erigon-pulse/p2p/dial.go

576 lines
16 KiB
Go
Raw Normal View History

2015-07-07 00:54:22 +00:00
// Copyright 2015 The go-ethereum Authors
// This file is part of the go-ethereum library.
2015-07-07 00:54:22 +00:00
//
// The go-ethereum library is free software: you can redistribute it and/or modify
2015-07-07 00:54:22 +00:00
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
2015-07-07 00:54:22 +00:00
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2015-07-07 00:54:22 +00:00
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
2015-07-07 00:54:22 +00:00
package p2p
import (
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
"context"
crand "crypto/rand"
"encoding/binary"
"errors"
"fmt"
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
mrand "math/rand"
"net"
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
"sync"
"sync/atomic"
"time"
"github.com/ledgerwatch/erigon/common/debug"
"github.com/ledgerwatch/erigon/common/mclock"
"github.com/ledgerwatch/erigon/p2p/enode"
"github.com/ledgerwatch/erigon/p2p/netutil"
2021-07-29 10:23:23 +00:00
"github.com/ledgerwatch/log/v3"
)
const (
// This is the amount of time spent waiting in between redialing a certain node. The
// limit is a bit higher than inboundThrottleTime to prevent failing dials in small
// private networks.
dialHistoryExpiration = inboundThrottleTime + 5*time.Second
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// Config for the "Looking for peers" message.
dialStatsLogInterval = 60 * time.Second // printed at most this often
// Endpoint resolution is throttled with bounded backoff.
initialResolveDelay = 60 * time.Second
maxResolveDelay = time.Hour
)
// NodeDialer is used to connect to nodes in the network, typically by using
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// an underlying net.Dialer but also using net.Pipe in tests.
type NodeDialer interface {
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
Dial(context.Context, *enode.Node) (net.Conn, error)
}
type nodeResolver interface {
Resolve(*enode.Node) *enode.Node
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// tcpDialer implements NodeDialer using real TCP connections.
type tcpDialer struct {
d *net.Dialer
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func (t tcpDialer) Dial(ctx context.Context, dest *enode.Node) (net.Conn, error) {
return t.d.DialContext(ctx, "tcp", nodeAddr(dest).String())
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func nodeAddr(n *enode.Node) net.Addr {
return &net.TCPAddr{IP: n.IP(), Port: n.TCP()}
}
// checkDial errors:
var (
errSelf = errors.New("is self")
errAlreadyDialing = errors.New("already dialing")
errAlreadyConnected = errors.New("already connected")
errRecentlyDialed = errors.New("recently dialed")
errNotWhitelisted = errors.New("not contained in netrestrict whitelist")
errNoPort = errors.New("node does not provide TCP port")
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
)
// dialer creates outbound connections and submits them into Server.
// Two types of peer connections can be created:
//
2022-08-10 12:04:13 +00:00
// - static dials are pre-configured connections. The dialer attempts
// keep these nodes connected at all times.
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
//
2022-08-10 12:04:13 +00:00
// - dynamic dials are created from node discovery results. The dialer
// continuously reads candidate nodes from its input iterator and attempts
// to create peer connections to nodes arriving through the iterator.
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
type dialScheduler struct {
dialConfig
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
mutex sync.Mutex
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
setupFunc dialSetupFunc
wg sync.WaitGroup
cancel context.CancelFunc
ctx context.Context
nodesIn chan *enode.Node
doneCh chan *dialTask
addStaticCh chan *enode.Node
remStaticCh chan *enode.Node
addPeerCh chan *conn
remPeerCh chan *conn
2021-07-03 03:35:11 +00:00
subProtocolVersion uint
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// Everything below here belongs to loop and
// should only be accessed by code on the loop goroutine.
dialing map[enode.ID]*dialTask // active tasks
peers map[enode.ID]connFlag // all connected peers
dialPeers int // current number of dialed peers
// The static map tracks all static dial tasks. The subset of usable static dial tasks
// (i.e. those passing checkDial) is kept in staticPool. The scheduler prefers
// launching random static tasks from the pool over launching dynamic dials from the
// iterator.
static map[enode.ID]*dialTask
staticPool []*dialTask
// The dial history keeps recently dialed nodes. Members of history are not dialed.
history expHeap
historyTimer mclock.Timer
historyTimerTime mclock.AbsTime
// for logStats
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
dialed int
errors map[string]uint
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
type dialSetupFunc func(net.Conn, connFlag, *enode.Node) error
type dialConfig struct {
self enode.ID // our own ID
maxDialPeers int // maximum number of dialed peers
maxActiveDials int // maximum number of active dials
netRestrict *netutil.Netlist // IP whitelist, disabled if nil
resolver nodeResolver
dialer NodeDialer
log log.Logger
clock mclock.Clock
rand *mrand.Rand
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func (cfg dialConfig) withDefaults() dialConfig {
if cfg.log == nil {
cfg.log = log.Root()
}
if cfg.clock == nil {
cfg.clock = mclock.System{}
}
if cfg.rand == nil {
seedb := make([]byte, 8)
2020-02-21 15:25:11 +00:00
if _, err := crand.Read(seedb); err != nil {
panic(err)
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
seed := int64(binary.BigEndian.Uint64(seedb))
cfg.rand = mrand.New(mrand.NewSource(seed)) // nolint: gosec
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
return cfg
}
2021-07-03 03:35:11 +00:00
func newDialScheduler(config dialConfig, it enode.Iterator, setupFunc dialSetupFunc, subProtocolVersion uint) *dialScheduler {
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d := &dialScheduler{
dialConfig: config.withDefaults(),
setupFunc: setupFunc,
dialing: make(map[enode.ID]*dialTask),
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
static: make(map[enode.ID]*dialTask),
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
peers: make(map[enode.ID]connFlag),
doneCh: make(chan *dialTask),
nodesIn: make(chan *enode.Node),
addStaticCh: make(chan *enode.Node),
remStaticCh: make(chan *enode.Node),
addPeerCh: make(chan *conn),
remPeerCh: make(chan *conn),
2021-07-03 03:35:11 +00:00
subProtocolVersion: subProtocolVersion,
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
errors: map[string]uint{},
}
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d.ctx, d.cancel = context.WithCancel(context.Background())
d.wg.Add(2)
go d.readNodes(it)
go d.loop(it)
return d
}
// stop shuts down the dialer, canceling all current dial tasks.
func (d *dialScheduler) stop() {
d.cancel()
d.wg.Wait()
}
// addStatic adds a static dial candidate.
func (d *dialScheduler) addStatic(n *enode.Node) {
select {
case d.addStaticCh <- n:
case <-d.ctx.Done():
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
// removeStatic removes a static dial candidate.
func (d *dialScheduler) removeStatic(n *enode.Node) {
select {
case d.remStaticCh <- n:
case <-d.ctx.Done():
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// peerAdded updates the peer set.
func (d *dialScheduler) peerAdded(c *conn) {
select {
case d.addPeerCh <- c:
case <-d.ctx.Done():
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// peerRemoved updates the peer set.
func (d *dialScheduler) peerRemoved(c *conn) {
select {
case d.remPeerCh <- c:
case <-d.ctx.Done():
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// loop is the main loop of the dialer.
func (d *dialScheduler) loop(it enode.Iterator) {
2021-06-22 10:09:45 +00:00
defer debug.LogPanic()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
var (
nodesCh chan *enode.Node
historyExp = make(chan struct{}, 1)
)
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
logTimer := time.NewTicker(dialStatsLogInterval)
defer logTimer.Stop()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
loop:
for {
// Launch new dials if slots are available.
slots := d.freeDialSlots()
d.startStaticDials()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
if slots > 0 {
nodesCh = d.nodesIn
} else {
nodesCh = nil
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d.rearmHistoryTimer(historyExp)
select {
case <-d.ctx.Done():
it.Close()
break loop
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
case <-logTimer.C:
d.logStats()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
case node := <-nodesCh:
if err := d.checkDial(node); err != nil {
d.log.Trace("Discarding dial candidate", "id", node.ID(), "ip", node.IP(), "reason", err)
} else {
d.startDial(newDialTask(node, dynDialedConn))
}
case task := <-d.doneCh:
id := task.dest.ID()
delete(d.dialing, id)
d.updateStaticPool(id)
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
d.dialed++
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
case c := <-d.addPeerCh:
if c.is(dynDialedConn) || c.is(staticDialedConn) {
d.dialPeers++
}
id := c.node.ID()
d.peers[id] = connFlag(atomic.LoadInt32((*int32)(&c.flags)))
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// Remove from static pool because the node is now connected.
task := d.static[id]
if task != nil && task.staticPoolIndex >= 0 {
d.removeFromStaticPool(task.staticPoolIndex)
}
// TODO: cancel dials to connected peers
case c := <-d.remPeerCh:
if c.is(dynDialedConn) || c.is(staticDialedConn) {
d.dialPeers--
}
delete(d.peers, c.node.ID())
d.updateStaticPool(c.node.ID())
case node := <-d.addStaticCh:
id := node.ID()
_, exists := d.static[id]
d.log.Trace("Adding static node", "id", id, "ip", node.IP(), "added", !exists)
if exists {
continue loop
}
task := newDialTask(node, staticDialedConn)
d.static[id] = task
if d.checkDial(node) == nil {
d.addToStaticPool(task)
}
case node := <-d.remStaticCh:
id := node.ID()
task := d.static[id]
d.log.Trace("Removing static node", "id", id, "ok", task != nil)
if task != nil {
delete(d.static, id)
if task.staticPoolIndex >= 0 {
d.removeFromStaticPool(task.staticPoolIndex)
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
case <-historyExp:
d.expireHistory()
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d.stopHistoryTimer(historyExp)
for range d.dialing {
<-d.doneCh
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d.wg.Done()
}
// readNodes runs in its own goroutine and delivers nodes from
// the input iterator to the nodesIn channel.
func (d *dialScheduler) readNodes(it enode.Iterator) {
2021-06-22 10:09:45 +00:00
defer debug.LogPanic()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
defer d.wg.Done()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
for it.Next() {
select {
case d.nodesIn <- it.Node():
case <-d.ctx.Done():
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// or comes back online.
2022-08-10 12:04:13 +00:00
// nolint
func (d *dialScheduler) logStats() {
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
vals := []interface{}{"protocol", d.subProtocolVersion,
"peers", fmt.Sprintf("%d/%d", len(d.peers), d.maxDialPeers), "tried", d.dialed, "static", len(d.static)}
d.mutex.Lock()
for err, count := range d.errors {
vals = append(vals, err, count)
}
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
d.mutex.Unlock()
d.log.Debug("[p2p] Dial scheduler", vals...)
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// rearmHistoryTimer configures d.historyTimer to fire when the
// next item in d.history expires.
func (d *dialScheduler) rearmHistoryTimer(ch chan struct{}) {
if len(d.history) == 0 || d.historyTimerTime == d.history.nextExpiry() {
return
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d.stopHistoryTimer(ch)
d.historyTimerTime = d.history.nextExpiry()
timeout := time.Duration(d.historyTimerTime - d.clock.Now())
d.historyTimer = d.clock.AfterFunc(timeout, func() { ch <- struct{}{} })
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// stopHistoryTimer stops the timer and drains the channel it sends on.
func (d *dialScheduler) stopHistoryTimer(ch chan struct{}) {
if d.historyTimer != nil && !d.historyTimer.Stop() {
<-ch
}
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// expireHistory removes expired items from d.history.
func (d *dialScheduler) expireHistory() {
d.historyTimer.Stop()
d.historyTimer = nil
d.historyTimerTime = 0
d.history.expire(d.clock.Now(), func(hkey string) {
var id enode.ID
copy(id[:], hkey)
d.updateStaticPool(id)
})
}
// freeDialSlots returns the number of free dial slots. The result can be negative
// when peers are connected while their task is still running.
func (d *dialScheduler) freeDialSlots() int {
slots := (d.maxDialPeers - d.dialPeers) * 2
if slots > d.maxActiveDials {
slots = d.maxActiveDials
}
free := slots - len(d.dialing)
return free
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// checkDial returns an error if node n should not be dialed.
func (d *dialScheduler) checkDial(n *enode.Node) error {
if n.ID() == d.self {
return errSelf
}
if n.IP() != nil && n.TCP() == 0 {
// This check can trigger if a non-TCP node is found
// by discovery. If there is no IP, the node is a static
// node and the actual endpoint will be resolved later in dialTask.
return errNoPort
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
if _, ok := d.dialing[n.ID()]; ok {
return errAlreadyDialing
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
if _, ok := d.peers[n.ID()]; ok {
return errAlreadyConnected
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
if d.netRestrict != nil && !d.netRestrict.Contains(n.IP()) {
return errNotWhitelisted
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
}
if d.history.contains(string(n.ID().Bytes())) {
return errRecentlyDialed
}
return nil
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// startStaticDials starts n static dial tasks.
func (d *dialScheduler) startStaticDials() {
for len(d.staticPool) > 0 {
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
idx := d.rand.Intn(len(d.staticPool))
task := d.staticPool[idx]
d.startDial(task)
d.removeFromStaticPool(idx)
}
}
// updateStaticPool attempts to move the given static dial back into staticPool.
func (d *dialScheduler) updateStaticPool(id enode.ID) {
task, ok := d.static[id]
if ok && task.staticPoolIndex < 0 && d.checkDial(task.dest) == nil {
d.addToStaticPool(task)
}
}
func (d *dialScheduler) addToStaticPool(task *dialTask) {
if task.staticPoolIndex >= 0 {
panic("attempt to add task to staticPool twice")
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
d.staticPool = append(d.staticPool, task)
task.staticPoolIndex = len(d.staticPool) - 1
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
// removeFromStaticPool removes the task at idx from staticPool. It does that by moving the
// current last element of the pool to idx and then shortening the pool by one.
func (d *dialScheduler) removeFromStaticPool(idx int) {
task := d.staticPool[idx]
end := len(d.staticPool) - 1
d.staticPool[idx] = d.staticPool[end]
d.staticPool[idx].staticPoolIndex = idx
d.staticPool[end] = nil
d.staticPool = d.staticPool[:end]
task.staticPoolIndex = -1
}
// startDial runs the given dial task in a separate goroutine.
func (d *dialScheduler) startDial(task *dialTask) {
d.log.Trace("Starting p2p dial", "id", task.dest.ID(), "ip", task.dest.IP(), "flag", task.flags)
hkey := string(task.dest.ID().Bytes())
d.history.add(hkey, d.clock.Now().Add(dialHistoryExpiration))
d.dialing[task.dest.ID()] = task
go func() {
2021-06-22 10:09:45 +00:00
defer debug.LogPanic()
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
task.run(d)
d.doneCh <- task
}()
}
// A dialTask generated for each node that is dialed.
type dialTask struct {
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
staticPoolIndex int
flags connFlag
// These fields are private to the task and should not be
// accessed by dialScheduler while the task is running.
dest *enode.Node
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
lastResolved mclock.AbsTime
resolveDelay time.Duration
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func newDialTask(dest *enode.Node, flags connFlag) *dialTask {
return &dialTask{dest: dest, flags: flags, staticPoolIndex: -1}
}
type dialError struct {
error
}
func (t *dialTask) run(d *dialScheduler) {
if t.needResolve() && !t.resolve(d) {
return
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
err := t.dial(d, t.dest)
if err != nil {
// For static nodes, resolve one more time if dialing fails.
if _, ok := err.(*dialError); ok && t.flags&staticDialedConn != 0 {
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
if t.resolve(d) {
2020-02-21 15:25:11 +00:00
t.dial(d, t.dest) //nolint:errcheck
}
}
}
}
func (t *dialTask) needResolve() bool {
return t.flags&staticDialedConn != 0 && t.dest.IP() == nil
}
// resolve attempts to find the current endpoint for the destination
// using discovery.
//
// Resolve operations are throttled with backoff to avoid flooding the
// discovery network with useless queries for nodes that don't exist.
// The backoff delay resets when the node is found.
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func (t *dialTask) resolve(d *dialScheduler) bool {
if d.resolver == nil {
return false
}
if t.resolveDelay == 0 {
t.resolveDelay = initialResolveDelay
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
if t.lastResolved > 0 && time.Duration(d.clock.Now()-t.lastResolved) < t.resolveDelay {
return false
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
resolved := d.resolver.Resolve(t.dest)
t.lastResolved = d.clock.Now()
if resolved == nil {
t.resolveDelay *= 2
if t.resolveDelay > maxResolveDelay {
t.resolveDelay = maxResolveDelay
}
d.log.Warn("Resolving node failed", "id", t.dest.ID(), "newdelay", t.resolveDelay)
return false
}
// The node was found.
t.resolveDelay = initialResolveDelay
t.dest = resolved
d.log.Trace("Resolved node", "id", t.dest.ID(), "addr", &net.TCPAddr{IP: t.dest.IP(), Port: t.dest.TCP()})
return true
}
// dial performs the actual connection attempt.
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func (t *dialTask) dial(d *dialScheduler, dest *enode.Node) error {
fd, err := d.dialer.Dial(d.ctx, t.dest)
if err != nil {
Discovery zero refresh timer (#8661) This fixes an issue where the mumbai testnet node struggle to find peers. Before this fix in general test peer numbers are typically around 20 in total between eth66, eth67 and eth68. For new peers some can struggle to find even a single peer after days of operation. These are the numbers after 12 hours or running on a node which previously could not find any peers: eth66=13, eth67=76, eth68=91. The root cause of this issue is the following: - A significant number of mumbai peers around the boot node return network ids which are different from those currently available in the DHT - The available nodes are all consequently busy and return 'too many peers' for long periods These issues case a significant number of discovery timeouts, some of the queries will never receive a response. This causes the discovery read loop to enter a channel deadlock - which means that no responses are processed, nor timeouts fired. This causes the discovery process in the node to stop. From then on it just re-requests handshakes from a relatively small number of peers. This check in fixes this situation with the following changes: - Remove the deadlock by running the timer in a separate go-routine so it can run independently of the main request processing. - Allow the discovery process matcher to match on port if no id match can be established on initial ping. This allows subsequent node validation to proceed and if the node proves to be valid via the remainder of the look-up and handshake process it us used as a valid peer. - Completely unsolicited responses, i.e. those which come from a completely unknown ip:port combination continue to be ignored. -
2023-11-07 08:48:58 +00:00
cleanErr := cleanupDialErr(err)
d.log.Trace("Dial error", "id", t.dest.ID(), "addr", nodeAddr(t.dest), "conn", t.flags, "err", cleanErr)
d.mutex.Lock()
d.errors[cleanErr.Error()] = d.errors[cleanErr.Error()] + 1
d.mutex.Unlock()
return &dialError{err}
}
mfd := newMeteredConn(fd, false, &net.TCPAddr{IP: dest.IP(), Port: dest.TCP()})
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
return d.setupFunc(mfd, t.flags, dest)
}
func (t *dialTask) String() string {
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
id := t.dest.ID()
return fmt.Sprintf("%v %x %v:%d", t.flags, id[:8], t.dest.IP(), t.dest.TCP())
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
func cleanupDialErr(err error) error {
if netErr, ok := err.(*net.OpError); ok && netErr.Op == "dial" {
return netErr.Err
}
p2p: new dial scheduler (#20592) * p2p: new dial scheduler This change replaces the peer-to-peer dial scheduler with a new and improved implementation. The new code is better than the previous implementation in two key aspects: - The time between discovery of a node and dialing that node is significantly lower in the new version. The old dialState kept a buffer of nodes and launched a task to refill it whenever the buffer became empty. This worked well with the discovery interface we used to have, but doesn't really work with the new iterator-based discovery API. - Selection of static dial candidates (created by Server.AddPeer or through static-nodes.json) performs much better for large amounts of static peers. Connections to static nodes are now limited like dynanic dials and can no longer overstep MaxPeers or the dial ratio. * p2p/simulations/adapters: adapt to new NodeDialer interface * p2p: re-add check for self in checkDial * p2p: remove peersetCh * p2p: allow static dials when discovery is disabled * p2p: add test for dialScheduler.removeStatic * p2p: remove blank line * p2p: fix documentation of maxDialPeers * p2p: change "ok" to "added" in static node log * p2p: improve dialTask docs Also increase log level for "Can't resolve node" * p2p: ensure dial resolver is truly nil without discovery * p2p: add "looking for peers" log message * p2p: clean up Server.run comments * p2p: fix maxDialedConns for maxpeers < dialRatio Always allocate at least one dial slot unless dialing is disabled using NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to dedicate the sole slot to dialing instead of listening. * p2p: fix RemovePeer to disconnect the peer again Also make RemovePeer synchronous and add a test. * p2p: remove "Connection set up" log message * p2p: clean up connection logging We previously logged outgoing connection failures up to three times. - in SetupConn() as "Setting up connection failed addr=..." - in setupConn() with an error-specific message and "id=... addr=..." - in dial() as "Dial error task=..." This commit ensures a single log message is emitted per failure and adds "id=... addr=... conn=..." everywhere (id= omitted when the ID isn't known yet). Also avoid printing a log message when a static dial fails but can't be resolved because discv4 is disabled. The light client hit this case all the time, increasing the message count to four lines per failed connection. * p2p: document that RemovePeer blocks
2020-02-13 10:10:03 +00:00
return err
}