *icemaker.ThreadSafeAgent
agentDialerCancel context.CancelFunc
agentConnecting bool // while it is true, drop all incoming offers
lastSuccess time.Time // with this avoid the too frequent ICE agent recreation
// connectedAgent is the agent whose connection was last reported ready; guarded by muxAgent
connectedAgent *icemaker.ThreadSafeAgent
// remoteSessionID represents the peer's session identifier from the latest remote offer.
remoteSessionID ICESessionID
remoteSessionID icemaker.SessionID
// sessionID is used to track the current session ID of the ICE agent
// increase by one when disconnecting the agent
// with it the remote peer can discard the already deprecated offer/answer
// Without it the remote peer may recreate a workable ICE connection
sessionID ICESessionID
sessionID icemaker.SessionID
remoteSessionChanged bool
muxAgent sync.Mutex
localUfrag string
localPwd string
// we record the last known state of the ICE agent to avoid duplicate on disconnected events
lastKnownState ice.ConnectionState
// portForwardAttempted tracks if we've already tried port forwarding this session
portForwardAttempted bool
}
func NewWorkerICE(ctx context.Context, log *log.Entry, config ConnConfig, conn *Conn, signaler *Signaler, ifaceDiscover stdnet.ExternalIFaceDiscover, statusRecorder *Status, hasRelayOnLocally bool) (*WorkerICE, error) {
sessionID, err := NewICESessionID()
func NewICE(log *log.Entry, key string, iceConfig icemaker.Config, isController bool, onConnReady func(ConnPriority, ICEConnInfo), onStatusDisconnect func(bool), services ICEDependencies, hasRelayOnLocally bool) (*ICE, error) {
sessionID, err := icemaker.NewSessionID()
if err != nil {
return nil, err
}
w := &WorkerICE{
ctx: ctx,
log: log,
config: config,
conn: conn,
signaler: signaler,
iFaceDiscover: ifaceDiscover,
statusRecorder: statusRecorder,
hasRelayOnLocally: hasRelayOnLocally,
lastKnownState: ice.ConnectionStateDisconnected,
sessionID: sessionID,
w := &ICE{
log: log,
key: key,
iceConfig: iceConfig,
isController: isController,
onConnReady: onConnReady,
onStatusDisconnect: onStatusDisconnect,
signaler: services.Signaler,
iFaceDiscover: services.IFaceDiscover,
statusRecorder: services.StatusRecorder,
portForwardManager: services.PortForwardManager,
hasRelayOnLocally: hasRelayOnLocally,
sessionID: sessionID,
}
localUfrag, localPwd, err := icemaker.GenerateICECredentials()
@@ -94,7 +106,7 @@ func NewWorkerICE(ctx context.Context, log *log.Entry, config ConnConfig, conn *
return w, nil
}
func (w *WorkerICE) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
func (w *ICE) OnNewOffer(ctx context.Context, remoteOfferAnswer *signaling.OfferAnswer) {
w.log.Debugf("OnNewOffer for ICE, serial: %s", remoteOfferAnswer.SessionIDString())
w.muxAgent.Lock()
defer w.muxAgent.Unlock()
@@ -118,7 +130,7 @@ func (w *WorkerICE) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
}
}
sessionID, err := NewICESessionID()
sessionID, err := icemaker.NewSessionID()
if err != nil {
w.log.Errorf("failed to create new session ID: %s", err)
}
@@ -136,8 +148,8 @@ func (w *WorkerICE) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
if remoteOfferAnswer.SessionID != nil {
w.log.Debugf("recreate ICE agent: %s / %s", w.sessionID, *remoteOfferAnswer.SessionID)
}
dialerCtx, dialerCancel := context.WithCancel(w.ctx)
agent, err := w.reCreateAgent(dialerCancel, preferredCandidateTypes)
dialerCtx, dialerCancel := context.WithCancel(ctx)
agent, err := w.reCreateAgent(ctx, dialerCancel, preferredCandidateTypes)
if err != nil {
w.log.Errorf("failed to recreate ICE Agent: %s", err)
return
@@ -151,14 +163,14 @@ func (w *WorkerICE) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
w.remoteSessionID = ""
}
go w.connect(dialerCtx, agent, remoteOfferAnswer)
go w.connect(dialerCtx, dialerCancel, agent, remoteOfferAnswer)
}
// OnRemoteCandidate Handles ICE connection Candidate provided by the remote peer.
func (w *WorkerICE) OnRemoteCandidate(candidate ice.Candidate, haRoutes route.HAMap) {
func (w *ICE) OnRemoteCandidate(candidate ice.Candidate, haRoutes route.HAMap) {
w.muxAgent.Lock()
defer w.muxAgent.Unlock()
w.log.Debugf("OnRemoteCandidate from peer %s -> %s", w.config.Key, candidate.String())
w.log.Debugf("OnRemoteCandidate from peer %s -> %s", w.key, candidate.String())
if w.agent == nil {
w.log.Warnf("ICE Agent is not initialized yet")
return
@@ -185,18 +197,24 @@ func (w *WorkerICE) OnRemoteCandidate(candidate ice.Candidate, haRoutes route.HA
}
}
func (w *WorkerICE) GetLocalUserCredentials() (frag string, pwd string) {
return w.localUfrag, w.localPwd
func (w *ICE) Credentials() signaling.Credentials {
w.muxAgent.Lock()
defer w.muxAgent.Unlock()
return signaling.Credentials{
UFrag: w.localUfrag,
Pwd: w.localPwd,
SessionID: w.sessionID,
}
}
func (w *WorkerICE) InProgress() bool {
func (w *ICE) InProgress() bool {
w.muxAgent.Lock()
defer w.muxAgent.Unlock()
return w.agentConnecting
}
func (w *WorkerICE) Close() {
func (w *ICE) Close() {
w.muxAgent.Lock()
defer w.muxAgent.Unlock()
@@ -212,10 +230,10 @@ func (w *WorkerICE) Close() {
w.agent = nil
}
func (w *WorkerICE) reCreateAgent(dialerCancel context.CancelFunc, candidates []ice.CandidateType) (*icemaker.ThreadSafeAgent, error) {
func (w *ICE) reCreateAgent(ctx context.Context, dialerCancel context.CancelFunc, candidates []ice.CandidateType) (*icemaker.ThreadSafeAgent, error) {
w.portForwardAttempted = false
agent, err := icemaker.NewAgent(w.ctx, w.iFaceDiscover, w.config.ICEConfig, candidates, w.localUfrag, w.localPwd)
agent, err := icemaker.NewAgent(ctx, w.iFaceDiscover, w.iceConfig, candidates, w.localUfrag, w.localPwd)
if err != nil {
return nil, fmt.Errorf("create agent: %w", err)
}
@@ -237,7 +255,7 @@ func (w *WorkerICE) reCreateAgent(dialerCancel context.CancelFunc, candidates []
return agent, nil
}
func (w *WorkerICE) SessionID() ICESessionID {
func (w *ICE) getSessionID() icemaker.SessionID {
w.muxAgent.Lock()
defer w.muxAgent.Unlock()
@@ -247,11 +265,11 @@ func (w *WorkerICE) SessionID() ICESessionID {
// will block until connection succeeded
// but it won't release if ICE Agent went into Disconnected or Failed state,
// so we have to cancel it with the provided context once agent detected a broken connection
func (w *WorkerICE) connect(ctx context.Context, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *OfferAnswer) {
func (w *ICE) connect(ctx context.Context, dialerCancel context.CancelFunc, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *signaling.OfferAnswer) {
w.log.Debugf("gather candidates")
if err := agent.GatherCandidates(); err != nil {
w.log.Warnf("failed to gather candidates: %s", err)
w.closeAgent(agent, w.agentDialerCancel)
w.closeAgent(agent, dialerCancel)
return
}
@@ -259,19 +277,19 @@ func (w *WorkerICE) connect(ctx context.Context, agent *icemaker.ThreadSafeAgent
remoteConn, err := w.turnAgentDial(ctx, agent, remoteOfferAnswer)
if err != nil {
w.log.Debugf("failed to dial the remote peer: %s", err)
w.closeAgent(agent, w.agentDialerCancel)
w.closeAgent(agent, dialerCancel)
return
}
w.log.Debugf("agent dial succeeded")
pair, err := agent.GetSelectedCandidatePair()
if err != nil {
w.closeAgent(agent, w.agentDialerCancel)
w.closeAgent(agent, dialerCancel)
return
}
if pair == nil {
w.log.Warnf("selected candidate pair is nil, cannot proceed")
w.closeAgent(agent, w.agentDialerCancel)
w.closeAgent(agent, dialerCancel)
return
}
@@ -299,17 +317,22 @@ func (w *WorkerICE) connect(ctx context.Context, agent *icemaker.ThreadSafeAgent
}
w.log.Debugf("on ICE conn is ready to use")
w.log.Infof("connection succeeded with offer session: %s", remoteOfferAnswer.SessionIDString())
w.muxAgent.Lock()
if w.agent != agent {
w.muxAgent.Unlock()
w.log.Debugf("agent has been replaced during connect, dropping obsolete connection")
return
}
w.agentConnecting = false
w.lastSuccess = time.Now()
w.connectedAgent = agent
w.muxAgent.Unlock()
// todo: the potential problem is a race between the onConnectionStateChange
w.conn.onICEConnectionIsReady(selectedPriority(pair), ci)
w.log.Infof("connection succeeded with offer session: %s", remoteOfferAnswer.SessionIDString())
w.onConnReady(selectedPriority(pair), ci)
}
func (w *WorkerICE) closeAgent(agent *icemaker.ThreadSafeAgent, cancel context.CancelFunc) bool {
func (w *ICE) closeAgent(agent *icemaker.ThreadSafeAgent, cancel context.CancelFunc) bool {
cancel()
if err := agent.Close(); err != nil {
w.log.Warnf("failed to close ICE agent: %s", err)
@@ -323,7 +346,7 @@ func (w *WorkerICE) closeAgent(agent *icemaker.ThreadSafeAgent, cancel context.C
if w.agent == agent {
// consider to remove from here and move to the OnNewOffer
sessionID, err := NewICESessionID()
sessionID, err := icemaker.NewSessionID()
if err != nil {
w.log.Errorf("failed to create new session ID: %s", err)
}
@@ -335,7 +358,7 @@ func (w *WorkerICE) closeAgent(agent *icemaker.ThreadSafeAgent, cancel context.C
return sessionChanged
}
func (w *WorkerICE) punchRemoteWGPort(pair *ice.CandidatePair, remoteWgPort int) {
func (w *ICE) punchRemoteWGPort(pair *ice.CandidatePair, remoteWgPort int) {
// wait local endpoint configuration
time.Sleep(time.Second)
addr, err := net.ResolveUDPAddr("udp", net.JoinHostPort(pair.Remote.Address(), strconv.Itoa(remoteWgPort)))
@@ -344,7 +367,7 @@ func (w *WorkerICE) punchRemoteWGPort(pair *ice.CandidatePair, remoteWgPort int)
return
}
mux, ok := w.config.ICEConfig.UDPMuxSrflx.(*udpmux.UniversalUDPMuxDefault)
mux, ok := w.iceConfig.UDPMuxSrflx.(*udpmux.UniversalUDPMuxDefault)
if !ok {
w.log.Warn("invalid udp mux conversion")
return
@@ -357,7 +380,7 @@ func (w *WorkerICE) punchRemoteWGPort(pair *ice.CandidatePair, remoteWgPort int)
// onICECandidate is a callback attached to an ICE Agent to receive new local connection candidates
// and then signals them to the remote peer
func (w *WorkerICE) onICECandidate(candidate ice.Candidate) {
func (w *ICE) onICECandidate(candidate ice.Candidate) {
// nil means candidate gathering has been ended
if candidate == nil {
return
@@ -366,9 +389,9 @@ func (w *WorkerICE) onICECandidate(candidate ice.Candidate) {
// TODO: reported port is incorrect for CandidateTypeHost, makes understanding ICE use via logs confusing as port is ignored
w.log.Debugf("discovered local candidate %s", candidate.String())
go func() {
err := w.signaler.SignalICECandidate(candidate, w.config.Key)
err := w.signaler.SignalICECandidate(candidate, w.key)
if err != nil {
w.log.Errorf("failed signaling candidate to the remote peer %s %s", w.config.Key, err)
w.log.Errorf("failed signaling candidate to the remote peer %s %s", w.key, err)
}
}()
@@ -378,8 +401,8 @@ func (w *WorkerICE) onICECandidate(candidate ice.Candidate) {
}
// injectPortForwardedCandidate signals an additional candidate using the pre-created port mapping.
func (w *WorkerICE) injectPortForwardedCandidate(srflxCandidate ice.Candidate) {
pfManager := w.conn.portForwardManager
func (w *ICE) injectPortForwardedCandidate(srflxCandidate ice.Candidate) {
pfManager := w.portForwardManager
if pfManager == nil {
return
}
@@ -407,7 +430,7 @@ func (w *WorkerICE) injectPortForwardedCandidate(srflxCandidate ice.Candidate) {
forwardedCandidate.String(), mapping.InternalPort, mapping.ExternalPort, mapping.NATType, forwardedCandidate.Priority())
go func() {
if err := w.signaler.SignalICECandidate(forwardedCandidate, w.config.Key); err != nil {
if err := w.signaler.SignalICECandidate(forwardedCandidate, w.key); err != nil {
w.log.Errorf("signal port-forwarded candidate: %v", err)
}
}()
@@ -415,7 +438,7 @@ func (w *WorkerICE) injectPortForwardedCandidate(srflxCandidate ice.Candidate) {
// createForwardedCandidate creates a new server reflexive candidate with the forwarded port.
// It uses the NAT gateway's external IP with the forwarded port.
func (w *WorkerICE) createForwardedCandidate(srflxCandidate ice.Candidate, mapping *portforward.Mapping) (ice.Candidate, error) {
func (w *ICE) createForwardedCandidate(srflxCandidate ice.Candidate, mapping *portforward.Mapping) (ice.Candidate, error) {
var externalIP string
if mapping.ExternalIP != nil && !mapping.ExternalIP.IsUnspecified() {
externalIP = mapping.ExternalIP.String()
@@ -460,9 +483,9 @@ func (w *WorkerICE) createForwardedCandidate(srflxCandidate ice.Candidate, mappi
return candidate, nil
}
func (w *WorkerICE) onICESelectedCandidatePair(agent *icemaker.ThreadSafeAgent, c1, c2 ice.Candidate) {
func (w *ICE) onICESelectedCandidatePair(agent *icemaker.ThreadSafeAgent, c1, c2 ice.Candidate) {
w.log.Debugf("selected candidate pair [local <-> remote] -> [%s <-> %s], peer %s", c1.String(), c2.String(),
w.config.Key)
w.key)
pairStat, ok := agent.GetSelectedCandidatePairStats()
if !ok {
@@ -471,14 +494,14 @@ func (w *WorkerICE) onICESelectedCandidatePair(agent *icemaker.ThreadSafeAgent,
}
duration := time.Duration(pairStat.CurrentRoundTripTime * float64(time.Second))
if err := w.statusRecorder.UpdateLatency(w.config.Key, duration); err != nil {
if err := w.statusRecorder.UpdateLatency(w.key, duration); err != nil {
w.log.Debugf("failed to update latency for peer: %s", err)
return
}
}
func (w *WorkerICE) logSuccessfulPaths(agent *icemaker.ThreadSafeAgent) {
sessionID := w.SessionID()
func (w *ICE) logSuccessfulPaths(agent *icemaker.ThreadSafeAgent) {
sessionID := w.getSessionID()
stats := agent.GetCandidatePairsStats()
localCandidates, _ := agent.GetLocalCandidates()
remoteCandidates, _ := agent.GetRemoteCandidates()
@@ -508,32 +531,44 @@ func (w *WorkerICE) logSuccessfulPaths(agent *icemaker.ThreadSafeAgent) {
}
}
func (w *WorkerICE) onConnectionStateChange(agent *icemaker.ThreadSafeAgent, dialerCancel context.CancelFunc) func(ice.ConnectionState) {
func (w *ICE) onConnectionStateChange(agent *icemaker.ThreadSafeAgent, dialerCancel context.CancelFunc) func(ice.ConnectionState) {
// per-agent state; pion delivers callbacks of one agent sequentially
var connected bool
return func(state ice.ConnectionState) {
w.log.Debugf("ICE ConnectionState has changed to %s", state.String())
switch state {
case ice.ConnectionStateConnected:
w.lastKnownState = ice.ConnectionStateConnected
connected = true
w.logSuccessfulPaths(agent)
return
case ice.ConnectionStateFailed, ice.ConnectionStateDisconnected, ice.ConnectionStateClosed:
// ice.ConnectionStateClosed happens when we recreate the agent. For the P2P to TURN switch important to
// notify the conn.onICEStateDisconnected changes to update the current used priority
sessionChanged := w.closeAgent(agent, dialerCancel)
if w.lastKnownState == ice.ConnectionStateConnected {
w.lastKnownState = ice.ConnectionStateDisconnected
w.conn.onICEStateDisconnected(sessionChanged)
if !connected {
return
}
default:
return
connected = false
w.muxAgent.Lock()
stale := w.connectedAgent != agent
if !stale {
w.connectedAgent = nil
}
w.muxAgent.Unlock()
if stale {
w.log.Debugf("suppress disconnected event of replaced ICE agent")
return
}
w.onStatusDisconnect(sessionChanged)
}
}
}
func (w *WorkerICE) turnAgentDial(ctx context.Context, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *OfferAnswer) (*ice.Conn, error) {
if isController(w.config) {
func (w *ICE) turnAgentDial(ctx context.Context, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *signaling.OfferAnswer) (*ice.Conn, error) {
if w.isController {
return agent.Dial(ctx, remoteOfferAnswer.IceCredentials.UFrag, remoteOfferAnswer.IceCredentials.Pwd)
} else {
return agent.Accept(ctx, remoteOfferAnswer.IceCredentials.UFrag, remoteOfferAnswer.IceCredentials.Pwd)
@@ -595,10 +630,10 @@ func isRelayed(pair *ice.CandidatePair) bool {
return false
}
func selectedPriority(pair *ice.CandidatePair) conntype.ConnPriority {
func selectedPriority(pair *ice.CandidatePair) ConnPriority {
if isRelayed(pair) {
return conntype.ICETurn
return ICETurn
} else {
return conntype.ICEP2P
return ICEP2P
}
}

View File

@@ -1,4 +1,4 @@
package peer
package worker
import (
"context"
@@ -10,22 +10,23 @@ import (
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/client/internal/peer/signaling"
relayClient "github.com/netbirdio/netbird/shared/relay/client"
)
type RelayConnInfo struct {
relayedConn net.Conn
rosenpassPubKey []byte
rosenpassAddr string
RelayedConn net.Conn
RosenpassPubKey []byte
RosenpassAddr string
}
type WorkerRelay struct {
peerCtx context.Context
log *log.Entry
isController bool
config ConnConfig
conn *Conn
relayManager *relayClient.Manager
log *log.Entry
key string
isController bool
onConnReady func(RelayConnInfo)
onDisconnected func()
relayManager *relayClient.Manager
relayedConn net.Conn
relayLock sync.Mutex
@@ -33,19 +34,19 @@ type WorkerRelay struct {
relaySupportedOnRemotePeer atomic.Bool
}
func NewWorkerRelay(ctx context.Context, log *log.Entry, ctrl bool, config ConnConfig, conn *Conn, relayManager *relayClient.Manager) *WorkerRelay {
func NewWorkerRelay(log *log.Entry, key string, isController bool, onConnReady func(RelayConnInfo), onDisconnected func(), relayManager *relayClient.Manager) *WorkerRelay {
r := &WorkerRelay{
peerCtx: ctx,
log: log,
isController: ctrl,
config: config,
conn: conn,
relayManager: relayManager,
log: log,
key: key,
isController: isController,
onConnReady: onConnReady,
onDisconnected: onDisconnected,
relayManager: relayManager,
}
return r
}
func (w *WorkerRelay) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
func (w *WorkerRelay) OnNewOffer(ctx context.Context, remoteOfferAnswer *signaling.OfferAnswer) {
if !w.isRelaySupported(remoteOfferAnswer) {
w.log.Infof("Relay is not supported by remote peer")
w.relaySupportedOnRemotePeer.Store(false)
@@ -66,7 +67,7 @@ func (w *WorkerRelay) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
serverIP = remoteOfferAnswer.RelaySrvIP
}
relayedConn, err := w.relayManager.OpenConn(w.peerCtx, srv, w.config.Key, serverIP)
relayedConn, err := w.relayManager.OpenConn(ctx, srv, w.key, serverIP)
if err != nil {
if errors.Is(err, relayClient.ErrConnAlreadyExists) {
w.log.Debugf("handled offer by reusing existing relay connection")
@@ -88,10 +89,10 @@ func (w *WorkerRelay) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
}
w.log.Debugf("peer conn opened via Relay: %s", srv)
go w.conn.onRelayConnectionIsReady(RelayConnInfo{
relayedConn: relayedConn,
rosenpassPubKey: remoteOfferAnswer.RosenpassPubKey,
rosenpassAddr: remoteOfferAnswer.RosenpassAddr,
w.onConnReady(RelayConnInfo{
RelayedConn: relayedConn,
RosenpassPubKey: remoteOfferAnswer.RosenpassPubKey,
RosenpassAddr: remoteOfferAnswer.RosenpassAddr,
})
}
@@ -119,7 +120,7 @@ func (w *WorkerRelay) CloseConn() {
}
}
func (w *WorkerRelay) isRelaySupported(answer *OfferAnswer) bool {
func (w *WorkerRelay) isRelaySupported(answer *signaling.OfferAnswer) bool {
if !w.relayManager.HasRelayAddress() {
return false
}
@@ -134,5 +135,5 @@ func (w *WorkerRelay) preferredRelayServer(myRelayAddress, remoteRelayAddress st
}
func (w *WorkerRelay) onRelayClientDisconnected() {
go w.conn.onRelayDisconnected()
w.onDisconnected()
}

View File

@@ -1,4 +1,4 @@
package worker
package peer
import (
"sync/atomic"
@@ -7,17 +7,17 @@ import (
)
const (
StatusDisconnected Status = iota
StatusConnected
WorkerStatusDisconnected WorkerStatus = iota
WorkerStatusConnected
)
type Status int32
type WorkerStatus int32
func (s Status) String() string {
func (s WorkerStatus) String() string {
switch s {
case StatusDisconnected:
case WorkerStatusDisconnected:
return "Disconnected"
case StatusConnected:
case WorkerStatusConnected:
return "Connected"
default:
log.Errorf("unknown status: %d", s)
@@ -37,16 +37,16 @@ func NewAtomicStatus() *AtomicWorkerStatus {
}
// Get returns the current connection status
func (acs *AtomicWorkerStatus) Get() Status {
return Status(acs.status.Load())
func (acs *AtomicWorkerStatus) Get() WorkerStatus {
return WorkerStatus(acs.status.Load())
}
func (acs *AtomicWorkerStatus) SetConnected() {
acs.status.Store(int32(StatusConnected))
acs.status.Store(int32(WorkerStatusConnected))
}
func (acs *AtomicWorkerStatus) SetDisconnected() {
acs.status.Store(int32(StatusDisconnected))
acs.status.Store(int32(WorkerStatusDisconnected))
}
// String returns the string representation of the current status