1 /* Copyright (c) 2010-2020. The SimGrid Team. All rights reserved. */
3 /* This program is free software; you can redistribute it and/or modify it
4 * under the terms of the license (GNU LGPL) which comes with this package. */
6 #include "s4u-dht-chord.hpp"
8 XBT_LOG_EXTERNAL_DEFAULT_CATEGORY(s4u_chord);
10 /* Returns whether an id belongs to the interval [start, end].
12 * The parameters are normalized to make sure they are between 0 and nb_keys - 1).
13 * 1 belongs to [62, 3]
14 * 1 does not belong to [3, 62]
15 * 63 belongs to [62, 3]
16 * 63 does not belong to [3, 62]
17 * 24 belongs to [21, 29]
18 * 24 does not belong to [29, 21]
20 * @param id id to check
21 * @param start lower bound
22 * @param end upper bound
23 * @return true if id in in [start, end]
25 static bool is_in_interval(int id, int start, int end)
28 int s = start % nb_keys;
29 int e = end % nb_keys;
31 // make sure end >= start and id >= start
43 void ChordMessage::destroy(void* message)
45 delete static_cast<ChordMessage*>(message);
48 /* Initializes the current node as the first one of the system */
49 Node::Node(std::vector<std::string> args)
51 xbt_assert(args.size() == 3 || args.size() == 5, "Wrong number of arguments for this node");
54 id_ = std::stoi(args[1]);
55 XBT_DEBUG("Initialize node with id: %d", id_);
57 mailbox_ = simgrid::s4u::Mailbox::by_name(std::to_string(id_));
58 next_finger_to_fix = 0;
59 fingers_.resize(nb_bits, id_);
61 if (args.size() == 3) { // first ring
62 deadline_ = std::stod(args[2]);
63 start_time_ = simgrid::s4u::Engine::get_clock();
64 XBT_DEBUG("Create a new Chord ring...");
66 known_id_ = std::stoi(args[2]);
67 start_time_ = std::stod(args[3]);
68 deadline_ = std::stod(args[4]);
69 XBT_DEBUG("Hey! Let's join the system in %f seconds (shall leave at time %f)", start_time_,
70 start_time_ + deadline_);
74 /* Makes the current node join the ring, knowing the id of a node already in the ring
76 * @param known_id id of a node already in the ring
77 * @return true if the join operation succeeded
80 void Node::join(int known_id)
82 XBT_INFO("Joining the ring with id %d, knowing node %d", id_, known_id);
83 setPredecessor(-1); // no predecessor (yet)
85 int successor_id = remoteFindSuccessor(known_id, id_);
86 if (successor_id == -1) {
87 XBT_INFO("Cannot join the ring.");
89 setFinger(0, successor_id);
95 /* Makes the current node quit the system */
98 XBT_INFO("Well Guys! I Think it's time for me to leave ;)");
103 /* Notifies the successor and the predecessor of the current node before leaving */
104 void Node::notifyAndQuit()
106 // send the PREDECESSOR_LEAVING to our successor
107 auto* pred_msg = new ChordMessage(MessageType::PREDECESSOR_LEAVING);
108 pred_msg->request_id = pred_id_;
109 pred_msg->answer_to = mailbox_;
111 XBT_DEBUG("Sending a 'PREDECESSOR_LEAVING' to my successor %d", fingers_[0]);
113 simgrid::s4u::Mailbox::by_name(std::to_string(fingers_[0]))->put(pred_msg, 10, timeout);
114 } catch (const simgrid::TimeoutException&) {
115 XBT_DEBUG("Timeout expired when sending a 'PREDECESSOR_LEAVING' to my successor %d", fingers_[0]);
119 if (pred_id_ != -1 && pred_id_ != id_) {
120 // send the SUCCESSOR_LEAVING to our predecessor (only if I have one that is not me)
121 auto* succ_msg = new ChordMessage(MessageType::SUCCESSOR_LEAVING);
122 succ_msg->request_id = fingers_[0];
123 succ_msg->answer_to = mailbox_;
124 XBT_DEBUG("Sending a 'SUCCESSOR_LEAVING' to my predecessor %d", pred_id_);
127 simgrid::s4u::Mailbox::by_name(std::to_string(pred_id_))->put(succ_msg, 10, timeout);
128 } catch (const simgrid::TimeoutException&) {
129 XBT_DEBUG("Timeout expired when sending a 'SUCCESSOR_LEAVING' to my predecessor %d", pred_id_);
135 /* Performs a find successor request to a random id */
136 void Node::randomLookup()
139 int random_index = random.uniform_int(0, nb_bits - 1);
140 int random_id = fingers_[random_index];
141 XBT_DEBUG("Making a lookup request for id %d", random_id);
142 if (random_id != id_)
143 res = findSuccessor(random_id);
144 XBT_DEBUG("The successor of node %d is %d", random_id, res);
147 /* Sets a finger of the current node.
149 * @param node the current node
150 * @param finger_index index of the finger to set (0 to nb_bits - 1)
151 * @param id the id to set for this finger
153 void Node::setFinger(int finger_index, int id)
155 if (id != fingers_[finger_index]) {
156 fingers_[finger_index] = id;
157 XBT_VERB("My new finger #%d is %d", finger_index, id);
161 /* Sets the predecessor of the current node.
162 * @param id the id to predecessor, or -1 to unset the predecessor
164 void Node::setPredecessor(int predecessor_id)
166 if (predecessor_id != pred_id_) {
167 pred_id_ = predecessor_id;
168 XBT_VERB("My new predecessor is %d", predecessor_id);
172 /** refreshes the finger table of the current node (called periodically) */
173 void Node::fixFingers()
175 XBT_DEBUG("Fixing fingers");
176 int id = findSuccessor(id_ + (1U << next_finger_to_fix));
178 if (id != fingers_[next_finger_to_fix]) {
179 setFinger(next_finger_to_fix, id);
182 next_finger_to_fix = (next_finger_to_fix + 1) % nb_bits;
186 /** Displays the finger table of a node. */
187 void Node::printFingerTable()
189 if (XBT_LOG_ISENABLED(s4u_chord, xbt_log_priority_verbose)) {
190 XBT_VERB("My finger table:");
191 XBT_VERB("Start | Succ");
192 for (int i = 0; i < nb_bits; i++) {
193 XBT_VERB(" %3u | %3d", (id_ + (1U << i)) % nb_keys, fingers_[i]);
196 XBT_VERB("Predecessor: %d", pred_id_);
200 /* checks whether the predecessor has failed (called periodically) */
201 void Node::checkPredecessor()
203 XBT_DEBUG("Checking whether my predecessor is alive");
207 simgrid::s4u::Mailbox* mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(pred_id_));
208 simgrid::s4u::Mailbox* return_mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(id_) + "_is_alive");
210 auto* message = new ChordMessage(MessageType::PREDECESSOR_ALIVE);
211 message->request_id = pred_id_;
212 message->answer_to = return_mailbox;
214 XBT_DEBUG("Sending a 'Predecessor Alive' request to my predecessor %d", pred_id_);
216 mailbox->put(message, 10, timeout);
217 } catch (const simgrid::TimeoutException&) {
218 XBT_DEBUG("Failed to send the 'Predecessor Alive' request to %d", pred_id_);
223 // receive the answer
224 XBT_DEBUG("Sent 'Predecessor Alive' request to %d, waiting for the answer on my mailbox '%s'", pred_id_,
225 message->answer_to->get_cname());
226 ChordMessage* answer = nullptr;
227 simgrid::s4u::CommPtr comm = return_mailbox->get_async<ChordMessage>(&answer);
230 comm->wait_for(timeout);
231 XBT_DEBUG("Received the answer to my 'Predecessor Alive': my predecessor %d is alive", pred_id_);
233 } catch (const simgrid::TimeoutException&) {
234 XBT_DEBUG("Failed to receive the answer to my 'Predecessor Alive' request");
239 /* Asks its predecessor to a remote node
241 * @param ask_to the node to ask to
242 * @return the id of its predecessor node, or -1 if the request failed (or if the node does not know its predecessor)
244 int Node::remoteGetPredecessor(int ask_to)
246 int predecessor_id = -1;
247 simgrid::s4u::Mailbox* mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(ask_to));
248 simgrid::s4u::Mailbox* return_mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(id_) + "_pred");
250 auto* message = new ChordMessage(MessageType::GET_PREDECESSOR);
251 message->request_id = id_;
252 message->answer_to = return_mailbox;
254 // send a "Get Predecessor" request to ask_to_id
255 XBT_DEBUG("Sending a 'Get Predecessor' request to %d", ask_to);
257 mailbox->put(message, 10, timeout);
258 } catch (const simgrid::TimeoutException&) {
259 XBT_DEBUG("Failed to send the 'Get Predecessor' request to %d", ask_to);
261 return predecessor_id;
264 // receive the answer
265 XBT_DEBUG("Sent 'Get Predecessor' request to %d, waiting for the answer on my mailbox '%s'", ask_to,
266 message->answer_to->get_cname());
267 ChordMessage* answer = nullptr;
268 simgrid::s4u::CommPtr comm = return_mailbox->get_async<ChordMessage>(&answer);
271 comm->wait_for(timeout);
272 XBT_DEBUG("Received the answer to my 'Get Predecessor' request: the predecessor of node %d is %d", ask_to,
274 predecessor_id = answer->answer_id;
276 } catch (const simgrid::TimeoutException&) {
277 XBT_DEBUG("Failed to receive the answer to my 'Get Predecessor' request");
281 return predecessor_id;
284 /* Returns the closest preceding finger of an id with respect to the finger table of the current node.
286 * @param id the id to find
287 * @return the closest preceding finger of that id
289 int Node::closestPrecedingFinger(int id)
291 for (int i = nb_bits - 1; i >= 0; i--) {
292 if (is_in_interval(fingers_[i], id_ + 1, id - 1)) {
299 /* Makes the current node find the successor node of an id.
301 * @param id the id to find
302 * @return the id of the successor node, or -1 if the request failed
304 int Node::findSuccessor(int id)
306 // is my successor the successor?
307 if (is_in_interval(id, id_ + 1, fingers_[0])) {
311 // otherwise, ask the closest preceding finger in my table
312 return remoteFindSuccessor(closestPrecedingFinger(id), id);
315 int Node::remoteFindSuccessor(int ask_to, int id)
318 simgrid::s4u::Mailbox* mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(ask_to));
319 simgrid::s4u::Mailbox* return_mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(id_) + "_succ");
321 auto* message = new ChordMessage(MessageType::FIND_SUCCESSOR);
322 message->request_id = id_;
323 message->answer_to = return_mailbox;
325 // send a "Find Successor" request to ask_to_id
326 XBT_DEBUG("Sending a 'Find Successor' request to %d for id %d", ask_to, id);
328 mailbox->put(message, 10, timeout);
329 } catch (const simgrid::TimeoutException&) {
330 XBT_DEBUG("Failed to send the 'Find Successor' request to %d for id %d", ask_to, id_);
334 // receive the answer
335 XBT_DEBUG("Sent a 'Find Successor' request to %d for key %d, waiting for the answer", ask_to, id);
336 ChordMessage* answer = nullptr;
337 simgrid::s4u::CommPtr comm = return_mailbox->get_async<ChordMessage>(&answer);
340 comm->wait_for(timeout);
341 XBT_DEBUG("Received the answer to my 'Find Successor' request for id %d: the successor of key %d is %d",
342 answer->request_id, id_, answer->answer_id);
343 successor = answer->answer_id;
345 } catch (const simgrid::TimeoutException&) {
346 XBT_DEBUG("Failed to receive the answer to my 'Find Successor' request");
353 /* Notifies the current node that its predecessor may have changed. */
354 void Node::notify(int predecessor_candidate_id)
356 if (pred_id_ == -1 || is_in_interval(predecessor_candidate_id, pred_id_ + 1, id_ - 1)) {
357 setPredecessor(predecessor_candidate_id);
360 XBT_DEBUG("I don't have to change my predecessor to %d", predecessor_candidate_id);
364 /* Notifies a remote node that its predecessor may have changed. */
365 void Node::remoteNotify(int notify_id, int predecessor_candidate_id) const
367 auto* message = new ChordMessage(MessageType::NOTIFY);
368 message->request_id = predecessor_candidate_id;
369 message->answer_to = nullptr;
371 // send a "Notify" request to notify_id
372 XBT_DEBUG("Sending a 'Notify' request to %d", notify_id);
373 simgrid::s4u::Mailbox* mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(notify_id));
374 mailbox->put_init(message, 10)->detach(ChordMessage::destroy);
377 /* This function is called periodically. It checks the immediate successor of the current node. */
378 void Node::stabilize()
380 XBT_DEBUG("Stabilizing node");
382 // get the predecessor of my immediate successor
383 int candidate_id = pred_id_;
384 int successor_id = fingers_[0];
385 if (successor_id != id_)
386 candidate_id = remoteGetPredecessor(successor_id);
388 // this node is a candidate to become my new successor
389 if (candidate_id != -1 && is_in_interval(candidate_id, id_ + 1, successor_id - 1)) {
390 setFinger(0, candidate_id);
392 if (successor_id != id_) {
393 remoteNotify(successor_id, id_);
397 /* This function is called when a node receives a message.
399 * @param message the message to handle (don't touch it afterward: it will be destroyed, reused or forwarded)
401 void Node::handleMessage(ChordMessage* message)
403 switch (message->type) {
404 case MessageType::FIND_SUCCESSOR:
405 XBT_DEBUG("Received a 'Find Successor' request from %s for id %d", message->issuer_host_name.c_str(),
406 message->request_id);
407 // is my successor the successor?
408 if (is_in_interval(message->request_id, id_ + 1, fingers_[0])) {
409 message->type = MessageType::FIND_SUCCESSOR_ANSWER;
410 message->answer_id = fingers_[0];
411 XBT_DEBUG("Sending back a 'Find Successor Answer' to %s (mailbox %s): the successor of %d is %d",
412 message->issuer_host_name.c_str(), message->answer_to->get_cname(), message->request_id,
414 message->answer_to->put_init(message, 10)->detach(ChordMessage::destroy);
416 // otherwise, forward the request to the closest preceding finger in my table
417 int closest = closestPrecedingFinger(message->request_id);
418 XBT_DEBUG("Forwarding the 'Find Successor' request for id %d to my closest preceding finger %d",
419 message->request_id, closest);
420 simgrid::s4u::Mailbox* mailbox = simgrid::s4u::Mailbox::by_name(std::to_string(closest));
421 mailbox->put_init(message, 10)->detach(ChordMessage::destroy);
425 case MessageType::GET_PREDECESSOR:
426 XBT_DEBUG("Receiving a 'Get Predecessor' request from %s", message->issuer_host_name.c_str());
427 message->type = MessageType::GET_PREDECESSOR_ANSWER;
428 message->answer_id = pred_id_;
429 XBT_DEBUG("Sending back a 'Get Predecessor Answer' to %s via mailbox '%s': my predecessor is %d",
430 message->issuer_host_name.c_str(), message->answer_to->get_cname(), message->answer_id);
431 message->answer_to->put_init(message, 10)->detach(ChordMessage::destroy);
434 case MessageType::NOTIFY:
435 // someone is telling me that he may be my new predecessor
436 XBT_DEBUG("Receiving a 'Notify' request from %s", message->issuer_host_name.c_str());
437 notify(message->request_id);
441 case MessageType::PREDECESSOR_LEAVING:
442 // my predecessor is about to quit
443 XBT_DEBUG("Receiving a 'Predecessor Leaving' message from %s", message->issuer_host_name.c_str());
444 // modify my predecessor
445 setPredecessor(message->request_id);
448 >> notify my new predecessor
449 >> send a notify_predecessors !!
453 case MessageType::SUCCESSOR_LEAVING:
454 // my successor is about to quit
455 XBT_DEBUG("Receiving a 'Successor Leaving' message from %s", message->issuer_host_name.c_str());
456 // modify my successor FIXME : this should be implicit ?
457 setFinger(0, message->request_id);
460 >> notify my new successor
461 >> update my table & predecessors table */
464 case MessageType::PREDECESSOR_ALIVE:
465 XBT_DEBUG("Receiving a 'Predecessor Alive' request from %s", message->issuer_host_name.c_str());
466 message->type = MessageType::PREDECESSOR_ALIVE_ANSWER;
467 XBT_DEBUG("Sending back a 'Predecessor Alive Answer' to %s (mailbox %s)", message->issuer_host_name.c_str(),
468 message->answer_to->get_cname());
469 message->answer_to->put_init(message, 10)->detach(ChordMessage::destroy);
473 XBT_DEBUG("Ignoring unexpected message: %d from %s", static_cast<int>(message->type),
474 message->issuer_host_name.c_str());
479 void Node::operator()()
481 simgrid::s4u::this_actor::sleep_for(start_time_);
482 if (known_id_ == -1) {
483 setPredecessor(-1); // -1 means that I have no predecessor
492 ChordMessage* message = nullptr;
493 double now = simgrid::s4u::Engine::get_clock();
494 double next_stabilize_date = start_time_ + PERIODIC_STABILIZE_DELAY;
495 double next_fix_fingers_date = start_time_ + PERIODIC_FIX_FINGERS_DELAY;
496 double next_check_predecessor_date = start_time_ + PERIODIC_CHECK_PREDECESSOR_DELAY;
497 double next_lookup_date = start_time_ + PERIODIC_LOOKUP_DELAY;
498 simgrid::s4u::CommPtr comm_receive = nullptr;
499 while (now < std::min(start_time_ + deadline_, MAX_SIMULATION_TIME)) {
500 if (comm_receive == nullptr)
501 comm_receive = mailbox_->get_async<ChordMessage>(&message);
502 bool comm_completed = true;
504 if (not comm_receive->test())
505 comm_completed = false;
506 } catch (const simgrid::TimeoutException&) {
507 XBT_DEBUG("Caught a timeout, go ahead.");
510 if (comm_completed) {
511 if (message != nullptr) {
512 handleMessage(message);
515 comm_receive = nullptr;
517 // no task was received: make some periodic calls
518 if (now >= next_stabilize_date) {
520 next_stabilize_date = simgrid::s4u::Engine::get_clock() + PERIODIC_STABILIZE_DELAY;
521 } else if (now >= next_fix_fingers_date) {
523 next_fix_fingers_date = simgrid::s4u::Engine::get_clock() + PERIODIC_FIX_FINGERS_DELAY;
524 } else if (now >= next_check_predecessor_date) {
526 next_check_predecessor_date = simgrid::s4u::Engine::get_clock() + PERIODIC_CHECK_PREDECESSOR_DELAY;
527 } else if (now >= next_lookup_date) {
529 next_lookup_date = simgrid::s4u::Engine::get_clock() + PERIODIC_LOOKUP_DELAY;
531 // nothing to do: sleep for a while
532 simgrid::s4u::this_actor::sleep_for(SLEEP_DELAY);
536 now = simgrid::s4u::Engine::get_clock();
538 if (comm_receive != nullptr) {
540 if (comm_receive->test())
543 comm_receive->cancel();
544 } catch (const simgrid::TimeoutException&) {
545 XBT_DEBUG("Caught a timeout for last message, nevermind.");