autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
package autopilot
|
|
|
|
|
|
|
|
import (
|
|
|
|
"io/ioutil"
|
|
|
|
"os"
|
|
|
|
"testing"
|
|
|
|
"time"
|
|
|
|
|
|
|
|
prand "math/rand"
|
|
|
|
|
|
|
|
"github.com/lightningnetwork/lnd/channeldb"
|
|
|
|
"github.com/lightningnetwork/lnd/lnwire"
|
|
|
|
"github.com/roasbeef/btcd/btcec"
|
|
|
|
"github.com/roasbeef/btcutil"
|
|
|
|
)
|
|
|
|
|
|
|
|
func TestConstrainedPrefAttachmentNeedMoreChan(t *testing.T) {
|
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
minChanSize = 0
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
|
|
|
|
chanLimit = 3
|
|
|
|
|
|
|
|
threshold = 0.5
|
|
|
|
)
|
|
|
|
|
|
|
|
randChanID := func() lnwire.ShortChannelID {
|
|
|
|
return lnwire.NewShortChanIDFromInt(uint64(prand.Int63()))
|
|
|
|
}
|
|
|
|
|
|
|
|
testCases := []struct {
|
|
|
|
channels []Channel
|
|
|
|
walletAmt btcutil.Amount
|
|
|
|
|
|
|
|
needMore bool
|
|
|
|
amtAvailable btcutil.Amount
|
|
|
|
}{
|
|
|
|
// Many available funds, but already have too many active open
|
|
|
|
// channels.
|
|
|
|
{
|
|
|
|
[]Channel{
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(prand.Int31()),
|
|
|
|
},
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(prand.Int31()),
|
|
|
|
},
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(prand.Int31()),
|
|
|
|
},
|
|
|
|
},
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin * 10),
|
|
|
|
false,
|
|
|
|
0,
|
|
|
|
},
|
|
|
|
|
|
|
|
// Ratio of funds in channels and total funds meets the
|
|
|
|
// threshold.
|
|
|
|
{
|
|
|
|
[]Channel{
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
},
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
},
|
|
|
|
},
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin * 2),
|
|
|
|
false,
|
|
|
|
0,
|
|
|
|
},
|
|
|
|
|
|
|
|
// Ratio of funds in channels and total funds is below the
|
|
|
|
// threshold. We have 10 BTC allocated amongst channels and
|
|
|
|
// funds, atm. We're targeting 50%, so 5 BTC should be
|
|
|
|
// allocated. Only 1 BTC is atm, so 4 BTC should be
|
|
|
|
// recommended.
|
|
|
|
{
|
|
|
|
[]Channel{
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
},
|
|
|
|
},
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin * 9),
|
|
|
|
true,
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin * 4),
|
|
|
|
},
|
|
|
|
|
|
|
|
// Ratio of funds in channels and total funds is below the
|
|
|
|
// threshold. We have 14 BTC total amongst the wallet's
|
|
|
|
// balance, and our currently opened channels. Since we're
|
|
|
|
// targeting a 50% allocation, we should commit 7 BTC. The
|
|
|
|
// current channels commit 4 BTC, so we should expected 3 bTC
|
|
|
|
// to be committed.
|
|
|
|
{
|
|
|
|
[]Channel{
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
},
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin * 3),
|
|
|
|
},
|
|
|
|
},
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin * 10),
|
|
|
|
true,
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin * 3),
|
|
|
|
},
|
|
|
|
|
|
|
|
// Ratio of funds in channels and total funds is above the
|
|
|
|
// threshold.
|
|
|
|
{
|
|
|
|
[]Channel{
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
},
|
|
|
|
{
|
|
|
|
ChanID: randChanID(),
|
|
|
|
Capacity: btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
},
|
|
|
|
},
|
|
|
|
btcutil.Amount(btcutil.SatoshiPerBitcoin),
|
|
|
|
false,
|
|
|
|
0,
|
|
|
|
},
|
|
|
|
}
|
|
|
|
|
2018-02-07 04:11:11 +01:00
|
|
|
prefAttach := NewConstrainedPrefAttachment(minChanSize, maxChanSize,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
chanLimit, threshold)
|
|
|
|
|
|
|
|
for i, testCase := range testCases {
|
2018-02-07 04:11:11 +01:00
|
|
|
amtToAllocate, needMore := prefAttach.NeedMoreChans(testCase.channels,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
testCase.walletAmt)
|
|
|
|
|
|
|
|
if amtToAllocate != testCase.amtAvailable {
|
|
|
|
t.Fatalf("test #%v: expected %v, got %v",
|
|
|
|
i, testCase.amtAvailable, amtToAllocate)
|
|
|
|
}
|
|
|
|
if needMore != testCase.needMore {
|
|
|
|
t.Fatalf("test #%v: expected %v, got %v",
|
|
|
|
i, testCase.needMore, needMore)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
type genGraphFunc func() (testGraph, func(), error)
|
|
|
|
|
|
|
|
type testGraph interface {
|
|
|
|
ChannelGraph
|
|
|
|
|
|
|
|
addRandChannel(*btcec.PublicKey, *btcec.PublicKey,
|
|
|
|
btcutil.Amount) (*ChannelEdge, *ChannelEdge, error)
|
|
|
|
}
|
|
|
|
|
|
|
|
func newDiskChanGraph() (testGraph, func(), error) {
|
|
|
|
// First, create a temporary directory to be used for the duration of
|
|
|
|
// this test.
|
|
|
|
tempDirName, err := ioutil.TempDir("", "channeldb")
|
|
|
|
if err != nil {
|
|
|
|
return nil, nil, err
|
|
|
|
}
|
|
|
|
|
|
|
|
// Next, create channeldb for the first time.
|
|
|
|
cdb, err := channeldb.Open(tempDirName)
|
|
|
|
if err != nil {
|
|
|
|
return nil, nil, err
|
|
|
|
}
|
|
|
|
|
|
|
|
cleanUp := func() {
|
|
|
|
cdb.Close()
|
|
|
|
os.RemoveAll(tempDirName)
|
|
|
|
}
|
|
|
|
|
|
|
|
return &databaseChannelGraph{
|
|
|
|
db: cdb.ChannelGraph(),
|
|
|
|
}, cleanUp, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
var _ testGraph = (*databaseChannelGraph)(nil)
|
|
|
|
|
|
|
|
func newMemChanGraph() (testGraph, func(), error) {
|
|
|
|
return newMemChannelGraph(), nil, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
var _ testGraph = (*memChannelGraph)(nil)
|
|
|
|
|
|
|
|
var chanGraphs = []struct {
|
|
|
|
name string
|
|
|
|
genFunc genGraphFunc
|
|
|
|
}{
|
|
|
|
{
|
|
|
|
name: "disk_graph",
|
|
|
|
genFunc: newDiskChanGraph,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
name: "mem_graph",
|
|
|
|
genFunc: newMemChanGraph,
|
|
|
|
},
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestConstrainedPrefAttachmentSelectEmptyGraph ensures that when passed en
|
|
|
|
// empty graph, the Select function always detects the state, and returns nil.
|
|
|
|
// Otherwise, it would be possible for the main Select loop to entire an
|
|
|
|
// infinite loop.
|
|
|
|
func TestConstrainedPrefAttachmentSelectEmptyGraph(t *testing.T) {
|
|
|
|
const (
|
|
|
|
minChanSize = 0
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
chanLimit = 3
|
|
|
|
threshold = 0.5
|
|
|
|
)
|
|
|
|
|
|
|
|
// First, we'll generate a random key that represents "us", and create
|
|
|
|
// a new instance of the heuristic with our set parameters.
|
|
|
|
self, err := randKey()
|
|
|
|
if err != nil {
|
|
|
|
t.Fatalf("unable to generate self key: %v", err)
|
|
|
|
}
|
2018-02-07 04:11:11 +01:00
|
|
|
prefAttach := NewConstrainedPrefAttachment(minChanSize, maxChanSize,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
chanLimit, threshold)
|
|
|
|
|
|
|
|
skipNodes := make(map[NodeID]struct{})
|
|
|
|
for _, graph := range chanGraphs {
|
|
|
|
success := t.Run(graph.name, func(t1 *testing.T) {
|
|
|
|
graph, cleanup, err := graph.genFunc()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create graph: %v", err)
|
|
|
|
}
|
|
|
|
if cleanup != nil {
|
|
|
|
defer cleanup()
|
|
|
|
}
|
|
|
|
|
|
|
|
// With the necessary state initialized, we'll not
|
|
|
|
// attempt to select a set of candidates channel for
|
|
|
|
// creation given the current state of the graph.
|
|
|
|
const walletFunds = btcutil.SatoshiPerBitcoin
|
2018-02-07 04:11:11 +01:00
|
|
|
directives, err := prefAttach.Select(self, graph,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
walletFunds, skipNodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// We shouldn't have selected any new directives as we
|
|
|
|
// started with an empty graph.
|
|
|
|
if len(directives) != 0 {
|
|
|
|
t1.Fatalf("zero attachment directives "+
|
2018-02-07 04:11:11 +01:00
|
|
|
"should have been returned instead %v were",
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
len(directives))
|
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestConstrainedPrefAttachmentSelectTwoVertexes ensures that when passed a
|
|
|
|
// graph with only two eligible vertexes, then both are selected (without any
|
|
|
|
// repeats), and the funds are appropriately allocated across each peer.
|
|
|
|
func TestConstrainedPrefAttachmentSelectTwoVertexes(t *testing.T) {
|
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
minChanSize = 0
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
chanLimit = 3
|
|
|
|
threshold = 0.5
|
|
|
|
)
|
|
|
|
|
|
|
|
skipNodes := make(map[NodeID]struct{})
|
|
|
|
for _, graph := range chanGraphs {
|
|
|
|
success := t.Run(graph.name, func(t1 *testing.T) {
|
|
|
|
graph, cleanup, err := graph.genFunc()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create graph: %v", err)
|
|
|
|
}
|
|
|
|
if cleanup != nil {
|
|
|
|
defer cleanup()
|
|
|
|
}
|
|
|
|
|
|
|
|
// First, we'll generate a random key that represents
|
|
|
|
// "us", and create a new instance of the heuristic
|
|
|
|
// with our set parameters.
|
|
|
|
self, err := randKey()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to generate self key: %v", err)
|
|
|
|
}
|
2018-02-07 04:11:11 +01:00
|
|
|
prefAttach := NewConstrainedPrefAttachment(minChanSize, maxChanSize,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
chanLimit, threshold)
|
|
|
|
|
|
|
|
// For this set, we'll load the memory graph with two
|
|
|
|
// nodes, and a random channel connecting them.
|
|
|
|
const chanCapacity = btcutil.SatoshiPerBitcoin
|
|
|
|
edge1, edge2, err := graph.addRandChannel(nil, nil, chanCapacity)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to generate channel: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// With the necessary state initialized, we'll not
|
|
|
|
// attempt to select a set of candidates channel for
|
|
|
|
// creation given the current state of the graph.
|
|
|
|
const walletFunds = btcutil.SatoshiPerBitcoin * 10
|
2018-02-07 04:11:11 +01:00
|
|
|
directives, err := prefAttach.Select(self, graph,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
walletFunds, skipNodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment directives: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Two new directives should have been selected, one
|
|
|
|
// for each node already present within the graph.
|
|
|
|
if len(directives) != 2 {
|
2018-02-07 04:11:11 +01:00
|
|
|
t1.Fatalf("two attachment directives should have been "+
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
"returned instead %v were", len(directives))
|
|
|
|
}
|
|
|
|
|
|
|
|
// The node attached to should be amongst the two edges
|
|
|
|
// created above.
|
|
|
|
for _, directive := range directives {
|
|
|
|
switch {
|
|
|
|
case directive.PeerKey.IsEqual(edge1.Peer.PubKey()):
|
|
|
|
case directive.PeerKey.IsEqual(edge2.Peer.PubKey()):
|
|
|
|
default:
|
|
|
|
t1.Fatalf("attache to unknown node: %x",
|
|
|
|
directive.PeerKey.SerializeCompressed())
|
|
|
|
}
|
|
|
|
|
|
|
|
// As the number of funds available exceed the
|
|
|
|
// max channel size, both edges should consume
|
|
|
|
// the maximum channel size.
|
|
|
|
if directive.ChanAmt != maxChanSize {
|
|
|
|
t1.Fatalf("max channel size should be allocated, "+
|
|
|
|
"instead %v was: ", maxChanSize)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestConstrainedPrefAttachmentSelectInsufficientFunds ensures that if the
|
|
|
|
// balance of the backing wallet is below the set min channel size, then it
|
|
|
|
// never recommends candidates to attach to.
|
|
|
|
func TestConstrainedPrefAttachmentSelectInsufficientFunds(t *testing.T) {
|
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
minChanSize = 0
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
chanLimit = 3
|
|
|
|
threshold = 0.5
|
|
|
|
)
|
|
|
|
|
|
|
|
skipNodes := make(map[NodeID]struct{})
|
|
|
|
for _, graph := range chanGraphs {
|
|
|
|
success := t.Run(graph.name, func(t1 *testing.T) {
|
|
|
|
graph, cleanup, err := graph.genFunc()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create graph: %v", err)
|
|
|
|
}
|
|
|
|
if cleanup != nil {
|
|
|
|
defer cleanup()
|
|
|
|
}
|
|
|
|
|
|
|
|
// First, we'll generate a random key that represents
|
|
|
|
// "us", and create a new instance of the heuristic
|
|
|
|
// with our set parameters.
|
|
|
|
self, err := randKey()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to generate self key: %v", err)
|
|
|
|
}
|
2018-02-07 04:11:11 +01:00
|
|
|
prefAttach := NewConstrainedPrefAttachment(
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
minChanSize, maxChanSize, chanLimit, threshold,
|
|
|
|
)
|
|
|
|
|
|
|
|
// Next, we'll attempt to select a set of candidates,
|
|
|
|
// passing zero for the amount of wallet funds. This
|
|
|
|
// should return an empty slice of directives.
|
2018-02-07 04:11:11 +01:00
|
|
|
directives, err := prefAttach.Select(self, graph, 0,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
skipNodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
if len(directives) != 0 {
|
|
|
|
t1.Fatalf("zero attachment directives "+
|
2018-02-07 04:11:11 +01:00
|
|
|
"should have been returned instead %v were",
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
len(directives))
|
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestConstrainedPrefAttachmentSelectGreedyAllocation tests that if upon
|
|
|
|
// deciding a set of candidates, we're unable to evenly split our funds, then
|
|
|
|
// we attempt to greedily allocate all funds to each selected vertex (up to the
|
|
|
|
// max channel size).
|
|
|
|
func TestConstrainedPrefAttachmentSelectGreedyAllocation(t *testing.T) {
|
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
minChanSize = 0
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
chanLimit = 3
|
|
|
|
threshold = 0.5
|
|
|
|
)
|
|
|
|
|
|
|
|
skipNodes := make(map[NodeID]struct{})
|
|
|
|
for _, graph := range chanGraphs {
|
|
|
|
success := t.Run(graph.name, func(t1 *testing.T) {
|
|
|
|
graph, cleanup, err := graph.genFunc()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create graph: %v", err)
|
|
|
|
}
|
|
|
|
if cleanup != nil {
|
|
|
|
defer cleanup()
|
|
|
|
}
|
|
|
|
|
|
|
|
// First, we'll generate a random key that represents
|
|
|
|
// "us", and create a new instance of the heuristic
|
|
|
|
// with our set parameters.
|
|
|
|
self, err := randKey()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to generate self key: %v", err)
|
|
|
|
}
|
2018-02-07 04:11:11 +01:00
|
|
|
prefAttach := NewConstrainedPrefAttachment(
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
minChanSize, maxChanSize, chanLimit, threshold,
|
|
|
|
)
|
|
|
|
|
2018-02-07 04:11:11 +01:00
|
|
|
const chanCapacity = btcutil.SatoshiPerBitcoin
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
|
|
|
|
// Next, we'll add 3 nodes to the graph, creating an
|
|
|
|
// "open triangle topology".
|
|
|
|
edge1, _, err := graph.addRandChannel(nil, nil,
|
2018-02-07 04:11:11 +01:00
|
|
|
chanCapacity)
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create channel: %v", err)
|
|
|
|
}
|
|
|
|
_, _, err = graph.addRandChannel(
|
2018-02-07 04:11:11 +01:00
|
|
|
edge1.Peer.PubKey(), nil, chanCapacity,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create channel: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// At this point, there should be three nodes in the
|
|
|
|
// graph, with node node having two edges.
|
|
|
|
numNodes := 0
|
|
|
|
twoChans := false
|
|
|
|
if err := graph.ForEachNode(func(n Node) error {
|
|
|
|
numNodes++
|
|
|
|
|
|
|
|
numChans := 0
|
|
|
|
err := n.ForEachChannel(func(c ChannelEdge) error {
|
|
|
|
numChans++
|
|
|
|
return nil
|
|
|
|
})
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
|
|
|
twoChans = twoChans || (numChans == 2)
|
|
|
|
|
|
|
|
return nil
|
|
|
|
}); err != nil {
|
|
|
|
t1.Fatalf("unable to traverse graph: %v", err)
|
|
|
|
}
|
|
|
|
if numNodes != 3 {
|
|
|
|
t1.Fatalf("expected 3 nodes, instead have: %v",
|
|
|
|
numNodes)
|
|
|
|
}
|
|
|
|
if !twoChans {
|
|
|
|
t1.Fatalf("expected node to have two channels")
|
|
|
|
}
|
|
|
|
|
|
|
|
// We'll now begin our test, modeling the available
|
|
|
|
// wallet balance to be 5.5 BTC. We're shooting for a
|
|
|
|
// 50/50 allocation, and have 3 BTC in channels. As a
|
|
|
|
// result, the heuristic should try to greedily
|
|
|
|
// allocate funds to channels.
|
|
|
|
const availableBalance = btcutil.SatoshiPerBitcoin * 2.5
|
2018-02-07 04:11:11 +01:00
|
|
|
directives, err := prefAttach.Select(self, graph,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
availableBalance, skipNodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Three directives should have been returned.
|
|
|
|
if len(directives) != 3 {
|
|
|
|
t1.Fatalf("expected 3 directives, instead "+
|
|
|
|
"got: %v", len(directives))
|
|
|
|
}
|
|
|
|
|
|
|
|
// The two directive should have the max channel amount
|
|
|
|
// allocated.
|
|
|
|
if directives[0].ChanAmt != maxChanSize {
|
|
|
|
t1.Fatalf("expected recommendation of %v, "+
|
|
|
|
"instead got %v", maxChanSize,
|
|
|
|
directives[0].ChanAmt)
|
|
|
|
}
|
|
|
|
if directives[1].ChanAmt != maxChanSize {
|
|
|
|
t1.Fatalf("expected recommendation of %v, "+
|
|
|
|
"instead got %v", maxChanSize,
|
|
|
|
directives[1].ChanAmt)
|
|
|
|
}
|
|
|
|
|
|
|
|
// The third channel should have been allocated the
|
|
|
|
// remainder, or 0.5 BTC.
|
|
|
|
if directives[2].ChanAmt != (btcutil.SatoshiPerBitcoin * 0.5) {
|
|
|
|
t1.Fatalf("expected recommendation of %v, "+
|
|
|
|
"instead got %v", maxChanSize,
|
|
|
|
directives[2].ChanAmt)
|
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// TestConstrainedPrefAttachmentSelectSkipNodes ensures that if a node was
|
|
|
|
// already select for attachment, then that node is excluded from the set of
|
|
|
|
// candidate nodes.
|
|
|
|
func TestConstrainedPrefAttachmentSelectSkipNodes(t *testing.T) {
|
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
minChanSize = 0
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
chanLimit = 3
|
|
|
|
threshold = 0.5
|
|
|
|
)
|
|
|
|
|
|
|
|
for _, graph := range chanGraphs {
|
|
|
|
success := t.Run(graph.name, func(t1 *testing.T) {
|
2017-08-22 07:21:35 +02:00
|
|
|
skipNodes := make(map[NodeID]struct{})
|
|
|
|
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
graph, cleanup, err := graph.genFunc()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create graph: %v", err)
|
|
|
|
}
|
|
|
|
if cleanup != nil {
|
|
|
|
defer cleanup()
|
|
|
|
}
|
|
|
|
|
|
|
|
// First, we'll generate a random key that represents
|
|
|
|
// "us", and create a new instance of the heuristic
|
|
|
|
// with our set parameters.
|
|
|
|
self, err := randKey()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to generate self key: %v", err)
|
|
|
|
}
|
2018-02-07 04:11:11 +01:00
|
|
|
prefAttach := NewConstrainedPrefAttachment(
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
minChanSize, maxChanSize, chanLimit, threshold,
|
|
|
|
)
|
|
|
|
|
|
|
|
// Next, we'll create a simple topology of two nodes,
|
|
|
|
// with a single channel connecting them.
|
2018-02-07 04:11:11 +01:00
|
|
|
const chanCapacity = btcutil.SatoshiPerBitcoin
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
_, _, err = graph.addRandChannel(nil, nil,
|
2018-02-07 04:11:11 +01:00
|
|
|
chanCapacity)
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create channel: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// With our graph created, we'll now execute the Select
|
|
|
|
// function to recommend potential attachment
|
|
|
|
// candidates.
|
|
|
|
const availableBalance = btcutil.SatoshiPerBitcoin * 2.5
|
2018-02-07 04:11:11 +01:00
|
|
|
directives, err := prefAttach.Select(self, graph,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
availableBalance, skipNodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
// As the channel limit is three, and two nodes are
|
|
|
|
// present in the graph, both should be selected.
|
|
|
|
if len(directives) != 2 {
|
|
|
|
t1.Fatalf("expected two directives, instead "+
|
|
|
|
"got %v", len(directives))
|
|
|
|
}
|
|
|
|
|
|
|
|
// We'll simulate a channel update by adding the nodes
|
|
|
|
// we just establish channel with the to set of nodes
|
|
|
|
// to be skipped.
|
|
|
|
skipNodes[NewNodeID(directives[0].PeerKey)] = struct{}{}
|
|
|
|
skipNodes[NewNodeID(directives[1].PeerKey)] = struct{}{}
|
|
|
|
|
|
|
|
// If we attempt to make a call to the Select function,
|
|
|
|
// without providing any new information, then we
|
|
|
|
// should get no new directives as both nodes has
|
|
|
|
// already been attached to.
|
2018-02-07 04:11:11 +01:00
|
|
|
directives, err = prefAttach.Select(self, graph,
|
autopilot: "Look ma no hands!", introducing autopilot mode
This commit introduces the initial implementation of the autopilot
mode. Autopilot is new mode within lnd that enables automatic channel
management. This means that if enabled lnd will attempt to
automatically manage channels according to a set of heuristic defined
within the main configuration for autopilot.Agent instance.
The autopilot.Agent implements a simple closed control loop. It takes
in external signals such as wallet balance updates, new open channel,
and channels that are now closed the updates its internal state. With
each external trigger it will consult the registered
AttachmentHeuristic to decide: if it needs to open any more channels,
and if so how much it should use to open the channels, ultimately
returning a set of recommended AttachmentDirectives. The
autopilot.Agent loop will then take those attempt to establish
connection, and go back in waiting for a new external signal.
With this first implementation the default heuristic is the
ConstrainedPrefAttachment implementation of AttachmentHeuristic. Given
a min and max channel size, a limit on the number of channels, and the
percentage of wallet funds to allocate to channels, it will attempt to
execute a heuristic drive by the Barabási–Albert model model in order
to attempt to drive the global graph towards a scale free topology.
This is commit implements a foundational layer for future simulations,
optimization, and additional heuristics.
2017-08-11 06:14:41 +02:00
|
|
|
availableBalance, skipNodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
if len(directives) != 0 {
|
|
|
|
t1.Fatalf("zero new directives should have been "+
|
|
|
|
"selected, but %v were", len(directives))
|
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|