Official Go implementation of the Ethereum protocol
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
go-ethereum/p2p/dial_test.go

673 lines
17 KiB

// Copyright 2015 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// 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,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// 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/>.
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
5 years ago
"context"
"errors"
"fmt"
"math/rand"
"net"
"reflect"
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
5 years ago
"sync"
"testing"
"time"
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
5 years ago
"github.com/ethereum/go-ethereum/common/mclock"
"github.com/ethereum/go-ethereum/internal/testlog"
"github.com/ethereum/go-ethereum/log"
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.
6 years ago
"github.com/ethereum/go-ethereum/p2p/enode"
"github.com/ethereum/go-ethereum/p2p/netutil"
)
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
5 years ago
// This test checks that dynamic dials are launched from discovery results.
func TestDialSchedDynDial(t *testing.T) {
t.Parallel()
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
5 years ago
config := dialConfig{
maxActiveDials: 5,
maxDialPeers: 4,
}
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
5 years ago
runDialTest(t, config, []dialTestRound{
// 3 out of 4 peers are connected, leaving 2 dial slots.
// 9 nodes are discovered, but only 2 are dialed.
{
peersAdded: []*conn{
{flags: staticDialedConn, node: newNode(uintID(0x00), "")},
{flags: dynDialedConn, node: newNode(uintID(0x01), "")},
{flags: dynDialedConn, node: newNode(uintID(0x02), "")},
},
discovered: []*enode.Node{
newNode(uintID(0x00), "127.0.0.1:30303"), // not dialed because already connected as static peer
newNode(uintID(0x02), "127.0.0.1:30303"), // ...
newNode(uintID(0x03), "127.0.0.1:30303"),
newNode(uintID(0x04), "127.0.0.1:30303"),
newNode(uintID(0x05), "127.0.0.1:30303"), // not dialed because there are only two slots
newNode(uintID(0x06), "127.0.0.1:30303"), // ...
newNode(uintID(0x07), "127.0.0.1:30303"), // ...
newNode(uintID(0x08), "127.0.0.1:30303"), // ...
},
wantNewDials: []*enode.Node{
newNode(uintID(0x03), "127.0.0.1:30303"),
newNode(uintID(0x04), "127.0.0.1:30303"),
},
},
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
5 years ago
// One dial completes, freeing one dial slot.
{
failed: []enode.ID{
uintID(0x04),
},
wantNewDials: []*enode.Node{
newNode(uintID(0x05), "127.0.0.1:30303"),
},
},
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
5 years ago
// Dial to 0x03 completes, filling the last remaining peer slot.
{
succeeded: []enode.ID{
uintID(0x03),
},
failed: []enode.ID{
uintID(0x05),
},
discovered: []*enode.Node{
newNode(uintID(0x09), "127.0.0.1:30303"), // not dialed because there are no free slots
},
},
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
5 years ago
// 3 peers drop off, creating 6 dial slots. Check that 5 of those slots
// (i.e. up to maxActiveDialTasks) are used.
{
peersRemoved: []enode.ID{
uintID(0x00),
uintID(0x01),
uintID(0x02),
},
discovered: []*enode.Node{
newNode(uintID(0x0a), "127.0.0.1:30303"),
newNode(uintID(0x0b), "127.0.0.1:30303"),
newNode(uintID(0x0c), "127.0.0.1:30303"),
newNode(uintID(0x0d), "127.0.0.1:30303"),
newNode(uintID(0x0f), "127.0.0.1:30303"),
},
wantNewDials: []*enode.Node{
newNode(uintID(0x06), "127.0.0.1:30303"),
newNode(uintID(0x07), "127.0.0.1:30303"),
newNode(uintID(0x08), "127.0.0.1:30303"),
newNode(uintID(0x09), "127.0.0.1:30303"),
newNode(uintID(0x0a), "127.0.0.1:30303"),
},
},
})
}
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
5 years ago
// This test checks that candidates that do not match the netrestrict list are not dialed.
func TestDialSchedNetRestrict(t *testing.T) {
t.Parallel()
nodes := []*enode.Node{
newNode(uintID(0x01), "127.0.0.1:30303"),
newNode(uintID(0x02), "127.0.0.2:30303"),
newNode(uintID(0x03), "127.0.0.3:30303"),
newNode(uintID(0x04), "127.0.0.4:30303"),
newNode(uintID(0x05), "127.0.2.5:30303"),
newNode(uintID(0x06), "127.0.2.6:30303"),
newNode(uintID(0x07), "127.0.2.7:30303"),
newNode(uintID(0x08), "127.0.2.8:30303"),
}
config := dialConfig{
netRestrict: new(netutil.Netlist),
maxActiveDials: 10,
maxDialPeers: 10,
}
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
5 years ago
config.netRestrict.Add("127.0.2.0/24")
runDialTest(t, config, []dialTestRound{
{
discovered: nodes,
wantNewDials: nodes[4:8],
},
{
succeeded: []enode.ID{
nodes[4].ID(),
nodes[5].ID(),
nodes[6].ID(),
nodes[7].ID(),
},
},
})
}
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
5 years ago
// This test checks that static dials work and obey the limits.
func TestDialSchedStaticDial(t *testing.T) {
t.Parallel()
config := dialConfig{
maxActiveDials: 5,
maxDialPeers: 4,
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.
6 years ago
}
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
5 years ago
runDialTest(t, config, []dialTestRound{
// Static dials are launched for the nodes that
// aren't yet connected.
{
peersAdded: []*conn{
{flags: dynDialedConn, node: newNode(uintID(0x01), "127.0.0.1:30303")},
{flags: dynDialedConn, node: newNode(uintID(0x02), "127.0.0.2:30303")},
},
update: func(d *dialScheduler) {
// These two are not dialed because they're already connected
// as dynamic peers.
d.addStatic(newNode(uintID(0x01), "127.0.0.1:30303"))
d.addStatic(newNode(uintID(0x02), "127.0.0.2:30303"))
// These nodes will be dialed:
d.addStatic(newNode(uintID(0x03), "127.0.0.3:30303"))
d.addStatic(newNode(uintID(0x04), "127.0.0.4:30303"))
d.addStatic(newNode(uintID(0x05), "127.0.0.5:30303"))
d.addStatic(newNode(uintID(0x06), "127.0.0.6:30303"))
d.addStatic(newNode(uintID(0x07), "127.0.0.7:30303"))
d.addStatic(newNode(uintID(0x08), "127.0.0.8:30303"))
d.addStatic(newNode(uintID(0x09), "127.0.0.9:30303"))
},
wantNewDials: []*enode.Node{
newNode(uintID(0x03), "127.0.0.3:30303"),
newNode(uintID(0x04), "127.0.0.4:30303"),
newNode(uintID(0x05), "127.0.0.5:30303"),
newNode(uintID(0x06), "127.0.0.6:30303"),
},
},
// Dial to 0x03 completes, filling a peer slot. One slot remains,
// two dials are launched to attempt to fill it.
{
succeeded: []enode.ID{
uintID(0x03),
},
failed: []enode.ID{
uintID(0x04),
uintID(0x05),
uintID(0x06),
},
wantResolves: map[enode.ID]*enode.Node{
uintID(0x04): nil,
uintID(0x05): nil,
uintID(0x06): nil,
},
wantNewDials: []*enode.Node{
newNode(uintID(0x08), "127.0.0.8:30303"),
newNode(uintID(0x09), "127.0.0.9:30303"),
},
},
// Peer 0x01 drops and 0x07 connects as inbound peer.
// Only 0x01 is dialed.
{
peersAdded: []*conn{
{flags: inboundConn, node: newNode(uintID(0x07), "127.0.0.7:30303")},
},
peersRemoved: []enode.ID{
uintID(0x01),
},
wantNewDials: []*enode.Node{
newNode(uintID(0x01), "127.0.0.1:30303"),
},
},
})
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.
6 years ago
}
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
5 years ago
// This test checks that removing static nodes stops connecting to them.
func TestDialSchedRemoveStatic(t *testing.T) {
t.Parallel()
config := dialConfig{
maxActiveDials: 1,
maxDialPeers: 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
5 years ago
runDialTest(t, config, []dialTestRound{
// Add static nodes.
{
update: func(d *dialScheduler) {
d.addStatic(newNode(uintID(0x01), "127.0.0.1:30303"))
d.addStatic(newNode(uintID(0x02), "127.0.0.2:30303"))
d.addStatic(newNode(uintID(0x03), "127.0.0.3:30303"))
},
wantNewDials: []*enode.Node{
newNode(uintID(0x01), "127.0.0.1:30303"),
},
},
// Dial to 0x01 fails.
{
failed: []enode.ID{
uintID(0x01),
},
wantResolves: map[enode.ID]*enode.Node{
uintID(0x01): nil,
},
wantNewDials: []*enode.Node{
newNode(uintID(0x02), "127.0.0.2:30303"),
},
},
// All static nodes are removed. 0x01 is in history, 0x02 is being
// dialed, 0x03 is in staticPool.
{
update: func(d *dialScheduler) {
d.removeStatic(newNode(uintID(0x01), "127.0.0.1:30303"))
d.removeStatic(newNode(uintID(0x02), "127.0.0.2:30303"))
d.removeStatic(newNode(uintID(0x03), "127.0.0.3:30303"))
},
failed: []enode.ID{
uintID(0x02),
},
wantResolves: map[enode.ID]*enode.Node{
uintID(0x02): 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
5 years ago
// Since all static nodes are removed, they should not be dialed again.
{}, {}, {},
})
}
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
5 years ago
// This test checks that static dials are selected at random.
func TestDialSchedManyStaticNodes(t *testing.T) {
t.Parallel()
config := dialConfig{maxDialPeers: 2}
runDialTest(t, config, []dialTestRound{
{
peersAdded: []*conn{
{flags: dynDialedConn, node: newNode(uintID(0xFFFE), "")},
{flags: dynDialedConn, node: newNode(uintID(0xFFFF), "")},
},
update: func(d *dialScheduler) {
for id := uint16(0); id < 2000; id++ {
n := newNode(uintID(id), "127.0.0.1:30303")
d.addStatic(n)
}
},
},
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
5 years ago
{
peersRemoved: []enode.ID{
uintID(0xFFFE),
uintID(0xFFFF),
},
wantNewDials: []*enode.Node{
newNode(uintID(0x0085), "127.0.0.1:30303"),
newNode(uintID(0x02dc), "127.0.0.1:30303"),
newNode(uintID(0x0285), "127.0.0.1:30303"),
newNode(uintID(0x00cb), "127.0.0.1:30303"),
},
},
})
}
// This test checks that past dials are not retried for some time.
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
5 years ago
func TestDialSchedHistory(t *testing.T) {
t.Parallel()
config := dialConfig{
maxActiveDials: 3,
maxDialPeers: 3,
}
runDialTest(t, config, []dialTestRound{
{
update: func(d *dialScheduler) {
d.addStatic(newNode(uintID(0x01), "127.0.0.1:30303"))
d.addStatic(newNode(uintID(0x02), "127.0.0.2:30303"))
d.addStatic(newNode(uintID(0x03), "127.0.0.3:30303"))
},
wantNewDials: []*enode.Node{
newNode(uintID(0x01), "127.0.0.1:30303"),
newNode(uintID(0x02), "127.0.0.2:30303"),
newNode(uintID(0x03), "127.0.0.3:30303"),
},
},
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
5 years ago
// No new tasks are launched in this round because all static
// nodes are either connected or still being dialed.
{
succeeded: []enode.ID{
uintID(0x01),
uintID(0x02),
},
failed: []enode.ID{
uintID(0x03),
},
wantResolves: map[enode.ID]*enode.Node{
uintID(0x03): nil,
},
},
// Nothing happens in this round because we're waiting for
// node 0x3's history entry to expire.
{},
// The cache entry for node 0x03 has expired and is retried.
{
wantNewDials: []*enode.Node{
newNode(uintID(0x03), "127.0.0.3:30303"),
},
},
})
}
func TestDialSchedResolve(t *testing.T) {
t.Parallel()
config := dialConfig{
maxActiveDials: 1,
maxDialPeers: 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
5 years ago
node := newNode(uintID(0x01), "")
resolved := newNode(uintID(0x01), "127.0.0.1:30303")
resolved2 := newNode(uintID(0x01), "127.0.0.55:30303")
runDialTest(t, config, []dialTestRound{
{
update: func(d *dialScheduler) {
d.addStatic(node)
},
wantResolves: map[enode.ID]*enode.Node{
uintID(0x01): resolved,
},
wantNewDials: []*enode.Node{
resolved,
},
},
{
failed: []enode.ID{
uintID(0x01),
},
wantResolves: map[enode.ID]*enode.Node{
uintID(0x01): resolved2,
},
wantNewDials: []*enode.Node{
resolved2,
},
},
})
}
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
5 years ago
// -------
// Code below here is the framework for the tests above.
type dialTestRound struct {
peersAdded []*conn
peersRemoved []enode.ID
update func(*dialScheduler) // called at beginning of round
discovered []*enode.Node // newly discovered nodes
succeeded []enode.ID // dials which succeed this round
failed []enode.ID // dials which fail this round
wantResolves map[enode.ID]*enode.Node
wantNewDials []*enode.Node // dials that should be launched in this round
}
func runDialTest(t *testing.T, config dialConfig, rounds []dialTestRound) {
var (
clock = new(mclock.Simulated)
iterator = newDialTestIterator()
dialer = newDialTestDialer()
resolver = new(dialTestResolver)
peers = make(map[enode.ID]*conn)
setupCh = make(chan *conn)
)
// Override config.
config.clock = clock
config.dialer = dialer
config.resolver = resolver
config.log = testlog.Logger(t, log.LvlTrace)
config.rand = rand.New(rand.NewSource(0x1111))
// Set up the dialer. The setup function below runs on the dialTask
// goroutine and adds the peer.
var dialsched *dialScheduler
setup := func(fd net.Conn, f connFlag, node *enode.Node) error {
conn := &conn{flags: f, node: node}
dialsched.peerAdded(conn)
setupCh <- conn
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
5 years ago
dialsched = newDialScheduler(config, iterator, setup)
defer dialsched.stop()
for i, round := range rounds {
// Apply peer set updates.
for _, c := range round.peersAdded {
if peers[c.node.ID()] != nil {
t.Fatalf("round %d: peer %v already connected", i, c.node.ID())
}
dialsched.peerAdded(c)
peers[c.node.ID()] = c
}
for _, id := range round.peersRemoved {
c := peers[id]
if c == nil {
t.Fatalf("round %d: can't remove non-existent peer %v", i, id)
}
dialsched.peerRemoved(c)
}
// Init round.
t.Logf("round %d (%d peers)", i, len(peers))
resolver.setAnswers(round.wantResolves)
if round.update != nil {
round.update(dialsched)
}
iterator.addNodes(round.discovered)
// Unblock dialTask goroutines.
if err := dialer.completeDials(round.succeeded, nil); err != nil {
t.Fatalf("round %d: %v", i, err)
}
for range round.succeeded {
conn := <-setupCh
peers[conn.node.ID()] = conn
}
if err := dialer.completeDials(round.failed, errors.New("oops")); err != nil {
t.Fatalf("round %d: %v", i, err)
}
// Wait for new tasks.
if err := dialer.waitForDials(round.wantNewDials); err != nil {
t.Fatalf("round %d: %v", i, err)
}
if !resolver.checkCalls() {
t.Fatalf("unexpected calls to Resolve: %v", resolver.calls)
}
clock.Run(16 * 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
5 years ago
}
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
5 years ago
// dialTestIterator is the input iterator for dialer tests. This works a bit like a channel
// with infinite buffer: nodes are added to the buffer with addNodes, which unblocks Next
// and returns them from the iterator.
type dialTestIterator struct {
cur *enode.Node
mu sync.Mutex
buf []*enode.Node
cond *sync.Cond
closed bool
}
func newDialTestIterator() *dialTestIterator {
it := &dialTestIterator{}
it.cond = sync.NewCond(&it.mu)
return it
}
// addNodes adds nodes to the iterator buffer and unblocks Next.
func (it *dialTestIterator) addNodes(nodes []*enode.Node) {
it.mu.Lock()
defer it.mu.Unlock()
it.buf = append(it.buf, nodes...)
it.cond.Signal()
}
// Node returns the current node.
func (it *dialTestIterator) Node() *enode.Node {
return it.cur
}
// Next moves to the next node.
func (it *dialTestIterator) Next() bool {
it.mu.Lock()
defer it.mu.Unlock()
it.cur = nil
for len(it.buf) == 0 && !it.closed {
it.cond.Wait()
}
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
5 years ago
if it.closed {
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
5 years ago
it.cur = it.buf[0]
copy(it.buf[:], it.buf[1:])
it.buf = it.buf[:len(it.buf)-1]
return true
}
// Close ends the iterator, unblocking Next.
func (it *dialTestIterator) Close() {
it.mu.Lock()
defer it.mu.Unlock()
it.closed = true
it.buf = nil
it.cond.Signal()
}
// dialTestDialer is the NodeDialer used by runDialTest.
type dialTestDialer struct {
init chan *dialTestReq
blocked map[enode.ID]*dialTestReq
}
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
5 years ago
type dialTestReq struct {
n *enode.Node
unblock chan error
}
func newDialTestDialer() *dialTestDialer {
return &dialTestDialer{
init: make(chan *dialTestReq),
blocked: make(map[enode.ID]*dialTestReq),
}
}
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
5 years ago
// Dial implements NodeDialer.
func (d *dialTestDialer) Dial(ctx context.Context, n *enode.Node) (net.Conn, error) {
req := &dialTestReq{n: n, unblock: make(chan error, 1)}
select {
case d.init <- req:
select {
case err := <-req.unblock:
pipe, _ := net.Pipe()
return pipe, err
case <-ctx.Done():
return nil, ctx.Err()
}
case <-ctx.Done():
return nil, ctx.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
5 years ago
}
// waitForDials waits for calls to Dial with the given nodes as argument.
// Those calls will be held blocking until completeDials is called with the same nodes.
func (d *dialTestDialer) waitForDials(nodes []*enode.Node) error {
waitset := make(map[enode.ID]*enode.Node)
for _, n := range nodes {
waitset[n.ID()] = n
}
timeout := time.NewTimer(1 * time.Second)
defer timeout.Stop()
for len(waitset) > 0 {
select {
case req := <-d.init:
want, ok := waitset[req.n.ID()]
if !ok {
return fmt.Errorf("attempt to dial unexpected node %v", req.n.ID())
}
if !reflect.DeepEqual(req.n, want) {
return fmt.Errorf("ENR of dialed node %v does not match test", req.n.ID())
}
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
5 years ago
delete(waitset, req.n.ID())
d.blocked[req.n.ID()] = req
case <-timeout.C:
var waitlist []enode.ID
for id := range waitset {
waitlist = append(waitlist, id)
}
return fmt.Errorf("timed out waiting for dials to %v", waitlist)
}
}
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
5 years ago
return d.checkUnexpectedDial()
}
func (d *dialTestDialer) checkUnexpectedDial() error {
select {
case req := <-d.init:
return fmt.Errorf("attempt to dial unexpected node %v", req.n.ID())
case <-time.After(150 * time.Millisecond):
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
5 years ago
// completeDials unblocks calls to Dial for the given nodes.
func (d *dialTestDialer) completeDials(ids []enode.ID, err error) error {
for _, id := range ids {
req := d.blocked[id]
if req == nil {
return fmt.Errorf("can't complete dial to %v", id)
}
req.unblock <- err
}
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
5 years ago
// dialTestResolver tracks calls to resolve.
type dialTestResolver struct {
mu sync.Mutex
calls []enode.ID
answers map[enode.ID]*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
5 years ago
func (t *dialTestResolver) setAnswers(m map[enode.ID]*enode.Node) {
t.mu.Lock()
defer t.mu.Unlock()
t.answers = m
t.calls = nil
}
func (t *dialTestResolver) checkCalls() bool {
t.mu.Lock()
defer t.mu.Unlock()
for _, id := range t.calls {
if _, ok := t.answers[id]; !ok {
return false
}
}
return true
}
func (t *dialTestResolver) Resolve(n *enode.Node) *enode.Node {
t.mu.Lock()
defer t.mu.Unlock()
t.calls = append(t.calls, n.ID())
return t.answers[n.ID()]
}