Reduce HandshakeManager complexity a little bit (#1701)
Some checks failed
gofmt / Run gofmt (push) Failing after 3s
smoke-extra / Run extra smoke tests (push) Failing after 3s
smoke / Run multi node smoke test (push) Failing after 2s
Build and test / Build all and test on ubuntu-linux (push) Failing after 3s
Build and test / Build and test on linux with boringcrypto (push) Failing after 2s
Build and test / Build and test on linux with pkcs11 (push) Failing after 2s
Build and test / Build and test on macos-latest (push) Has been cancelled
Build and test / Build and test on windows-latest (push) Has been cancelled

This commit is contained in:
Nate Brown
2026-05-01 13:21:38 -05:00
committed by GitHub
parent f141cebe8d
commit 33c2d7277c
3 changed files with 163 additions and 157 deletions

View File

@@ -23,7 +23,6 @@ const (
DefaultHandshakeTryInterval = time.Millisecond * 100 DefaultHandshakeTryInterval = time.Millisecond * 100
DefaultHandshakeRetries = 10 DefaultHandshakeRetries = 10
DefaultHandshakeTriggerBuffer = 64 DefaultHandshakeTriggerBuffer = 64
DefaultUseRelays = true
// maxCachedPackets is how many unsent packets we'll buffer per pending // maxCachedPackets is how many unsent packets we'll buffer per pending
// handshake before dropping further ones. // handshake before dropping further ones.
@@ -43,7 +42,6 @@ var (
tryInterval: DefaultHandshakeTryInterval, tryInterval: DefaultHandshakeTryInterval,
retries: DefaultHandshakeRetries, retries: DefaultHandshakeRetries,
triggerBuffer: DefaultHandshakeTriggerBuffer, triggerBuffer: DefaultHandshakeTriggerBuffer,
useRelays: DefaultUseRelays,
} }
) )
@@ -51,7 +49,6 @@ type HandshakeConfig struct {
tryInterval time.Duration tryInterval time.Duration
retries int64 retries int64
triggerBuffer int triggerBuffer int
useRelays bool
messageMetrics *MessageMetrics messageMetrics *MessageMetrics
} }
@@ -326,146 +323,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
) )
} }
if hm.config.useRelays && len(hostinfo.remotes.relays) > 0 { hm.f.relayManager.StartRelays(hm.f, vpnIp, hostinfo, stage0)
hostinfo.logger(hm.l).Info("Attempt to relay through hosts", "relays", hostinfo.remotes.relays)
// Send a RelayRequest to all known Relay IP's
for _, relay := range hostinfo.remotes.relays {
// Don't relay through the host I'm trying to connect to
if relay == vpnIp {
continue
}
// Don't relay to myself
if hm.f.myVpnAddrsTable.Contains(relay) {
continue
}
relayHostInfo := hm.mainHostMap.QueryVpnAddr(relay)
if relayHostInfo == nil || !relayHostInfo.remote.IsValid() {
hostinfo.logger(hm.l).Info("Establish tunnel to relay target", "relay", relay.String())
hm.f.Handshake(relay)
continue
}
// Check the relay HostInfo to see if we already established a relay through
existingRelay, ok := relayHostInfo.relayState.QueryRelayForByIp(vpnIp)
if !ok {
// No relays exist or requested yet.
if relayHostInfo.remote.IsValid() {
idx, err := AddRelay(hm.l, relayHostInfo, hm.mainHostMap, vpnIp, nil, TerminalType, Requested)
if err != nil {
hostinfo.logger(hm.l).Info("Failed to add relay to hostmap", "relay", relay.String(), "error", err)
}
m := NebulaControl{
Type: NebulaControl_CreateRelayRequest,
InitiatorRelayIndex: idx,
}
switch relayHostInfo.GetCert().Certificate.Version() {
case cert.Version1:
if !hm.f.myVpnAddrs[0].Is4() {
hostinfo.logger(hm.l).Error("can not establish v1 relay with a v6 network because the relay is not running a current nebula version")
continue
}
if !vpnIp.Is4() {
hostinfo.logger(hm.l).Error("can not establish v1 relay with a v6 remote network because the relay is not running a current nebula version")
continue
}
b := hm.f.myVpnAddrs[0].As4()
m.OldRelayFromAddr = binary.BigEndian.Uint32(b[:])
b = vpnIp.As4()
m.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
case cert.Version2:
m.RelayFromAddr = netAddrToProtoAddr(hm.f.myVpnAddrs[0])
m.RelayToAddr = netAddrToProtoAddr(vpnIp)
default:
hostinfo.logger(hm.l).Error("Unknown certificate version found while creating relay")
continue
}
msg, err := m.Marshal()
if err != nil {
hostinfo.logger(hm.l).Error("Failed to marshal Control message to create relay", "error", err)
} else {
hm.f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
hm.l.Info("send CreateRelayRequest",
"relayFrom", hm.f.myVpnAddrs[0],
"relayTo", vpnIp,
"initiatorRelayIndex", idx,
"relay", relay,
)
}
}
continue
}
switch existingRelay.State {
case Established:
hostinfo.logger(hm.l).Info("Send handshake via relay", "relay", relay.String())
hm.f.SendVia(relayHostInfo, existingRelay, hostinfo.HandshakePacket[handshakePacketStage0], make([]byte, 12), make([]byte, mtu), false)
case Disestablished:
// Mark this relay as 'requested'
relayHostInfo.relayState.UpdateRelayForByIpState(vpnIp, Requested)
fallthrough
case Requested:
hostinfo.logger(hm.l).Info("Re-send CreateRelay request", "relay", relay.String())
// Re-send the CreateRelay request, in case the previous one was lost.
m := NebulaControl{
Type: NebulaControl_CreateRelayRequest,
InitiatorRelayIndex: existingRelay.LocalIndex,
}
switch relayHostInfo.GetCert().Certificate.Version() {
case cert.Version1:
if !hm.f.myVpnAddrs[0].Is4() {
hostinfo.logger(hm.l).Error("can not establish v1 relay with a v6 network because the relay is not running a current nebula version")
continue
}
if !vpnIp.Is4() {
hostinfo.logger(hm.l).Error("can not establish v1 relay with a v6 remote network because the relay is not running a current nebula version")
continue
}
b := hm.f.myVpnAddrs[0].As4()
m.OldRelayFromAddr = binary.BigEndian.Uint32(b[:])
b = vpnIp.As4()
m.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
case cert.Version2:
m.RelayFromAddr = netAddrToProtoAddr(hm.f.myVpnAddrs[0])
m.RelayToAddr = netAddrToProtoAddr(vpnIp)
default:
hostinfo.logger(hm.l).Error("Unknown certificate version found while creating relay")
continue
}
msg, err := m.Marshal()
if err != nil {
hostinfo.logger(hm.l).Error("Failed to marshal Control message to create relay", "error", err)
} else {
// This must send over the hostinfo, not over hm.Hosts[ip]
hm.f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
hm.l.Info("send CreateRelayRequest",
"relayFrom", hm.f.myVpnAddrs[0],
"relayTo", vpnIp,
"initiatorRelayIndex", existingRelay.LocalIndex,
"relay", relay,
)
}
case PeerRequested:
// PeerRequested only occurs in Forwarding relays, not Terminal relays, and this is a Terminal relay case.
fallthrough
default:
hostinfo.logger(hm.l).Error("Relay unexpected state",
"vpnIp", vpnIp,
"state", existingRelay.State,
"relay", relay,
)
}
}
}
// If a lighthouse triggered this attempt then we are still in the timer wheel and do not need to re-add // If a lighthouse triggered this attempt then we are still in the timer wheel and do not need to re-add
if !lighthouseTriggered { if !lighthouseTriggered {

View File

@@ -184,14 +184,10 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
messageMetrics = newMessageMetricsOnlyRecvError() messageMetrics = newMessageMetricsOnlyRecvError()
} }
useRelays := c.GetBool("relay.use_relays", DefaultUseRelays) && !c.GetBool("relay.am_relay", false)
handshakeConfig := HandshakeConfig{ handshakeConfig := HandshakeConfig{
tryInterval: c.GetDuration("handshakes.try_interval", DefaultHandshakeTryInterval), tryInterval: c.GetDuration("handshakes.try_interval", DefaultHandshakeTryInterval),
retries: int64(c.GetInt("handshakes.retries", DefaultHandshakeRetries)), retries: int64(c.GetInt("handshakes.retries", DefaultHandshakeRetries)),
triggerBuffer: c.GetInt("handshakes.trigger_buffer", DefaultHandshakeTriggerBuffer), triggerBuffer: c.GetInt("handshakes.trigger_buffer", DefaultHandshakeTriggerBuffer),
useRelays: useRelays,
messageMetrics: messageMetrics, messageMetrics: messageMetrics,
} }

View File

@@ -18,6 +18,7 @@ type relayManager struct {
l *slog.Logger l *slog.Logger
hostmap *HostMap hostmap *HostMap
amRelay atomic.Bool amRelay atomic.Bool
useRelays atomic.Bool
} }
func NewRelayManager(ctx context.Context, l *slog.Logger, hostmap *HostMap, c *config.C) *relayManager { func NewRelayManager(ctx context.Context, l *slog.Logger, hostmap *HostMap, c *config.C) *relayManager {
@@ -36,8 +37,10 @@ func NewRelayManager(ctx context.Context, l *slog.Logger, hostmap *HostMap, c *c
} }
func (rm *relayManager) reload(c *config.C, initial bool) error { func (rm *relayManager) reload(c *config.C, initial bool) error {
if initial || c.HasChanged("relay.am_relay") { if initial || c.HasChanged("relay.am_relay") || c.HasChanged("relay.use_relays") {
rm.setAmRelay(c.GetBool("relay.am_relay", false)) amRelay := c.GetBool("relay.am_relay", false)
rm.amRelay.Store(amRelay)
rm.useRelays.Store(c.GetBool("relay.use_relays", true) && !amRelay)
} }
return nil return nil
} }
@@ -46,8 +49,157 @@ func (rm *relayManager) GetAmRelay() bool {
return rm.amRelay.Load() return rm.amRelay.Load()
} }
func (rm *relayManager) setAmRelay(v bool) { func (rm *relayManager) GetUseRelays() bool {
rm.amRelay.Store(v) return rm.useRelays.Load()
}
// StartRelays drives the relay-establishment side of an outbound handshake attempt.
// For each candidate relay it either kicks off a handshake to the relay, sends a CreateRelayRequest, retransmits
// one that may have been lost, or, once the relay is Established, forwards the in-progress
// stage 0 handshake packet for vpnIp through it.
func (rm *relayManager) StartRelays(f *Interface, vpnIp netip.Addr, hostinfo *HostInfo, stage0 []byte) {
if !rm.GetUseRelays() || len(hostinfo.remotes.relays) == 0 {
return
}
hostinfo.logger(rm.l).Info("Attempt to relay through hosts", "relays", hostinfo.remotes.relays)
// Send a RelayRequest to all known Relay IP's
for _, relay := range hostinfo.remotes.relays {
// Don't relay through the host I'm trying to connect to
if relay == vpnIp {
continue
}
// Don't relay to myself
if f.myVpnAddrsTable.Contains(relay) {
continue
}
relayHostInfo := rm.hostmap.QueryVpnAddr(relay)
if relayHostInfo == nil || !relayHostInfo.remote.IsValid() {
hostinfo.logger(rm.l).Info("Establish tunnel to relay target", "relay", relay.String())
f.Handshake(relay)
continue
}
// Check the relay HostInfo to see if we already established a relay through
existingRelay, ok := relayHostInfo.relayState.QueryRelayForByIp(vpnIp)
if !ok {
// No relays exist or requested yet.
if relayHostInfo.remote.IsValid() {
idx, err := AddRelay(rm.l, relayHostInfo, rm.hostmap, vpnIp, nil, TerminalType, Requested)
if err != nil {
hostinfo.logger(rm.l).Info("Failed to add relay to hostmap", "relay", relay.String(), "error", err)
}
m := NebulaControl{
Type: NebulaControl_CreateRelayRequest,
InitiatorRelayIndex: idx,
}
switch relayHostInfo.GetCert().Certificate.Version() {
case cert.Version1:
if !f.myVpnAddrs[0].Is4() {
hostinfo.logger(rm.l).Error("can not establish v1 relay with a v6 network because the relay is not running a current nebula version")
continue
}
if !vpnIp.Is4() {
hostinfo.logger(rm.l).Error("can not establish v1 relay with a v6 remote network because the relay is not running a current nebula version")
continue
}
b := f.myVpnAddrs[0].As4()
m.OldRelayFromAddr = binary.BigEndian.Uint32(b[:])
b = vpnIp.As4()
m.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
case cert.Version2:
m.RelayFromAddr = netAddrToProtoAddr(f.myVpnAddrs[0])
m.RelayToAddr = netAddrToProtoAddr(vpnIp)
default:
hostinfo.logger(rm.l).Error("Unknown certificate version found while creating relay")
continue
}
msg, err := m.Marshal()
if err != nil {
hostinfo.logger(rm.l).Error("Failed to marshal Control message to create relay", "error", err)
} else {
f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
rm.l.Info("send CreateRelayRequest",
"relayFrom", f.myVpnAddrs[0],
"relayTo", vpnIp,
"initiatorRelayIndex", idx,
"relay", relay,
)
}
}
continue
}
switch existingRelay.State {
case Established:
hostinfo.logger(rm.l).Info("Send handshake via relay", "relay", relay.String())
f.SendVia(relayHostInfo, existingRelay, stage0, make([]byte, 12), make([]byte, mtu), false)
case Disestablished:
// Mark this relay as 'requested'
relayHostInfo.relayState.UpdateRelayForByIpState(vpnIp, Requested)
fallthrough
case Requested:
hostinfo.logger(rm.l).Info("Re-send CreateRelay request", "relay", relay.String())
// Re-send the CreateRelay request, in case the previous one was lost.
m := NebulaControl{
Type: NebulaControl_CreateRelayRequest,
InitiatorRelayIndex: existingRelay.LocalIndex,
}
switch relayHostInfo.GetCert().Certificate.Version() {
case cert.Version1:
if !f.myVpnAddrs[0].Is4() {
hostinfo.logger(rm.l).Error("can not establish v1 relay with a v6 network because the relay is not running a current nebula version")
continue
}
if !vpnIp.Is4() {
hostinfo.logger(rm.l).Error("can not establish v1 relay with a v6 remote network because the relay is not running a current nebula version")
continue
}
b := f.myVpnAddrs[0].As4()
m.OldRelayFromAddr = binary.BigEndian.Uint32(b[:])
b = vpnIp.As4()
m.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
case cert.Version2:
m.RelayFromAddr = netAddrToProtoAddr(f.myVpnAddrs[0])
m.RelayToAddr = netAddrToProtoAddr(vpnIp)
default:
hostinfo.logger(rm.l).Error("Unknown certificate version found while creating relay")
continue
}
msg, err := m.Marshal()
if err != nil {
hostinfo.logger(rm.l).Error("Failed to marshal Control message to create relay", "error", err)
} else {
// This must send over the hostinfo, not over hm.Hosts[ip]
f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
rm.l.Info("send CreateRelayRequest",
"relayFrom", f.myVpnAddrs[0],
"relayTo", vpnIp,
"initiatorRelayIndex", existingRelay.LocalIndex,
"relay", relay,
)
}
case PeerRequested:
// PeerRequested only occurs in Forwarding relays, not Terminal relays, and this is a Terminal relay case.
fallthrough
default:
hostinfo.logger(rm.l).Error("Relay unexpected state",
"vpnIp", vpnIp,
"state", existingRelay.State,
"relay", relay,
)
}
}
} }
// AddRelay finds an available relay index on the hostmap, and associates the relay info with it. // AddRelay finds an available relay index on the hostmap, and associates the relay info with it.