server: start htlcswitch early in the pipeline

chanbackup/pubsub.go

@@ -137,7 +137,7 @@ func NewSubSwapper(startingChans []Single, chanNotifier ChannelNotifier,
 func (s *SubSwapper) Start() error {
 	var startErr error
 	s.started.Do(func() {
-		log.Infof("Starting chanbackup.SubSwapper")
+		log.Infof("chanbackup.SubSwapper starting")
 
 		// Before we enter our main loop, we'll update the on-disk
 		// state with the latest Single state, as nodes may have new

chanfitness/chaneventstore.go

@@ -142,6 +142,8 @@ func NewChannelEventStore(config *Config) *ChannelEventStore {
 // information from the store. If this function fails, it cancels its existing
 // subscriptions and returns an error.
 func (c *ChannelEventStore) Start() error {
+	log.Info("ChannelEventStore starting")
+
 	// Create a subscription to channel events.
 	channelClient, err := c.cfg.SubscribeChannelEvents()
 	if err != nil {

channelnotifier/channelnotifier.go

@@ -87,7 +87,7 @@ func New(chanDB *channeldb.ChannelStateDB) *ChannelNotifier {
 func (c *ChannelNotifier) Start() error {
 	var err error
 	c.started.Do(func() {
-		log.Trace("ChannelNotifier starting")
+		log.Info("ChannelNotifier starting")
 		err = c.ntfnServer.Start()
 	})
 	return err

contractcourt/breacharbiter.go

@@ -206,14 +206,13 @@ func NewBreachArbiter(cfg *BreachConfig) *BreachArbiter {
 func (b *BreachArbiter) Start() error {
 	var err error
 	b.started.Do(func() {
+		brarLog.Info("Breach arbiter starting")
 		err = b.start()
 	})
 	return err
 }
 
 func (b *BreachArbiter) start() error {
-	brarLog.Tracef("Starting breach arbiter")
-
 	// Load all retributions currently persisted in the retribution store.
 	var breachRetInfos map[wire.OutPoint]retributionInfo
 	if err := b.cfg.Store.ForAll(func(ret *retributionInfo) error {

contractcourt/chain_arbitrator.go

@@ -487,7 +487,7 @@ func (c *ChainArbitrator) Start() error {
 		return nil
 	}
 
-	log.Tracef("Starting ChainArbitrator")
+	log.Info("ChainArbitrator starting")
 
 	// First, we'll fetch all the channels that are still open, in order to
 	// collect them within our set of active contracts.

contractcourt/utxonursery.go

@@ -241,7 +241,7 @@ func (u *UtxoNursery) Start() error {
 		return nil
 	}
 
-	utxnLog.Tracef("Starting UTXO nursery")
+	utxnLog.Info("UTXO nursery starting")
 
 	// Retrieve the currently best known block. This is needed to have the
 	// state machine catch up with the blocks we missed when we were down.

discovery/gossiper.go

@@ -498,14 +498,13 @@ func (d *AuthenticatedGossiper) PropagateChanPolicyUpdate(
 func (d *AuthenticatedGossiper) Start() error {
 	var err error
 	d.started.Do(func() {
+		log.Info("Authenticated Gossiper starting")
 		err = d.start()
 	})
 	return err
 }
 
 func (d *AuthenticatedGossiper) start() error {
-	log.Info("Authenticated Gossiper is starting")
-
 	// First we register for new notifications of newly discovered blocks.
 	// We do this immediately so we'll later be able to consume any/all
 	// blocks which were discovered.

docs/release-notes/release-notes-0.15.0.md

@@ -36,6 +36,10 @@
   could result in an "invoice too large" error when creating invoices. Hints 
   are now properly limited to our maximum of 20.
 
+* [Fixed an edge case where the lnd might be stuck at starting due to channel
+  arbitrator relying on htlcswitch to be started
+  first](https://github.com/lightningnetwork/lnd/pull/6214).
+
 ## Misc
 
 * [An example systemd service file](https://github.com/lightningnetwork/lnd/pull/6033)

funding/manager.go

@@ -588,14 +588,13 @@ func NewFundingManager(cfg Config) (*Manager, error) {
 func (f *Manager) Start() error {
 	var err error
 	f.started.Do(func() {
+		log.Info("Funding manager starting")
 		err = f.start()
 	})
 	return err
 }
 
 func (f *Manager) start() error {
-	log.Tracef("Funding manager running")
-
 	// Upon restart, the Funding Manager will check the database to load any
 	// channels that were  waiting for their funding transactions to be
 	// confirmed on the blockchain at the time when the daemon last went
@@ -1235,7 +1234,7 @@ func (f *Manager) handleFundingOpen(peer lnpeer.Peer,
 	if amt < f.cfg.MinChanSize {
 		f.failFundingFlow(
 			peer, msg.PendingChannelID,
-			lnwallet.ErrChanTooSmall(amt, btcutil.Amount(f.cfg.MinChanSize)),
+			lnwallet.ErrChanTooSmall(amt, f.cfg.MinChanSize),
 		)
 		return
 	}

htlcswitch/hop/iterator.go

@@ -132,6 +132,7 @@ func NewOnionProcessor(router *sphinx.Router) *OnionProcessor {
 
 // Start spins up the onion processor's sphinx router.
 func (p *OnionProcessor) Start() error {
+	log.Info("Onion processor starting")
 	return p.router.Start()
 }
 

htlcswitch/htlcnotifier.go

@@ -81,7 +81,7 @@ func NewHtlcNotifier(now func() time.Time) *HtlcNotifier {
 func (h *HtlcNotifier) Start() error {
 	var err error
 	h.started.Do(func() {
-		log.Trace("HtlcNotifier starting")
+		log.Info("HtlcNotifier starting")
 		err = h.ntfnServer.Start()
 	})
 	return err

htlcswitch/switch.go

@@ -1846,7 +1846,7 @@ func (s *Switch) Start() error {
 		return errors.New("htlc switch already started")
 	}
 
-	log.Infof("Starting HTLC Switch")
+	log.Infof("HTLC Switch starting")
 
 	blockEpochStream, err := s.cfg.Notifier.RegisterBlockEpochNtfn(nil)
 	if err != nil {

invoices/invoiceregistry.go

@@ -238,6 +238,8 @@ func (i *InvoiceRegistry) Start() error {
 		return err
 	}
 
+	log.Info("InvoiceRegistry starting")
+
 	i.wg.Add(1)
 	go i.invoiceEventLoop()
 

lnwallet/sigpool.go

@@ -155,6 +155,7 @@ func NewSigPool(numWorkers int, signer input.Signer) *SigPool {
 // carry out its duties.
 func (s *SigPool) Start() error {
 	s.started.Do(func() {
+		walletLog.Info("SigPool starting")
 		for i := 0; i < s.numWorkers; i++ {
 			s.wg.Add(1)
 			go s.poolWorker()

netann/chan_status_manager.go

@@ -172,6 +172,7 @@ func NewChanStatusManager(cfg *ChanStatusConfig) (*ChanStatusManager, error) {
 func (m *ChanStatusManager) Start() error {
 	var err error
 	m.started.Do(func() {
+		log.Info("Channel Status Manager starting")
 		err = m.start()
 	})
 	return err

netann/host_ann.go

@@ -58,6 +58,7 @@ func NewHostAnnouncer(cfg HostAnnouncerConfig) *HostAnnouncer {
 // Start starts the HostAnnouncer.
 func (h *HostAnnouncer) Start() error {
 	h.startOnce.Do(func() {
+		log.Info("HostAnnouncer starting")
 		h.wg.Add(1)
 		go h.hostWatcher()
 	})

routing/router.go

@@ -490,7 +490,7 @@ func (r *ChannelRouter) Start() error {
 		return nil
 	}
 
-	log.Tracef("Channel Router starting")
+	log.Info("Channel Router starting")
 
 	bestHash, bestHeight, err := r.cfg.Chain.GetBestBlock()
 	if err != nil {

server.go

@@ -1777,6 +1777,15 @@ func (s *server) Start() error {
 		}
 		cleanup = cleanup.add(s.fundingMgr.Stop)
 
+		// htlcSwitch must be started before chainArb since the latter
+		// relies on htlcSwitch to deliver resolution message upon
+		// start.
+		if err := s.htlcSwitch.Start(); err != nil {
+			startErr = err
+			return
+		}
+		cleanup = cleanup.add(s.htlcSwitch.Stop)
+
 		if err := s.chainArb.Start(); err != nil {
 			startErr = err
 			return
@@ -1807,12 +1816,6 @@ func (s *server) Start() error {
 		}
 		cleanup = cleanup.add(s.sphinx.Stop)
 
-		if err := s.htlcSwitch.Start(); err != nil {
-			startErr = err
-			return
-		}
-		cleanup = cleanup.add(s.htlcSwitch.Stop)
-
 		if err := s.chanStatusMgr.Start(); err != nil {
 			startErr = err
 			return

sweep/sweeper.go

@@ -338,7 +338,7 @@ func (s *UtxoSweeper) Start() error {
 		return nil
 	}
 
-	log.Tracef("Sweeper starting")
+	log.Info("Sweeper starting")
 
 	// Retrieve last published tx from database.
 	lastTx, err := s.cfg.Store.GetLastPublishedTx()

watchtower/wtclient/client.go

@@ -439,7 +439,7 @@ func (c *TowerClient) buildHighestCommitHeights() {
 func (c *TowerClient) Start() error {
 	var err error
 	c.started.Do(func() {
-		c.log.Infof("Starting watchtower client")
+		c.log.Infof("Watchtower client starting")
 
 		// First, restart a session queue for any sessions that have
 		// committed but unacked state updates. This ensures that these