lnd

README

Lightning Network Daemon

The Lightning Network Daemon (lnd) - is a complete implementation of a Lightning Network node. lnd has several pluggable back-end chain services including btcd (a full-node), bitcoind, and neutrino (a new experimental light client). The project's codebase uses the btcsuite set of Bitcoin libraries, and also exports a large set of isolated re-usable Lightning Network related libraries within it. In the current state lnd is capable of:

• Creating channels.
• Closing channels.
• Completely managing all channel states (including the exceptional ones!).
• Maintaining a fully authenticated+validated channel graph.
• Performing path finding within the network, passively forwarding incoming payments.
• Sending outgoing onion-encrypted payments through the network.
• Updating advertised fee schedules.
• Automatic channel management (autopilot).

Lightning Network Specification Compliance

lnd fully conforms to the Lightning Network specification (BOLTs). BOLT stands for: Basis of Lightning Technology. The specifications are currently being drafted by several groups of implementers based around the world including the developers of lnd. The set of specification documents as well as our implementation of the specification are still a work-in-progress. With that said, the current status of lnd's BOLT compliance is:

• BOLT 1: Base Protocol
• BOLT 2: Peer Protocol for Channel Management
• BOLT 3: Bitcoin Transaction and Script Formats
• BOLT 4: Onion Routing Protocol
• BOLT 5: Recommendations for On-chain Transaction Handling
• BOLT 7: P2P Node and Channel Discovery
• BOLT 8: Encrypted and Authenticated Transport
• BOLT 9: Assigned Feature Flags
• BOLT 10: DNS Bootstrap and Assisted Node Location
• BOLT 11: Invoice Protocol for Lightning Payments

Developer Resources

The daemon has been designed to be as developer friendly as possible in order to facilitate application development on top of lnd. Two primary RPC interfaces are exported: an HTTP REST API, and a gRPC service. The exported API's are not yet stable, so be warned: they may change drastically in the near future.

An automatically generated set of documentation for the RPC APIs can be found at api.lightning.community. A set of developer resources including talks, articles, and example applications can be found at: dev.lightning.community.

Finally, we also have an active Slack where protocol developers, application developers, testers and users gather to discuss various aspects of lnd and also Lightning in general.

Installation

In order to build from source, please see the installation instructions.

Docker

To run lnd from Docker, please see the main Docker instructions

IRC

• irc.freenode.net
• channel #lnd
• webchat

Security

The developers of lnd take security very seriously. The disclosure of security vulnerabilities helps us secure the health of lnd, privacy of our users, and also the health of the Lightning Network as a whole. If you find any issues regarding security or privacy, please disclose the information responsibly by sending an email to security at lightning dot engineering, preferably encrypted using our designated PGP key (91FE464CD75101DA6B6BAB60555C6465E5BCB3AF) which can be found here.

Further reading

• Step-by-step send payment guide with docker
• Contribution guide

Documentation

Index

• Constants
• Variables
• func Main() error
• type BreachConfig
• type ContractBreachEvent
• type NurseryConfig
• type NurseryStore
• type RetributionStore
• type WalletUnlockParams

Constants

const UnassignedConnID uint64 = 0

UnassignedConnID is the default connection ID that a request can have before it actually is submitted to the connmgr. TODO(conner): move into connmgr package, or better, add connmgr method for generating atomic IDs

Variables

var (
	// ErrChanAlreadyClosing is returned when a channel shutdown is
	// attempted more than once.
	ErrChanAlreadyClosing = fmt.Errorf("channel shutdown already initiated")

	// ErrChanCloseNotFinished is returned when a caller attempts to access
	// a field or function that is contingent on the channel closure
	// negotiation already being completed.
	ErrChanCloseNotFinished = fmt.Errorf("close negotiation not finished")

	// ErrInvalidState is returned when the closing state machine receives
	// a message while it is in an unknown state.
	ErrInvalidState = fmt.Errorf("invalid state")
)

var (
	// ErrPeerNotConnected signals that the server has no connection to the
	// given peer.
	ErrPeerNotConnected = errors.New("peer is not connected")

	// ErrServerShuttingDown indicates that the server is in the process of
	// gracefully exiting.
	ErrServerShuttingDown = errors.New("server is shutting down")
)

var (

	// ErrChannelNotFound is an error returned when a channel is not known
	// to us. In this case of the fundingManager, this error is returned
	// when the channel in question is not considered being in an opening
	// state.
	ErrChannelNotFound = fmt.Errorf("channel not found")
)

var (
	// ErrContractNotFound is returned when the nursery is unable to
	// retrieve information about a queried contract.
	ErrContractNotFound = fmt.Errorf("unable to locate contract")
)

var (

	// ErrFundingManagerShuttingDown is an error returned when attempting to
	// process a funding request/message but the funding manager has already
	// been signaled to shut down.
	ErrFundingManagerShuttingDown = errors.New("funding manager shutting " +
		"down")
)

var ErrImmatureChannel = errors.New("cannot remove immature channel, " +
	"still has ungraduated outputs")

ErrImmatureChannel signals a channel cannot be removed because not all of its outputs have graduated.

Functions

func Main

func Main() error

Main is the true entry point for lnd. This function is required since defers created in the top-level scope of a main method aren't executed if os.Exit() is called.

Types

type BreachConfig

type BreachConfig struct {
	// CloseLink allows the breach arbiter to shutdown any channel links for
	// which it detects a breach, ensuring now further activity will
	// continue across the link. The method accepts link's channel point and
	// a close type to be included in the channel close summary.
	CloseLink func(*wire.OutPoint, htlcswitch.ChannelCloseType)

	// DB provides access to the user's channels, allowing the breach
	// arbiter to determine the current state of a user's channels, and how
	// it should respond to channel closure.
	DB *channeldb.DB

	// Estimator is used by the breach arbiter to determine an appropriate
	// fee level when generating, signing, and broadcasting sweep
	// transactions.
	Estimator lnwallet.FeeEstimator

	// GenSweepScript generates the receiving scripts for swept outputs.
	GenSweepScript func() ([]byte, error)

	// Notifier provides a publish/subscribe interface for event driven
	// notifications regarding the confirmation of txids.
	Notifier chainntnfs.ChainNotifier

	// PublishTransaction facilitates the process of broadcasting a
	// transaction to the network.
	PublishTransaction func(*wire.MsgTx) error

	// ContractBreaches is a channel where the breachArbiter will receive
	// notifications in the event of a contract breach being observed. A
	// ContractBreachEvent must be ACKed by the breachArbiter, such that
	// the sending subsystem knows that the event is properly handed off.
	ContractBreaches <-chan *ContractBreachEvent

	// Signer is used by the breach arbiter to generate sweep transactions,
	// which move coins from previously open channels back to the user's
	// wallet.
	Signer input.Signer

	// Store is a persistent resource that maintains information regarding
	// breached channels. This is used in conjunction with DB to recover
	// from crashes, restarts, or other failures.
	Store RetributionStore
}

BreachConfig bundles the required subsystems used by the breach arbiter. An instance of BreachConfig is passed to newBreachArbiter during instantiation.

type ContractBreachEvent

type ContractBreachEvent struct {
	// ChanPoint is the channel point of the breached channel.
	ChanPoint wire.OutPoint

	// ProcessACK is an error channel where a nil error should be sent
	// iff the breach retribution info is safely stored in the retribution
	// store. In case storing the information to the store fails, a non-nil
	// error should be sent.
	ProcessACK chan error

	// BreachRetribution is the information needed to act on this contract
	// breach.
	BreachRetribution *lnwallet.BreachRetribution
}

ContractBreachEvent is an event the breachArbiter will receive in case a contract breach is observed on-chain. It contains the necessary information to handle the breach, and a ProcessACK channel we will use to ACK the event when we have safely stored all the necessary information.

type NurseryConfig

type NurseryConfig struct {
	// ChainIO is used by the utxo nursery to determine the current block
	// height, which drives the incubation of the nursery's outputs.
	ChainIO lnwallet.BlockChainIO

	// ConfDepth is the number of blocks the nursery store waits before
	// determining outputs in the chain as confirmed.
	ConfDepth uint32

	// FetchClosedChannels provides access to a user's channels, such that
	// they can be marked fully closed after incubation has concluded.
	FetchClosedChannels func(pendingOnly bool) (
		[]*channeldb.ChannelCloseSummary, error)

	// FetchClosedChannel provides access to the close summary to extract a
	// height hint from.
	FetchClosedChannel func(chanID *wire.OutPoint) (
		*channeldb.ChannelCloseSummary, error)

	// Notifier provides the utxo nursery the ability to subscribe to
	// transaction confirmation events, which advance outputs through their
	// persistence state transitions.
	Notifier chainntnfs.ChainNotifier

	// PublishTransaction facilitates the process of broadcasting a signed
	// transaction to the appropriate network.
	PublishTransaction func(*wire.MsgTx) error

	// Store provides access to and modification of the persistent state
	// maintained about the utxo nursery's incubating outputs.
	Store NurseryStore

	// Sweep sweeps an input back to the wallet.
	SweepInput func(input input.Input) (chan sweep.Result, error)
}

NurseryConfig abstracts the required subsystems used by the utxo nursery. An instance of NurseryConfig is passed to newUtxoNursery during instantiation.

type NurseryStore

type NurseryStore interface {
	// Incubate registers a set of CSV delayed outputs (incoming HTLC's on
	// our commitment transaction, or a commitment output), and a slice of
	// outgoing htlc outputs to be swept back into the user's wallet. The
	// event is persisted to disk, such that the nursery can resume the
	// incubation process after a potential crash.
	Incubate([]kidOutput, []babyOutput) error

	// CribToKinder atomically moves a babyOutput in the crib bucket to the
	// kindergarten bucket. Baby outputs are outgoing HTLC's which require
	// us to go to the second-layer to claim. The now mature kidOutput
	// contained in the babyOutput will be stored as it waits out the
	// kidOutput's CSV delay.
	CribToKinder(*babyOutput) error

	// PreschoolToKinder atomically moves a kidOutput from the preschool
	// bucket to the kindergarten bucket. This transition should be executed
	// after receiving confirmation of the preschool output. Incoming HTLC's
	// we need to go to the second-layer to claim, and also our commitment
	// outputs fall into this class.
	//
	// An additional parameter specifies the last graduated height. This is
	// used in case of late registration. It schedules the output for sweep
	// at the next epoch even though it has already expired earlier.
	PreschoolToKinder(kid *kidOutput, lastGradHeight uint32) error

	// GraduateKinder atomically moves an output at the provided height into
	// the graduated status. This involves removing the kindergarten entries
	// from both the height and channel indexes. The height bucket will be
	// opportunistically pruned from the height index as outputs are
	// removed.
	GraduateKinder(height uint32, output *kidOutput) error

	// FetchPreschools returns a list of all outputs currently stored in
	// the preschool bucket.
	FetchPreschools() ([]kidOutput, error)

	// FetchClass returns a list of kindergarten and crib outputs whose
	// timelocks expire at the given height.
	FetchClass(height uint32) ([]kidOutput, []babyOutput, error)

	// HeightsBelowOrEqual returns the lowest non-empty heights in the
	// height index, that exist at or below the provided upper bound.
	HeightsBelowOrEqual(height uint32) ([]uint32, error)

	// ForChanOutputs iterates over all outputs being incubated for a
	// particular channel point. This method accepts a callback that allows
	// the caller to process each key-value pair. The key will be a prefixed
	// outpoint, and the value will be the serialized bytes for an output,
	// whose type should be inferred from the key's prefix.
	ForChanOutputs(*wire.OutPoint, func([]byte, []byte) error) error

	// ListChannels returns all channels the nursery is currently tracking.
	ListChannels() ([]wire.OutPoint, error)

	// IsMatureChannel determines the whether or not all of the outputs in a
	// particular channel bucket have been marked as graduated.
	IsMatureChannel(*wire.OutPoint) (bool, error)

	// RemoveChannel channel erases all entries from the channel bucket for
	// the provided channel point, this method should only be called if
	// IsMatureChannel indicates the channel is ready for removal.
	RemoveChannel(*wire.OutPoint) error
}

NurseryStore abstracts the persistent storage layer for the utxo nursery. Concretely, it stores commitment and htlc outputs until any time-bounded constraints have fully matured. The store exposes methods for enumerating its contents, and persisting state transitions detected by the utxo nursery.

type RetributionStore

type RetributionStore interface {
	// Add persists the retributionInfo to disk, using the information's
	// chanPoint as the key. This method should overwrite any existing
	// entries found under the same key, and an error should be raised if
	// the addition fails.
	Add(retInfo *retributionInfo) error

	// IsBreached queries the retribution store to see if the breach arbiter
	// is aware of any breaches for the provided channel point.
	IsBreached(chanPoint *wire.OutPoint) (bool, error)

	// Finalize persists the finalized justice transaction for a particular
	// channel.
	Finalize(chanPoint *wire.OutPoint, finalTx *wire.MsgTx) error

	// GetFinalizedTxn loads the finalized justice transaction, if any, from
	// the retribution store. The finalized transaction will be nil if
	// Finalize has not yet been called for this channel point.
	GetFinalizedTxn(chanPoint *wire.OutPoint) (*wire.MsgTx, error)

	// Remove deletes the retributionInfo from disk, if any exists, under
	// the given key. An error should be re raised if the removal fails.
	Remove(key *wire.OutPoint) error

	// ForAll iterates over the existing on-disk contents and applies a
	// chosen, read-only callback to each. This method should ensure that it
	// immediately propagate any errors generated by the callback.
	ForAll(cb func(*retributionInfo) error) error
}

RetributionStore provides an interface for managing a persistent map from wire.OutPoint -> retributionInfo. Upon learning of a breach, a BreachArbiter should record the retributionInfo for the breached channel, which serves a checkpoint in the event that retribution needs to be resumed after failure. A RetributionStore provides an interface for managing the persisted set, as well as mapping user defined functions over the entire on-disk contents.

Calls to RetributionStore may occur concurrently. A concrete instance of RetributionStore should use appropriate synchronization primitives, or be otherwise safe for concurrent access.

type WalletUnlockParams

type WalletUnlockParams struct {
	// Password is the public and private wallet passphrase.
	Password []byte

	// Birthday specifies the approximate time that this wallet was created.
	// This is used to bound any rescans on startup.
	Birthday time.Time

	// RecoveryWindow specifies the address lookahead when entering recovery
	// mode. A recovery will be attempted if this value is non-zero.
	RecoveryWindow uint32

	// Wallet is the loaded and unlocked Wallet. This is returned
	// from the unlocker service to avoid it being unlocked twice (once in
	// the unlocker service to check if the password is correct and again
	// later when lnd actually uses it). Because unlocking involves scrypt
	// which is resource intensive, we want to avoid doing it twice.
	Wallet *wallet.Wallet

	// ChansToRestore a set of static channel backups that should be
	// restored before the main server instance starts up.
	ChansToRestore walletunlocker.ChannelsToRecover
}

WalletUnlockParams holds the variables used to parameterize the unlocking of lnd's wallet after it has already been created.