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 07:14:41 +03:00
|
|
|
package autopilot
|
|
|
|
|
|
|
|
import (
|
2018-08-30 01:44:26 +03:00
|
|
|
"bytes"
|
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 07:14:41 +03:00
|
|
|
"io/ioutil"
|
|
|
|
"os"
|
|
|
|
"testing"
|
|
|
|
"time"
|
|
|
|
|
|
|
|
prand "math/rand"
|
|
|
|
|
2018-06-05 04:34:16 +03:00
|
|
|
"github.com/btcsuite/btcd/btcec"
|
|
|
|
"github.com/btcsuite/btcutil"
|
2018-07-31 10:17:17 +03:00
|
|
|
"github.com/lightningnetwork/lnd/channeldb"
|
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 07:14:41 +03:00
|
|
|
)
|
|
|
|
|
|
|
|
type genGraphFunc func() (testGraph, func(), error)
|
|
|
|
|
|
|
|
type testGraph interface {
|
|
|
|
ChannelGraph
|
|
|
|
|
|
|
|
addRandChannel(*btcec.PublicKey, *btcec.PublicKey,
|
|
|
|
btcutil.Amount) (*ChannelEdge, *ChannelEdge, error)
|
2018-12-10 16:56:54 +03:00
|
|
|
|
|
|
|
addRandNode() (*btcec.PublicKey, error)
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
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,
|
|
|
|
},
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
// TestPrefAttachmentSelectEmptyGraph ensures that when passed an
|
2018-11-23 01:18:09 +03:00
|
|
|
// empty graph, the NodeSores function always returns a score of 0.
|
2018-12-19 16:54:54 +03:00
|
|
|
func TestPrefAttachmentSelectEmptyGraph(t *testing.T) {
|
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 07:14:41 +03:00
|
|
|
const (
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
|
|
|
)
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
prefAttach := NewPrefAttachment()
|
2018-11-23 01:18:09 +03:00
|
|
|
|
|
|
|
// Create a random public key, which we will query to get a score for.
|
|
|
|
pub, err := randKey()
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
2018-11-23 01:18:09 +03:00
|
|
|
t.Fatalf("unable to generate key: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
nodes := map[NodeID]struct{}{
|
|
|
|
NewNodeID(pub): {},
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
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()
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// With the necessary state initialized, we'll now
|
|
|
|
// attempt to get the score for this one node.
|
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 07:14:41 +03:00
|
|
|
const walletFunds = btcutil.SatoshiPerBitcoin
|
2018-11-23 01:18:09 +03:00
|
|
|
scores, err := prefAttach.NodeScores(graph, nil,
|
|
|
|
walletFunds, nodes)
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// Since the graph is empty, we expect the score to be
|
|
|
|
// 0, giving an empty return map.
|
|
|
|
if len(scores) != 0 {
|
|
|
|
t1.Fatalf("expected empty score map, "+
|
|
|
|
"instead got %v ", len(scores))
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// completeGraph is a helper method that adds numNodes fully connected nodes to
|
|
|
|
// the graph.
|
|
|
|
func completeGraph(t *testing.T, g testGraph, numNodes int) {
|
|
|
|
const chanCapacity = btcutil.SatoshiPerBitcoin
|
|
|
|
nodes := make(map[int]*btcec.PublicKey)
|
|
|
|
for i := 0; i < numNodes; i++ {
|
|
|
|
for j := i + 1; j < numNodes; j++ {
|
|
|
|
|
|
|
|
node1 := nodes[i]
|
|
|
|
node2 := nodes[j]
|
|
|
|
edge1, edge2, err := g.addRandChannel(
|
|
|
|
node1, node2, chanCapacity)
|
|
|
|
if err != nil {
|
|
|
|
t.Fatalf("unable to generate channel: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
if node1 == nil {
|
|
|
|
pubKeyBytes := edge1.Peer.PubKey()
|
|
|
|
nodes[i], err = btcec.ParsePubKey(
|
|
|
|
pubKeyBytes[:], btcec.S256(),
|
|
|
|
)
|
|
|
|
if err != nil {
|
|
|
|
t.Fatalf("unable to parse pubkey: %v",
|
|
|
|
err)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if node2 == nil {
|
|
|
|
pubKeyBytes := edge2.Peer.PubKey()
|
|
|
|
nodes[j], err = btcec.ParsePubKey(
|
|
|
|
pubKeyBytes[:], btcec.S256(),
|
|
|
|
)
|
|
|
|
if err != nil {
|
|
|
|
t.Fatalf("unable to parse pubkey: %v",
|
|
|
|
err)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
// TestPrefAttachmentSelectTwoVertexes ensures that when passed a
|
2018-11-23 01:18:09 +03:00
|
|
|
// graph with only two eligible vertexes, then both are given the same score,
|
|
|
|
// and the funds are appropriately allocated across each peer.
|
2018-12-19 16:54:54 +03:00
|
|
|
func TestPrefAttachmentSelectTwoVertexes(t *testing.T) {
|
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 07:14:41 +03:00
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
2018-12-19 16:54:53 +03:00
|
|
|
)
|
|
|
|
|
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 07:14:41 +03:00
|
|
|
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()
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
prefAttach := NewPrefAttachment()
|
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 07:14:41 +03:00
|
|
|
|
|
|
|
// 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)
|
|
|
|
}
|
|
|
|
|
2018-12-10 16:56:54 +03:00
|
|
|
// We also add a third, non-connected node to the graph.
|
|
|
|
_, err = graph.addRandNode()
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to add random node: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// Get the score for all nodes found in the graph at
|
|
|
|
// this point.
|
|
|
|
nodes := make(map[NodeID]struct{})
|
|
|
|
if err := graph.ForEachNode(func(n Node) error {
|
|
|
|
nodes[n.PubKey()] = struct{}{}
|
|
|
|
return nil
|
|
|
|
}); err != nil {
|
|
|
|
t1.Fatalf("unable to traverse graph: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-12-10 16:56:54 +03:00
|
|
|
if len(nodes) != 3 {
|
2018-11-23 01:18:09 +03:00
|
|
|
t1.Fatalf("expected 2 nodes, found %d", len(nodes))
|
|
|
|
}
|
|
|
|
|
|
|
|
// With the necessary state initialized, we'll now
|
|
|
|
// attempt to get our candidates channel score given
|
|
|
|
// the current state of the graph.
|
|
|
|
candidates, err := prefAttach.NodeScores(graph, nil,
|
2018-12-19 16:54:53 +03:00
|
|
|
maxChanSize, nodes)
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
2018-11-23 01:18:09 +03:00
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
2018-12-10 16:56:54 +03:00
|
|
|
// We expect two candidates, since one of the nodes
|
|
|
|
// doesn't have any channels.
|
|
|
|
if len(candidates) != 2 {
|
|
|
|
t1.Fatalf("2 nodes should be scored, "+
|
2018-11-23 01:18:09 +03:00
|
|
|
"instead %v were", len(candidates))
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// The candidates should be amongst the two edges
|
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 07:14:41 +03:00
|
|
|
// created above.
|
2018-11-23 01:18:09 +03:00
|
|
|
for nodeID, candidate := range candidates {
|
2018-08-30 01:44:26 +03:00
|
|
|
edge1Pub := edge1.Peer.PubKey()
|
|
|
|
edge2Pub := edge2.Peer.PubKey()
|
|
|
|
|
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 07:14:41 +03:00
|
|
|
switch {
|
2018-11-23 01:18:09 +03:00
|
|
|
case bytes.Equal(nodeID[:], edge1Pub[:]):
|
|
|
|
case bytes.Equal(nodeID[:], edge2Pub[:]):
|
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 07:14:41 +03:00
|
|
|
default:
|
2018-08-30 01:44:26 +03:00
|
|
|
t1.Fatalf("attached to unknown node: %x",
|
2018-11-23 01:18:09 +03:00
|
|
|
nodeID[:])
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// Since each of the nodes has 1 channel, out
|
|
|
|
// of only one channel in the graph, we expect
|
|
|
|
// their score to be 0.5.
|
|
|
|
expScore := float64(0.5)
|
|
|
|
if candidate.Score != expScore {
|
|
|
|
t1.Fatalf("expected candidate score "+
|
|
|
|
"to be %v, instead was %v",
|
|
|
|
expScore, candidate.Score)
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
// TestPrefAttachmentSelectGreedyAllocation tests that if upon
|
2018-11-23 01:18:09 +03:00
|
|
|
// returning node scores, the NodeScores method will attempt to greedily
|
|
|
|
// allocate all funds to each vertex (up to the max channel size).
|
2018-12-19 16:54:54 +03:00
|
|
|
func TestPrefAttachmentSelectGreedyAllocation(t *testing.T) {
|
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 07:14:41 +03:00
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
2018-12-19 16:54:53 +03:00
|
|
|
)
|
2018-11-23 01:18:08 +03:00
|
|
|
|
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 07:14:41 +03:00
|
|
|
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()
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
prefAttach := NewPrefAttachment()
|
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 07:14:41 +03:00
|
|
|
|
2018-02-07 06:11:11 +03:00
|
|
|
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 07:14:41 +03:00
|
|
|
|
|
|
|
// Next, we'll add 3 nodes to the graph, creating an
|
|
|
|
// "open triangle topology".
|
|
|
|
edge1, _, err := graph.addRandChannel(nil, nil,
|
2018-02-07 06:11:11 +03:00
|
|
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create channel: %v", err)
|
|
|
|
}
|
2018-08-30 01:44:26 +03:00
|
|
|
peerPubBytes := edge1.Peer.PubKey()
|
|
|
|
peerPub, err := btcec.ParsePubKey(
|
|
|
|
peerPubBytes[:], btcec.S256(),
|
|
|
|
)
|
|
|
|
if err != nil {
|
|
|
|
t.Fatalf("unable to parse pubkey: %v", err)
|
|
|
|
}
|
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 07:14:41 +03:00
|
|
|
_, _, err = graph.addRandChannel(
|
2018-08-30 01:44:26 +03:00
|
|
|
peerPub, 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 07:14:41 +03:00
|
|
|
)
|
|
|
|
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
|
2018-11-23 01:18:09 +03:00
|
|
|
nodes := 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 07:14:41 +03:00
|
|
|
if err := graph.ForEachNode(func(n Node) error {
|
|
|
|
numNodes++
|
2018-11-23 01:18:09 +03:00
|
|
|
nodes[n.PubKey()] = 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 07:14:41 +03:00
|
|
|
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.
|
2018-11-23 01:18:09 +03:00
|
|
|
scores, err := prefAttach.NodeScores(graph, nil,
|
2018-12-19 16:54:53 +03:00
|
|
|
maxChanSize, nodes)
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
if len(scores) != len(nodes) {
|
|
|
|
t1.Fatalf("all nodes should be scored, "+
|
|
|
|
"instead %v were", len(scores))
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// The candidates should have a non-zero score, and
|
|
|
|
// have the max chan size funds recommended channel
|
|
|
|
// size.
|
|
|
|
for _, candidate := range scores {
|
|
|
|
if candidate.Score == 0 {
|
|
|
|
t1.Fatalf("Expected non-zero score")
|
|
|
|
}
|
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 07:14:41 +03:00
|
|
|
}
|
2018-11-23 01:18:09 +03:00
|
|
|
|
|
|
|
// Imagine a few channels are being opened, and there's
|
|
|
|
// only 0.5 BTC left. That should leave us with channel
|
|
|
|
// candidates of that size.
|
|
|
|
const remBalance = btcutil.SatoshiPerBitcoin * 0.5
|
|
|
|
scores, err = prefAttach.NodeScores(graph, nil,
|
|
|
|
remBalance, nodes)
|
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
if len(scores) != len(nodes) {
|
|
|
|
t1.Fatalf("all nodes should be scored, "+
|
|
|
|
"instead %v were", len(scores))
|
|
|
|
}
|
|
|
|
|
|
|
|
// Check that the recommended channel sizes are now the
|
|
|
|
// remaining channel balance.
|
|
|
|
for _, candidate := range scores {
|
|
|
|
if candidate.Score == 0 {
|
|
|
|
t1.Fatalf("Expected non-zero score")
|
|
|
|
}
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
// TestPrefAttachmentSelectSkipNodes ensures that if a node was
|
2018-11-23 01:18:09 +03:00
|
|
|
// already selected as a channel counterparty, then that node will get a score
|
|
|
|
// of zero during scoring.
|
2018-12-19 16:54:54 +03:00
|
|
|
func TestPrefAttachmentSelectSkipNodes(t *testing.T) {
|
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 07:14:41 +03:00
|
|
|
t.Parallel()
|
|
|
|
|
|
|
|
prand.Seed(time.Now().Unix())
|
|
|
|
|
|
|
|
const (
|
|
|
|
maxChanSize = btcutil.Amount(btcutil.SatoshiPerBitcoin)
|
2018-12-19 16:54:53 +03:00
|
|
|
)
|
2018-11-23 01:18:08 +03:00
|
|
|
|
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 07:14:41 +03:00
|
|
|
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()
|
|
|
|
}
|
|
|
|
|
2018-12-19 16:54:54 +03:00
|
|
|
prefAttach := NewPrefAttachment()
|
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 07:14:41 +03:00
|
|
|
|
|
|
|
// Next, we'll create a simple topology of two nodes,
|
|
|
|
// with a single channel connecting them.
|
2018-02-07 06:11:11 +03:00
|
|
|
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 07:14:41 +03:00
|
|
|
_, _, err = graph.addRandChannel(nil, nil,
|
2018-02-07 06:11:11 +03:00
|
|
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to create channel: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
nodes := make(map[NodeID]struct{})
|
|
|
|
if err := graph.ForEachNode(func(n Node) error {
|
|
|
|
nodes[n.PubKey()] = struct{}{}
|
|
|
|
return nil
|
|
|
|
}); err != nil {
|
|
|
|
t1.Fatalf("unable to traverse graph: %v", err)
|
|
|
|
}
|
|
|
|
|
|
|
|
if len(nodes) != 2 {
|
|
|
|
t1.Fatalf("expected 2 nodes, found %d", len(nodes))
|
|
|
|
}
|
|
|
|
|
|
|
|
// With our graph created, we'll now get the scores for
|
|
|
|
// all nodes in the graph.
|
|
|
|
scores, err := prefAttach.NodeScores(graph, nil,
|
2018-12-19 16:54:53 +03:00
|
|
|
maxChanSize, nodes)
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
if len(scores) != len(nodes) {
|
|
|
|
t1.Fatalf("all nodes should be scored, "+
|
|
|
|
"instead %v were", len(scores))
|
|
|
|
}
|
|
|
|
|
|
|
|
// THey should all have a score, and a maxChanSize
|
|
|
|
// channel size recommendation.
|
|
|
|
for _, candidate := range scores {
|
|
|
|
if candidate.Score == 0 {
|
|
|
|
t1.Fatalf("Expected non-zero score")
|
|
|
|
}
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
// We'll simulate a channel update by adding the nodes
|
2018-11-23 01:18:09 +03:00
|
|
|
// to our set of channels.
|
|
|
|
var chans []Channel
|
|
|
|
for _, candidate := range scores {
|
|
|
|
chans = append(chans,
|
|
|
|
Channel{
|
|
|
|
Node: candidate.NodeID,
|
|
|
|
},
|
|
|
|
)
|
|
|
|
}
|
|
|
|
|
|
|
|
// If we attempt to make a call to the NodeScores
|
|
|
|
// function, without providing any new information,
|
|
|
|
// then all nodes should have a score of zero, since we
|
|
|
|
// already got channels to them.
|
|
|
|
scores, err = prefAttach.NodeScores(graph, chans,
|
2018-12-19 16:54:53 +03:00
|
|
|
maxChanSize, nodes)
|
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 07:14:41 +03:00
|
|
|
if err != nil {
|
|
|
|
t1.Fatalf("unable to select attachment "+
|
|
|
|
"directives: %v", err)
|
|
|
|
}
|
|
|
|
|
2018-11-23 01:18:09 +03:00
|
|
|
// Since all should be given a score of 0, the map
|
|
|
|
// should be empty.
|
|
|
|
if len(scores) != 0 {
|
|
|
|
t1.Fatalf("expected empty score map, "+
|
|
|
|
"instead got %v ", len(scores))
|
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 07:14:41 +03:00
|
|
|
}
|
|
|
|
})
|
|
|
|
if !success {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|