gitee.com/liu-zhao234568/cntest@v1.0.0/p2p/simulations/network.go (about) 1 // Copyright 2017 The go-ethereum Authors 2 // This file is part of the go-ethereum library. 3 // 4 // The go-ethereum library is free software: you can redistribute it and/or modify 5 // it under the terms of the GNU Lesser General Public License as published by 6 // the Free Software Foundation, either version 3 of the License, or 7 // (at your option) any later version. 8 // 9 // The go-ethereum library is distributed in the hope that it will be useful, 10 // but WITHOUT ANY WARRANTY; without even the implied warranty of 11 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 12 // GNU Lesser General Public License for more details. 13 // 14 // You should have received a copy of the GNU Lesser General Public License 15 // along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>. 16 17 package simulations 18 19 import ( 20 "bytes" 21 "context" 22 "encoding/json" 23 "errors" 24 "fmt" 25 "io" 26 "math/rand" 27 "sync" 28 "time" 29 30 "gitee.com/liu-zhao234568/cntest/event" 31 "gitee.com/liu-zhao234568/cntest/log" 32 "gitee.com/liu-zhao234568/cntest/p2p" 33 "gitee.com/liu-zhao234568/cntest/p2p/enode" 34 "gitee.com/liu-zhao234568/cntest/p2p/simulations/adapters" 35 ) 36 37 var DialBanTimeout = 200 * time.Millisecond 38 39 // NetworkConfig defines configuration options for starting a Network 40 type NetworkConfig struct { 41 ID string `json:"id"` 42 DefaultService string `json:"default_service,omitempty"` 43 } 44 45 // Network models a p2p simulation network which consists of a collection of 46 // simulated nodes and the connections which exist between them. 47 // 48 // The Network has a single NodeAdapter which is responsible for actually 49 // starting nodes and connecting them together. 50 // 51 // The Network emits events when nodes are started and stopped, when they are 52 // connected and disconnected, and also when messages are sent between nodes. 53 type Network struct { 54 NetworkConfig 55 56 Nodes []*Node `json:"nodes"` 57 nodeMap map[enode.ID]int 58 59 // Maps a node property string to node indexes of all nodes that hold this property 60 propertyMap map[string][]int 61 62 Conns []*Conn `json:"conns"` 63 connMap map[string]int 64 65 nodeAdapter adapters.NodeAdapter 66 events event.Feed 67 lock sync.RWMutex 68 quitc chan struct{} 69 } 70 71 // NewNetwork returns a Network which uses the given NodeAdapter and NetworkConfig 72 func NewNetwork(nodeAdapter adapters.NodeAdapter, conf *NetworkConfig) *Network { 73 return &Network{ 74 NetworkConfig: *conf, 75 nodeAdapter: nodeAdapter, 76 nodeMap: make(map[enode.ID]int), 77 propertyMap: make(map[string][]int), 78 connMap: make(map[string]int), 79 quitc: make(chan struct{}), 80 } 81 } 82 83 // Events returns the output event feed of the Network. 84 func (net *Network) Events() *event.Feed { 85 return &net.events 86 } 87 88 // NewNodeWithConfig adds a new node to the network with the given config, 89 // returning an error if a node with the same ID or name already exists 90 func (net *Network) NewNodeWithConfig(conf *adapters.NodeConfig) (*Node, error) { 91 net.lock.Lock() 92 defer net.lock.Unlock() 93 94 if conf.Reachable == nil { 95 conf.Reachable = func(otherID enode.ID) bool { 96 _, err := net.InitConn(conf.ID, otherID) 97 if err != nil && bytes.Compare(conf.ID.Bytes(), otherID.Bytes()) < 0 { 98 return false 99 } 100 return true 101 } 102 } 103 104 // check the node doesn't already exist 105 if node := net.getNode(conf.ID); node != nil { 106 return nil, fmt.Errorf("node with ID %q already exists", conf.ID) 107 } 108 if node := net.getNodeByName(conf.Name); node != nil { 109 return nil, fmt.Errorf("node with name %q already exists", conf.Name) 110 } 111 112 // if no services are configured, use the default service 113 if len(conf.Lifecycles) == 0 { 114 conf.Lifecycles = []string{net.DefaultService} 115 } 116 117 // use the NodeAdapter to create the node 118 adapterNode, err := net.nodeAdapter.NewNode(conf) 119 if err != nil { 120 return nil, err 121 } 122 node := newNode(adapterNode, conf, false) 123 log.Trace("Node created", "id", conf.ID) 124 125 nodeIndex := len(net.Nodes) 126 net.nodeMap[conf.ID] = nodeIndex 127 net.Nodes = append(net.Nodes, node) 128 129 // Register any node properties with the network-level propertyMap 130 for _, property := range conf.Properties { 131 net.propertyMap[property] = append(net.propertyMap[property], nodeIndex) 132 } 133 134 // emit a "control" event 135 net.events.Send(ControlEvent(node)) 136 137 return node, nil 138 } 139 140 // Config returns the network configuration 141 func (net *Network) Config() *NetworkConfig { 142 return &net.NetworkConfig 143 } 144 145 // StartAll starts all nodes in the network 146 func (net *Network) StartAll() error { 147 for _, node := range net.Nodes { 148 if node.Up() { 149 continue 150 } 151 if err := net.Start(node.ID()); err != nil { 152 return err 153 } 154 } 155 return nil 156 } 157 158 // StopAll stops all nodes in the network 159 func (net *Network) StopAll() error { 160 for _, node := range net.Nodes { 161 if !node.Up() { 162 continue 163 } 164 if err := net.Stop(node.ID()); err != nil { 165 return err 166 } 167 } 168 return nil 169 } 170 171 // Start starts the node with the given ID 172 func (net *Network) Start(id enode.ID) error { 173 return net.startWithSnapshots(id, nil) 174 } 175 176 // startWithSnapshots starts the node with the given ID using the give 177 // snapshots 178 func (net *Network) startWithSnapshots(id enode.ID, snapshots map[string][]byte) error { 179 net.lock.Lock() 180 defer net.lock.Unlock() 181 182 node := net.getNode(id) 183 if node == nil { 184 return fmt.Errorf("node %v does not exist", id) 185 } 186 if node.Up() { 187 return fmt.Errorf("node %v already up", id) 188 } 189 log.Trace("Starting node", "id", id, "adapter", net.nodeAdapter.Name()) 190 if err := node.Start(snapshots); err != nil { 191 log.Warn("Node startup failed", "id", id, "err", err) 192 return err 193 } 194 node.SetUp(true) 195 log.Info("Started node", "id", id) 196 ev := NewEvent(node) 197 net.events.Send(ev) 198 199 // subscribe to peer events 200 client, err := node.Client() 201 if err != nil { 202 return fmt.Errorf("error getting rpc client for node %v: %s", id, err) 203 } 204 events := make(chan *p2p.PeerEvent) 205 sub, err := client.Subscribe(context.Background(), "admin", events, "peerEvents") 206 if err != nil { 207 return fmt.Errorf("error getting peer events for node %v: %s", id, err) 208 } 209 go net.watchPeerEvents(id, events, sub) 210 return nil 211 } 212 213 // watchPeerEvents reads peer events from the given channel and emits 214 // corresponding network events 215 func (net *Network) watchPeerEvents(id enode.ID, events chan *p2p.PeerEvent, sub event.Subscription) { 216 defer func() { 217 sub.Unsubscribe() 218 219 // assume the node is now down 220 net.lock.Lock() 221 defer net.lock.Unlock() 222 223 node := net.getNode(id) 224 if node == nil { 225 return 226 } 227 node.SetUp(false) 228 ev := NewEvent(node) 229 net.events.Send(ev) 230 }() 231 for { 232 select { 233 case event, ok := <-events: 234 if !ok { 235 return 236 } 237 peer := event.Peer 238 switch event.Type { 239 240 case p2p.PeerEventTypeAdd: 241 net.DidConnect(id, peer) 242 243 case p2p.PeerEventTypeDrop: 244 net.DidDisconnect(id, peer) 245 246 case p2p.PeerEventTypeMsgSend: 247 net.DidSend(id, peer, event.Protocol, *event.MsgCode) 248 249 case p2p.PeerEventTypeMsgRecv: 250 net.DidReceive(peer, id, event.Protocol, *event.MsgCode) 251 252 } 253 254 case err := <-sub.Err(): 255 if err != nil { 256 log.Error("Error in peer event subscription", "id", id, "err", err) 257 } 258 return 259 } 260 } 261 } 262 263 // Stop stops the node with the given ID 264 func (net *Network) Stop(id enode.ID) error { 265 // IMPORTANT: node.Stop() must NOT be called under net.lock as 266 // node.Reachable() closure has a reference to the network and 267 // calls net.InitConn() what also locks the network. => DEADLOCK 268 // That holds until the following ticket is not resolved: 269 270 var err error 271 272 node, err := func() (*Node, error) { 273 net.lock.Lock() 274 defer net.lock.Unlock() 275 276 node := net.getNode(id) 277 if node == nil { 278 return nil, fmt.Errorf("node %v does not exist", id) 279 } 280 if !node.Up() { 281 return nil, fmt.Errorf("node %v already down", id) 282 } 283 node.SetUp(false) 284 return node, nil 285 }() 286 if err != nil { 287 return err 288 } 289 290 err = node.Stop() // must be called without net.lock 291 292 net.lock.Lock() 293 defer net.lock.Unlock() 294 295 if err != nil { 296 node.SetUp(true) 297 return err 298 } 299 log.Info("Stopped node", "id", id, "err", err) 300 ev := ControlEvent(node) 301 net.events.Send(ev) 302 return nil 303 } 304 305 // Connect connects two nodes together by calling the "admin_addPeer" RPC 306 // method on the "one" node so that it connects to the "other" node 307 func (net *Network) Connect(oneID, otherID enode.ID) error { 308 net.lock.Lock() 309 defer net.lock.Unlock() 310 return net.connect(oneID, otherID) 311 } 312 313 func (net *Network) connect(oneID, otherID enode.ID) error { 314 log.Debug("Connecting nodes with addPeer", "id", oneID, "other", otherID) 315 conn, err := net.initConn(oneID, otherID) 316 if err != nil { 317 return err 318 } 319 client, err := conn.one.Client() 320 if err != nil { 321 return err 322 } 323 net.events.Send(ControlEvent(conn)) 324 return client.Call(nil, "admin_addPeer", string(conn.other.Addr())) 325 } 326 327 // Disconnect disconnects two nodes by calling the "admin_removePeer" RPC 328 // method on the "one" node so that it disconnects from the "other" node 329 func (net *Network) Disconnect(oneID, otherID enode.ID) error { 330 conn := net.GetConn(oneID, otherID) 331 if conn == nil { 332 return fmt.Errorf("connection between %v and %v does not exist", oneID, otherID) 333 } 334 if !conn.Up { 335 return fmt.Errorf("%v and %v already disconnected", oneID, otherID) 336 } 337 client, err := conn.one.Client() 338 if err != nil { 339 return err 340 } 341 net.events.Send(ControlEvent(conn)) 342 return client.Call(nil, "admin_removePeer", string(conn.other.Addr())) 343 } 344 345 // DidConnect tracks the fact that the "one" node connected to the "other" node 346 func (net *Network) DidConnect(one, other enode.ID) error { 347 net.lock.Lock() 348 defer net.lock.Unlock() 349 conn, err := net.getOrCreateConn(one, other) 350 if err != nil { 351 return fmt.Errorf("connection between %v and %v does not exist", one, other) 352 } 353 if conn.Up { 354 return fmt.Errorf("%v and %v already connected", one, other) 355 } 356 conn.Up = true 357 net.events.Send(NewEvent(conn)) 358 return nil 359 } 360 361 // DidDisconnect tracks the fact that the "one" node disconnected from the 362 // "other" node 363 func (net *Network) DidDisconnect(one, other enode.ID) error { 364 net.lock.Lock() 365 defer net.lock.Unlock() 366 conn := net.getConn(one, other) 367 if conn == nil { 368 return fmt.Errorf("connection between %v and %v does not exist", one, other) 369 } 370 if !conn.Up { 371 return fmt.Errorf("%v and %v already disconnected", one, other) 372 } 373 conn.Up = false 374 conn.initiated = time.Now().Add(-DialBanTimeout) 375 net.events.Send(NewEvent(conn)) 376 return nil 377 } 378 379 // DidSend tracks the fact that "sender" sent a message to "receiver" 380 func (net *Network) DidSend(sender, receiver enode.ID, proto string, code uint64) error { 381 msg := &Msg{ 382 One: sender, 383 Other: receiver, 384 Protocol: proto, 385 Code: code, 386 Received: false, 387 } 388 net.events.Send(NewEvent(msg)) 389 return nil 390 } 391 392 // DidReceive tracks the fact that "receiver" received a message from "sender" 393 func (net *Network) DidReceive(sender, receiver enode.ID, proto string, code uint64) error { 394 msg := &Msg{ 395 One: sender, 396 Other: receiver, 397 Protocol: proto, 398 Code: code, 399 Received: true, 400 } 401 net.events.Send(NewEvent(msg)) 402 return nil 403 } 404 405 // GetNode gets the node with the given ID, returning nil if the node does not 406 // exist 407 func (net *Network) GetNode(id enode.ID) *Node { 408 net.lock.RLock() 409 defer net.lock.RUnlock() 410 return net.getNode(id) 411 } 412 413 func (net *Network) getNode(id enode.ID) *Node { 414 i, found := net.nodeMap[id] 415 if !found { 416 return nil 417 } 418 return net.Nodes[i] 419 } 420 421 // GetNodeByName gets the node with the given name, returning nil if the node does 422 // not exist 423 func (net *Network) GetNodeByName(name string) *Node { 424 net.lock.RLock() 425 defer net.lock.RUnlock() 426 return net.getNodeByName(name) 427 } 428 429 func (net *Network) getNodeByName(name string) *Node { 430 for _, node := range net.Nodes { 431 if node.Config.Name == name { 432 return node 433 } 434 } 435 return nil 436 } 437 438 // GetNodeIDs returns the IDs of all existing nodes 439 // Nodes can optionally be excluded by specifying their enode.ID. 440 func (net *Network) GetNodeIDs(excludeIDs ...enode.ID) []enode.ID { 441 net.lock.RLock() 442 defer net.lock.RUnlock() 443 444 return net.getNodeIDs(excludeIDs) 445 } 446 447 func (net *Network) getNodeIDs(excludeIDs []enode.ID) []enode.ID { 448 // Get all current nodeIDs 449 nodeIDs := make([]enode.ID, 0, len(net.nodeMap)) 450 for id := range net.nodeMap { 451 nodeIDs = append(nodeIDs, id) 452 } 453 454 if len(excludeIDs) > 0 { 455 // Return the difference of nodeIDs and excludeIDs 456 return filterIDs(nodeIDs, excludeIDs) 457 } 458 return nodeIDs 459 } 460 461 // GetNodes returns the existing nodes. 462 // Nodes can optionally be excluded by specifying their enode.ID. 463 func (net *Network) GetNodes(excludeIDs ...enode.ID) []*Node { 464 net.lock.RLock() 465 defer net.lock.RUnlock() 466 467 return net.getNodes(excludeIDs) 468 } 469 470 func (net *Network) getNodes(excludeIDs []enode.ID) []*Node { 471 if len(excludeIDs) > 0 { 472 nodeIDs := net.getNodeIDs(excludeIDs) 473 return net.getNodesByID(nodeIDs) 474 } 475 return net.Nodes 476 } 477 478 // GetNodesByID returns existing nodes with the given enode.IDs. 479 // If a node doesn't exist with a given enode.ID, it is ignored. 480 func (net *Network) GetNodesByID(nodeIDs []enode.ID) []*Node { 481 net.lock.RLock() 482 defer net.lock.RUnlock() 483 484 return net.getNodesByID(nodeIDs) 485 } 486 487 func (net *Network) getNodesByID(nodeIDs []enode.ID) []*Node { 488 nodes := make([]*Node, 0, len(nodeIDs)) 489 for _, id := range nodeIDs { 490 node := net.getNode(id) 491 if node != nil { 492 nodes = append(nodes, node) 493 } 494 } 495 496 return nodes 497 } 498 499 // GetNodesByProperty returns existing nodes that have the given property string registered in their NodeConfig 500 func (net *Network) GetNodesByProperty(property string) []*Node { 501 net.lock.RLock() 502 defer net.lock.RUnlock() 503 504 return net.getNodesByProperty(property) 505 } 506 507 func (net *Network) getNodesByProperty(property string) []*Node { 508 nodes := make([]*Node, 0, len(net.propertyMap[property])) 509 for _, nodeIndex := range net.propertyMap[property] { 510 nodes = append(nodes, net.Nodes[nodeIndex]) 511 } 512 513 return nodes 514 } 515 516 // GetNodeIDsByProperty returns existing node's enode IDs that have the given property string registered in the NodeConfig 517 func (net *Network) GetNodeIDsByProperty(property string) []enode.ID { 518 net.lock.RLock() 519 defer net.lock.RUnlock() 520 521 return net.getNodeIDsByProperty(property) 522 } 523 524 func (net *Network) getNodeIDsByProperty(property string) []enode.ID { 525 nodeIDs := make([]enode.ID, 0, len(net.propertyMap[property])) 526 for _, nodeIndex := range net.propertyMap[property] { 527 node := net.Nodes[nodeIndex] 528 nodeIDs = append(nodeIDs, node.ID()) 529 } 530 531 return nodeIDs 532 } 533 534 // GetRandomUpNode returns a random node on the network, which is running. 535 func (net *Network) GetRandomUpNode(excludeIDs ...enode.ID) *Node { 536 net.lock.RLock() 537 defer net.lock.RUnlock() 538 return net.getRandomUpNode(excludeIDs...) 539 } 540 541 // GetRandomUpNode returns a random node on the network, which is running. 542 func (net *Network) getRandomUpNode(excludeIDs ...enode.ID) *Node { 543 return net.getRandomNode(net.getUpNodeIDs(), excludeIDs) 544 } 545 546 func (net *Network) getUpNodeIDs() (ids []enode.ID) { 547 for _, node := range net.Nodes { 548 if node.Up() { 549 ids = append(ids, node.ID()) 550 } 551 } 552 return ids 553 } 554 555 // GetRandomDownNode returns a random node on the network, which is stopped. 556 func (net *Network) GetRandomDownNode(excludeIDs ...enode.ID) *Node { 557 net.lock.RLock() 558 defer net.lock.RUnlock() 559 return net.getRandomNode(net.getDownNodeIDs(), excludeIDs) 560 } 561 562 func (net *Network) getDownNodeIDs() (ids []enode.ID) { 563 for _, node := range net.Nodes { 564 if !node.Up() { 565 ids = append(ids, node.ID()) 566 } 567 } 568 return ids 569 } 570 571 // GetRandomNode returns a random node on the network, regardless of whether it is running or not 572 func (net *Network) GetRandomNode(excludeIDs ...enode.ID) *Node { 573 net.lock.RLock() 574 defer net.lock.RUnlock() 575 return net.getRandomNode(net.getNodeIDs(nil), excludeIDs) // no need to exclude twice 576 } 577 578 func (net *Network) getRandomNode(ids []enode.ID, excludeIDs []enode.ID) *Node { 579 filtered := filterIDs(ids, excludeIDs) 580 581 l := len(filtered) 582 if l == 0 { 583 return nil 584 } 585 return net.getNode(filtered[rand.Intn(l)]) 586 } 587 588 func filterIDs(ids []enode.ID, excludeIDs []enode.ID) []enode.ID { 589 exclude := make(map[enode.ID]bool) 590 for _, id := range excludeIDs { 591 exclude[id] = true 592 } 593 var filtered []enode.ID 594 for _, id := range ids { 595 if _, found := exclude[id]; !found { 596 filtered = append(filtered, id) 597 } 598 } 599 return filtered 600 } 601 602 // GetConn returns the connection which exists between "one" and "other" 603 // regardless of which node initiated the connection 604 func (net *Network) GetConn(oneID, otherID enode.ID) *Conn { 605 net.lock.RLock() 606 defer net.lock.RUnlock() 607 return net.getConn(oneID, otherID) 608 } 609 610 // GetOrCreateConn is like GetConn but creates the connection if it doesn't 611 // already exist 612 func (net *Network) GetOrCreateConn(oneID, otherID enode.ID) (*Conn, error) { 613 net.lock.Lock() 614 defer net.lock.Unlock() 615 return net.getOrCreateConn(oneID, otherID) 616 } 617 618 func (net *Network) getOrCreateConn(oneID, otherID enode.ID) (*Conn, error) { 619 if conn := net.getConn(oneID, otherID); conn != nil { 620 return conn, nil 621 } 622 623 one := net.getNode(oneID) 624 if one == nil { 625 return nil, fmt.Errorf("node %v does not exist", oneID) 626 } 627 other := net.getNode(otherID) 628 if other == nil { 629 return nil, fmt.Errorf("node %v does not exist", otherID) 630 } 631 conn := &Conn{ 632 One: oneID, 633 Other: otherID, 634 one: one, 635 other: other, 636 } 637 label := ConnLabel(oneID, otherID) 638 net.connMap[label] = len(net.Conns) 639 net.Conns = append(net.Conns, conn) 640 return conn, nil 641 } 642 643 func (net *Network) getConn(oneID, otherID enode.ID) *Conn { 644 label := ConnLabel(oneID, otherID) 645 i, found := net.connMap[label] 646 if !found { 647 return nil 648 } 649 return net.Conns[i] 650 } 651 652 // InitConn(one, other) retrieves the connection model for the connection between 653 // peers one and other, or creates a new one if it does not exist 654 // the order of nodes does not matter, i.e., Conn(i,j) == Conn(j, i) 655 // it checks if the connection is already up, and if the nodes are running 656 // NOTE: 657 // it also checks whether there has been recent attempt to connect the peers 658 // this is cheating as the simulation is used as an oracle and know about 659 // remote peers attempt to connect to a node which will then not initiate the connection 660 func (net *Network) InitConn(oneID, otherID enode.ID) (*Conn, error) { 661 net.lock.Lock() 662 defer net.lock.Unlock() 663 return net.initConn(oneID, otherID) 664 } 665 666 func (net *Network) initConn(oneID, otherID enode.ID) (*Conn, error) { 667 if oneID == otherID { 668 return nil, fmt.Errorf("refusing to connect to self %v", oneID) 669 } 670 conn, err := net.getOrCreateConn(oneID, otherID) 671 if err != nil { 672 return nil, err 673 } 674 if conn.Up { 675 return nil, fmt.Errorf("%v and %v already connected", oneID, otherID) 676 } 677 if time.Since(conn.initiated) < DialBanTimeout { 678 return nil, fmt.Errorf("connection between %v and %v recently attempted", oneID, otherID) 679 } 680 681 err = conn.nodesUp() 682 if err != nil { 683 log.Trace("Nodes not up", "err", err) 684 return nil, fmt.Errorf("nodes not up: %v", err) 685 } 686 log.Debug("Connection initiated", "id", oneID, "other", otherID) 687 conn.initiated = time.Now() 688 return conn, nil 689 } 690 691 // Shutdown stops all nodes in the network and closes the quit channel 692 func (net *Network) Shutdown() { 693 for _, node := range net.Nodes { 694 log.Debug("Stopping node", "id", node.ID()) 695 if err := node.Stop(); err != nil { 696 log.Warn("Can't stop node", "id", node.ID(), "err", err) 697 } 698 // If the node has the close method, call it. 699 if closer, ok := node.Node.(io.Closer); ok { 700 if err := closer.Close(); err != nil { 701 log.Warn("Can't close node", "id", node.ID(), "err", err) 702 } 703 } 704 } 705 close(net.quitc) 706 } 707 708 // Reset resets all network properties: 709 // empties the nodes and the connection list 710 func (net *Network) Reset() { 711 net.lock.Lock() 712 defer net.lock.Unlock() 713 714 //re-initialize the maps 715 net.connMap = make(map[string]int) 716 net.nodeMap = make(map[enode.ID]int) 717 net.propertyMap = make(map[string][]int) 718 719 net.Nodes = nil 720 net.Conns = nil 721 } 722 723 // Node is a wrapper around adapters.Node which is used to track the status 724 // of a node in the network 725 type Node struct { 726 adapters.Node `json:"-"` 727 728 // Config if the config used to created the node 729 Config *adapters.NodeConfig `json:"config"` 730 731 // up tracks whether or not the node is running 732 up bool 733 upMu *sync.RWMutex 734 } 735 736 func newNode(an adapters.Node, ac *adapters.NodeConfig, up bool) *Node { 737 return &Node{Node: an, Config: ac, up: up, upMu: new(sync.RWMutex)} 738 } 739 740 func (n *Node) copy() *Node { 741 configCpy := *n.Config 742 return newNode(n.Node, &configCpy, n.Up()) 743 } 744 745 // Up returns whether the node is currently up (online) 746 func (n *Node) Up() bool { 747 n.upMu.RLock() 748 defer n.upMu.RUnlock() 749 return n.up 750 } 751 752 // SetUp sets the up (online) status of the nodes with the given value 753 func (n *Node) SetUp(up bool) { 754 n.upMu.Lock() 755 defer n.upMu.Unlock() 756 n.up = up 757 } 758 759 // ID returns the ID of the node 760 func (n *Node) ID() enode.ID { 761 return n.Config.ID 762 } 763 764 // String returns a log-friendly string 765 func (n *Node) String() string { 766 return fmt.Sprintf("Node %v", n.ID().TerminalString()) 767 } 768 769 // NodeInfo returns information about the node 770 func (n *Node) NodeInfo() *p2p.NodeInfo { 771 // avoid a panic if the node is not started yet 772 if n.Node == nil { 773 return nil 774 } 775 info := n.Node.NodeInfo() 776 info.Name = n.Config.Name 777 return info 778 } 779 780 // MarshalJSON implements the json.Marshaler interface so that the encoded 781 // JSON includes the NodeInfo 782 func (n *Node) MarshalJSON() ([]byte, error) { 783 return json.Marshal(struct { 784 Info *p2p.NodeInfo `json:"info,omitempty"` 785 Config *adapters.NodeConfig `json:"config,omitempty"` 786 Up bool `json:"up"` 787 }{ 788 Info: n.NodeInfo(), 789 Config: n.Config, 790 Up: n.Up(), 791 }) 792 } 793 794 // UnmarshalJSON implements json.Unmarshaler interface so that we don't lose Node.up 795 // status. IMPORTANT: The implementation is incomplete; we lose p2p.NodeInfo. 796 func (n *Node) UnmarshalJSON(raw []byte) error { 797 // TODO: How should we turn back NodeInfo into n.Node? 798 // Ticket: https://github.com/ethersphere/go-ethereum/issues/1177 799 var node struct { 800 Config *adapters.NodeConfig `json:"config,omitempty"` 801 Up bool `json:"up"` 802 } 803 if err := json.Unmarshal(raw, &node); err != nil { 804 return err 805 } 806 *n = *newNode(nil, node.Config, node.Up) 807 return nil 808 } 809 810 // Conn represents a connection between two nodes in the network 811 type Conn struct { 812 // One is the node which initiated the connection 813 One enode.ID `json:"one"` 814 815 // Other is the node which the connection was made to 816 Other enode.ID `json:"other"` 817 818 // Up tracks whether or not the connection is active 819 Up bool `json:"up"` 820 // Registers when the connection was grabbed to dial 821 initiated time.Time 822 823 one *Node 824 other *Node 825 } 826 827 // nodesUp returns whether both nodes are currently up 828 func (c *Conn) nodesUp() error { 829 if !c.one.Up() { 830 return fmt.Errorf("one %v is not up", c.One) 831 } 832 if !c.other.Up() { 833 return fmt.Errorf("other %v is not up", c.Other) 834 } 835 return nil 836 } 837 838 // String returns a log-friendly string 839 func (c *Conn) String() string { 840 return fmt.Sprintf("Conn %v->%v", c.One.TerminalString(), c.Other.TerminalString()) 841 } 842 843 // Msg represents a p2p message sent between two nodes in the network 844 type Msg struct { 845 One enode.ID `json:"one"` 846 Other enode.ID `json:"other"` 847 Protocol string `json:"protocol"` 848 Code uint64 `json:"code"` 849 Received bool `json:"received"` 850 } 851 852 // String returns a log-friendly string 853 func (m *Msg) String() string { 854 return fmt.Sprintf("Msg(%d) %v->%v", m.Code, m.One.TerminalString(), m.Other.TerminalString()) 855 } 856 857 // ConnLabel generates a deterministic string which represents a connection 858 // between two nodes, used to compare if two connections are between the same 859 // nodes 860 func ConnLabel(source, target enode.ID) string { 861 var first, second enode.ID 862 if bytes.Compare(source.Bytes(), target.Bytes()) > 0 { 863 first = target 864 second = source 865 } else { 866 first = source 867 second = target 868 } 869 return fmt.Sprintf("%v-%v", first, second) 870 } 871 872 // Snapshot represents the state of a network at a single point in time and can 873 // be used to restore the state of a network 874 type Snapshot struct { 875 Nodes []NodeSnapshot `json:"nodes,omitempty"` 876 Conns []Conn `json:"conns,omitempty"` 877 } 878 879 // NodeSnapshot represents the state of a node in the network 880 type NodeSnapshot struct { 881 Node Node `json:"node,omitempty"` 882 883 // Snapshots is arbitrary data gathered from calling node.Snapshots() 884 Snapshots map[string][]byte `json:"snapshots,omitempty"` 885 } 886 887 // Snapshot creates a network snapshot 888 func (net *Network) Snapshot() (*Snapshot, error) { 889 return net.snapshot(nil, nil) 890 } 891 892 func (net *Network) SnapshotWithServices(addServices []string, removeServices []string) (*Snapshot, error) { 893 return net.snapshot(addServices, removeServices) 894 } 895 896 func (net *Network) snapshot(addServices []string, removeServices []string) (*Snapshot, error) { 897 net.lock.Lock() 898 defer net.lock.Unlock() 899 snap := &Snapshot{ 900 Nodes: make([]NodeSnapshot, len(net.Nodes)), 901 } 902 for i, node := range net.Nodes { 903 snap.Nodes[i] = NodeSnapshot{Node: *node.copy()} 904 if !node.Up() { 905 continue 906 } 907 snapshots, err := node.Snapshots() 908 if err != nil { 909 return nil, err 910 } 911 snap.Nodes[i].Snapshots = snapshots 912 for _, addSvc := range addServices { 913 haveSvc := false 914 for _, svc := range snap.Nodes[i].Node.Config.Lifecycles { 915 if svc == addSvc { 916 haveSvc = true 917 break 918 } 919 } 920 if !haveSvc { 921 snap.Nodes[i].Node.Config.Lifecycles = append(snap.Nodes[i].Node.Config.Lifecycles, addSvc) 922 } 923 } 924 if len(removeServices) > 0 { 925 var cleanedServices []string 926 for _, svc := range snap.Nodes[i].Node.Config.Lifecycles { 927 haveSvc := false 928 for _, rmSvc := range removeServices { 929 if rmSvc == svc { 930 haveSvc = true 931 break 932 } 933 } 934 if !haveSvc { 935 cleanedServices = append(cleanedServices, svc) 936 } 937 938 } 939 snap.Nodes[i].Node.Config.Lifecycles = cleanedServices 940 } 941 } 942 for _, conn := range net.Conns { 943 if conn.Up { 944 snap.Conns = append(snap.Conns, *conn) 945 } 946 } 947 return snap, nil 948 } 949 950 // longrunning tests may need a longer timeout 951 var snapshotLoadTimeout = 900 * time.Second 952 953 // Load loads a network snapshot 954 func (net *Network) Load(snap *Snapshot) error { 955 // Start nodes. 956 for _, n := range snap.Nodes { 957 if _, err := net.NewNodeWithConfig(n.Node.Config); err != nil { 958 return err 959 } 960 if !n.Node.Up() { 961 continue 962 } 963 if err := net.startWithSnapshots(n.Node.Config.ID, n.Snapshots); err != nil { 964 return err 965 } 966 } 967 968 // Prepare connection events counter. 969 allConnected := make(chan struct{}) // closed when all connections are established 970 done := make(chan struct{}) // ensures that the event loop goroutine is terminated 971 defer close(done) 972 973 // Subscribe to event channel. 974 // It needs to be done outside of the event loop goroutine (created below) 975 // to ensure that the event channel is blocking before connect calls are made. 976 events := make(chan *Event) 977 sub := net.Events().Subscribe(events) 978 defer sub.Unsubscribe() 979 980 go func() { 981 // Expected number of connections. 982 total := len(snap.Conns) 983 // Set of all established connections from the snapshot, not other connections. 984 // Key array element 0 is the connection One field value, and element 1 connection Other field. 985 connections := make(map[[2]enode.ID]struct{}, total) 986 987 for { 988 select { 989 case e := <-events: 990 // Ignore control events as they do not represent 991 // connect or disconnect (Up) state change. 992 if e.Control { 993 continue 994 } 995 // Detect only connection events. 996 if e.Type != EventTypeConn { 997 continue 998 } 999 connection := [2]enode.ID{e.Conn.One, e.Conn.Other} 1000 // Nodes are still not connected or have been disconnected. 1001 if !e.Conn.Up { 1002 // Delete the connection from the set of established connections. 1003 // This will prevent false positive in case disconnections happen. 1004 delete(connections, connection) 1005 log.Warn("load snapshot: unexpected disconnection", "one", e.Conn.One, "other", e.Conn.Other) 1006 continue 1007 } 1008 // Check that the connection is from the snapshot. 1009 for _, conn := range snap.Conns { 1010 if conn.One == e.Conn.One && conn.Other == e.Conn.Other { 1011 // Add the connection to the set of established connections. 1012 connections[connection] = struct{}{} 1013 if len(connections) == total { 1014 // Signal that all nodes are connected. 1015 close(allConnected) 1016 return 1017 } 1018 1019 break 1020 } 1021 } 1022 case <-done: 1023 // Load function returned, terminate this goroutine. 1024 return 1025 } 1026 } 1027 }() 1028 1029 // Start connecting. 1030 for _, conn := range snap.Conns { 1031 1032 if !net.GetNode(conn.One).Up() || !net.GetNode(conn.Other).Up() { 1033 //in this case, at least one of the nodes of a connection is not up, 1034 //so it would result in the snapshot `Load` to fail 1035 continue 1036 } 1037 if err := net.Connect(conn.One, conn.Other); err != nil { 1038 return err 1039 } 1040 } 1041 1042 select { 1043 // Wait until all connections from the snapshot are established. 1044 case <-allConnected: 1045 // Make sure that we do not wait forever. 1046 case <-time.After(snapshotLoadTimeout): 1047 return errors.New("snapshot connections not established") 1048 } 1049 return nil 1050 } 1051 1052 // Subscribe reads control events from a channel and executes them 1053 func (net *Network) Subscribe(events chan *Event) { 1054 for { 1055 select { 1056 case event, ok := <-events: 1057 if !ok { 1058 return 1059 } 1060 if event.Control { 1061 net.executeControlEvent(event) 1062 } 1063 case <-net.quitc: 1064 return 1065 } 1066 } 1067 } 1068 1069 func (net *Network) executeControlEvent(event *Event) { 1070 log.Trace("Executing control event", "type", event.Type, "event", event) 1071 switch event.Type { 1072 case EventTypeNode: 1073 if err := net.executeNodeEvent(event); err != nil { 1074 log.Error("Error executing node event", "event", event, "err", err) 1075 } 1076 case EventTypeConn: 1077 if err := net.executeConnEvent(event); err != nil { 1078 log.Error("Error executing conn event", "event", event, "err", err) 1079 } 1080 case EventTypeMsg: 1081 log.Warn("Ignoring control msg event") 1082 } 1083 } 1084 1085 func (net *Network) executeNodeEvent(e *Event) error { 1086 if !e.Node.Up() { 1087 return net.Stop(e.Node.ID()) 1088 } 1089 1090 if _, err := net.NewNodeWithConfig(e.Node.Config); err != nil { 1091 return err 1092 } 1093 return net.Start(e.Node.ID()) 1094 } 1095 1096 func (net *Network) executeConnEvent(e *Event) error { 1097 if e.Conn.Up { 1098 return net.Connect(e.Conn.One, e.Conn.Other) 1099 } 1100 return net.Disconnect(e.Conn.One, e.Conn.Other) 1101 }