forked from zeromq/gyre
-
Notifications
You must be signed in to change notification settings - Fork 0
/
node.go
910 lines (764 loc) · 21.9 KB
/
node.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
package gyre
import (
"bytes"
crand "crypto/rand"
"encoding/binary"
"errors"
"fmt"
"io"
"log"
"math/rand"
"net"
"net/url"
"os"
"strconv"
"strings"
"sync"
"time"
"github.com/armen/goviral/zgossip"
zmq "github.com/pebbe/zmq4"
"github.com/zeromq/gyre/beacon"
"github.com/zeromq/gyre/zre/msg"
)
type node struct {
reactor *zmq.Reactor
terminated chan interface{} // API shut us down
wg sync.WaitGroup // wait group is used to wait until actor() is done
events chan *Event // We send all Gyre events to the events channel
cmds chan interface{} // Receive commands from the cmds channel
replies chan interface{} // Send command replies to the replies channel
verbose bool // Log all traffic
beaconPort int // Beacon port number
interval time.Duration // Beacon interval
beacon *beacon.Beacon // Beacon object
uuid []byte // Our UUID
inbox *zmq.Socket // Our inbox socket (ROUTER)
name string // Our public name
endpoint string // Our public endpoint
port uint16 // Our inbox port number
bound bool // Did app bind node explicitly?
status byte // Our own change counter
peers map[string]*peer // Hash of known peers, fast lookup
peerGroups map[string]*group // Groups that our peers are in
ownGroups map[string]*group // Groups that we are in
headers map[string]string // Our header values
gossip *zgossip.Zgossip // Gossip discovery service, if any
gossipBind string // Gossip bind endpoint, if any
gossipConnect string // Gossip connect endpoint, if any
}
// Beacon frame has this format:
//
// Z R E 3 bytes
// Version 1 byte, %x01
// UUID 16 bytes
// Port 2 bytes in network order
type aBeacon struct {
Protocol [3]byte
Version byte
UUID []byte
Port uint16
}
const (
beaconVersion = 0x1
// IANA-assigned port for ZRE discovery protocol
zreDiscoveryPort = 5670
// Port range 0xc000~0xffff is defined by IANA for dynamic or private ports
// We use this when choosing a port for dynamic binding
dynPortFrom uint16 = 0xc000
dynPortTo uint16 = 0xffff
)
// newNode creates a new node.
func newNode(events chan *Event, cmds chan interface{}, replies chan interface{}) (n *node, err error) {
n = &node{
reactor: zmq.NewReactor(),
events: events,
cmds: cmds,
replies: replies,
beaconPort: zreDiscoveryPort,
peers: make(map[string]*peer),
peerGroups: make(map[string]*group),
ownGroups: make(map[string]*group),
headers: make(map[string]string),
terminated: make(chan interface{}),
}
n.beacon = beacon.New()
n.inbox, err = zmq.NewSocket(zmq.ROUTER)
if err != nil {
return nil, err // Could not create new socket
}
err = n.inbox.SetIpv6(true)
if err != nil {
return nil, err
}
// Generate random uuid
n.uuid = make([]byte, 16)
io.ReadFull(crand.Reader, n.uuid)
// Default name for node is first 6 characters of UUID:
// the shorter string is more readable in logs
n.name = fmt.Sprintf("%.6s", fmt.Sprintf("%X", n.uuid))
n.wg.Add(1) // We're going to wait until actor() is done
return
}
// If we haven't already set-up the gossip network, do so
func (n *node) gossipStart() (err error) {
n.beaconPort = 0 // Disable UDP beaconing
if n.gossip == nil {
n.gossip, err = zgossip.New(n.identity())
if err != nil {
return
}
if n.verbose {
n.gossip.SendCmd("VERBOSE", nil, 100*time.Millisecond)
}
}
return err
}
// Start node, return nil if OK, error if not possible
func (n *node) start() (err error) {
// If application didn't bind explicitly, we grab an ephemeral port
// on all available network interfaces. This is orthogonal to
// beaconing, since we can connect to other peers and they will
// gossip our endpoint to others.
if !n.bound {
_, n.port, err = bind(n.inbox, "tcp://*:*")
if err != nil {
return err
}
n.bound = true
}
// Start UDP beaconing, if the application didn't disable it
if n.beaconPort > 0 {
b := &aBeacon{}
b.Protocol[0] = 'Z'
b.Protocol[1] = 'R'
b.Protocol[2] = 'E'
b.Version = beaconVersion
b.UUID = n.uuid
b.Port = n.port
buffer := new(bytes.Buffer)
binary.Write(buffer, binary.BigEndian, b.Protocol)
binary.Write(buffer, binary.BigEndian, b.Version)
binary.Write(buffer, binary.BigEndian, b.UUID)
binary.Write(buffer, binary.BigEndian, b.Port)
if n.interval > 0 {
n.beacon.SetInterval(n.interval)
}
n.beacon.SetPort(n.beaconPort)
n.beacon.NoEcho()
n.beacon.Subscribe([]byte("ZRE"))
err := n.beacon.Publish(buffer.Bytes())
if err != nil {
return err
}
// Our own host endpoint is provided by the beacon
if n.endpoint != "" {
panic("Endpoint is already set")
}
ip := net.ParseIP(n.beacon.Addr())
if ip.To4() == nil {
n.endpoint = fmt.Sprintf("tcp://[%s]:%d", ip.String(), n.port)
} else {
n.endpoint = fmt.Sprintf("tcp://%s:%d", ip.String(), n.port)
}
n.reactor.AddChannel(n.beacon.Signals(), 1, func(s interface{}) error {
n.recvFromBeacon(s.(*beacon.Signal))
return nil
})
} else {
if n.endpoint == "" {
hostname, err := os.Hostname()
if err != nil {
return err
}
n.endpoint = fmt.Sprintf("tcp://%s:%d", hostname, n.port)
}
if n.gossip == nil {
return errors.New("Gossip engine hasn't been started yet, use SetEndpoint, GossipBind or GossipConnect")
}
n.gossip.SendCmd("PUBLISH", map[string]string{n.identity(): n.endpoint}, 100*time.Millisecond)
// Start polling on zgossip
n.reactor.AddChannel(n.gossip.Resp(), 1, func(r interface{}) error {
n.recvFromGossip(r)
return nil
})
}
return
}
// Stop node discovery and interconnection
func (n *node) stop() {
if n.beacon != nil {
// Stop broadcast/listen beacon
b := &aBeacon{}
b.Protocol[0] = 'Z'
b.Protocol[1] = 'R'
b.Protocol[2] = 'E'
b.Version = beaconVersion
b.UUID = n.uuid
b.Port = 0 // Zero means we're stopping
buffer := new(bytes.Buffer)
binary.Write(buffer, binary.BigEndian, b.Protocol)
binary.Write(buffer, binary.BigEndian, b.Version)
binary.Write(buffer, binary.BigEndian, b.UUID)
binary.Write(buffer, binary.BigEndian, b.Port)
n.beacon.Publish(buffer.Bytes())
time.Sleep(1 * time.Millisecond) // Allow 1 msec for beacon to go out
n.beacon.Close()
}
}
// recvFromAPI handles a new command received from front-end
func (n *node) recvFromAPI(c *cmd) {
if n.verbose {
log.Printf("[%s] Received a %q command from API", n.name, c.cmd)
}
switch c.cmd {
case cmdUUID:
n.replies <- &reply{cmd: cmdUUID, payload: n.identity()}
case cmdName:
n.replies <- &reply{cmd: cmdName, payload: n.name}
case cmdSetName:
n.name = c.payload.(string)
case cmdSetHeader:
n.headers[c.key] = c.payload.(string)
case cmdSetVerbose:
n.verbose = c.payload.(bool)
// n.reactor.SetVerbose(n.verbose)
case cmdSetPort:
n.beaconPort = c.payload.(int)
case cmdSetInterval:
// Set beacon interval
n.interval = c.payload.(time.Duration)
case cmdSetIface:
n.beacon.SetInterface(c.payload.(string))
case cmdSetEndpoint:
err := n.gossipStart()
if err != nil {
// Signal the caller and send back the error if any
n.replies <- &reply{cmd: cmdSetEndpoint, err: err}
break
}
endpoint := c.payload.(string)
n.endpoint, _, err = bind(n.inbox, endpoint)
if err != nil {
n.replies <- &reply{cmd: cmdSetEndpoint, err: err}
break
}
n.bound = true
n.beaconPort = 0
n.replies <- &reply{cmd: cmdSetEndpoint}
case cmdGossipBind:
err := n.gossipStart()
if err != nil {
n.replies <- &reply{cmd: cmdGossipBind, err: err}
break
}
endpoint := c.payload.(string)
err = n.gossip.SendCmd("BIND", endpoint, 5*time.Second)
n.replies <- &reply{cmd: cmdGossipBind, err: err}
case cmdGossipPort:
err := n.gossip.SendCmd("PORT", nil, 5*time.Second)
if err != nil {
n.replies <- &reply{cmd: cmdGossipPort, err: err}
break
}
port, err := n.gossip.RecvResp(5 * time.Second)
if err != nil {
n.replies <- &reply{cmd: cmdGossipPort, err: err}
break
}
n.replies <- &reply{cmd: cmdGossipPort, payload: strconv.FormatUint(uint64(port.(uint16)), 10)}
case cmdGossipConnect:
err := n.gossipStart()
if err != nil {
n.replies <- &reply{cmd: cmdGossipConnect, err: err}
break
}
endpoint := c.payload.(string)
err = n.gossip.SendCmd("CONNECT", endpoint, 5*time.Second)
n.replies <- &reply{cmd: cmdGossipConnect, err: err}
case cmdStart:
// Add the ping ticker just right before start so that it reads the latest
// value of loopInterval
n.reactor.AddChannelTime(time.Tick(loopInterval), 1, func(interface{}) error {
n.ping()
return nil
})
err := n.start()
// Signal the caller and send back the error if any
n.replies <- &reply{cmd: cmdStart, err: err}
case cmdStop, cmdTerm:
if n.terminated != nil {
close(n.terminated)
}
// Wait and send the signal in a separate go routine
// because closing terminated channel
go func() {
n.wg.Wait()
// Signal the caller
n.replies <- &reply{}
}()
case cmdWhisper:
// Get peer to send message to
peer, ok := n.peers[c.key]
// Send frame on out to peer's mailbox, drop message
// if peer doesn't exist (may have been destroyed)
if ok {
m := msg.NewWhisper()
m.Content = c.payload.([]byte)
peer.send(m)
}
case cmdShout:
group := c.key
// Get group to send message to
if g, ok := n.peerGroups[group]; ok {
m := msg.NewShout()
m.Group = group
m.Content = c.payload.([]byte)
g.send(m)
}
case cmdJoin:
group := c.key
if _, ok := n.ownGroups[group]; !ok {
// Only send if we're not already in group
n.ownGroups[group] = newGroup(group)
m := msg.NewJoin()
m.Group = group
// Update status before sending command
n.status++
m.Status = n.status
for _, peer := range n.peers {
cloned := msg.Clone(m)
peer.send(cloned)
}
}
case cmdLeave:
group := c.key
if _, ok := n.ownGroups[group]; ok {
// Only send if we are actually in group
m := msg.NewLeave()
m.Group = group
// Update status before sending command
n.status++
m.Status = n.status
for _, peer := range n.peers {
cloned := msg.Clone(m)
peer.send(cloned)
}
delete(n.ownGroups, group)
}
case cmdDump:
// TODO: implement DUMP
case cmdAddr:
if n.beaconPort > 0 {
n.replies <- &reply{cmd: cmdAddr, payload: n.beacon.Addr()}
} else {
u, err := url.Parse(n.endpoint)
if err != nil {
n.replies <- &reply{cmd: cmdHeader, err: err}
return
}
ip, _, err := net.SplitHostPort(u.Host)
if err != nil {
n.replies <- &reply{cmd: cmdHeader, err: err}
return
}
n.replies <- &reply{cmd: cmdAddr, payload: ip}
}
case cmdHeader:
header, ok := n.headers[c.key]
var err error
if !ok {
err = errors.New("Header doesn't exist")
}
n.replies <- &reply{cmd: cmdHeader, err: err, payload: header}
case cmdHeaders:
n.replies <- &reply{cmd: cmdHeader, payload: n.headers}
default:
log.Printf("Invalid command %q %#v", c.cmd, c)
}
}
func (n *node) identity() string {
return fmt.Sprintf("%X", n.uuid)
}
// requirePeer finds or creates peer via its UUID string
func (n *node) requirePeer(identity string, endpoint string) (peer *peer, err error) {
peer, ok := n.peers[identity]
if !ok {
// Purge any previous peer on same endpoint
for _, p := range n.peers {
if p.endpoint == endpoint {
p.disconnect()
}
}
peer = newPeer(identity)
err = peer.connect(n.uuid, endpoint)
if err != nil {
return nil, err
}
// Handshake discovery by sending HELLO as first message
m := msg.NewHello()
m.Endpoint = n.endpoint
m.Status = n.status
m.Name = n.name
for key := range n.ownGroups {
m.Groups = append(m.Groups, key)
}
for key, header := range n.headers {
m.Headers[key] = header
}
peer.send(m)
n.peers[identity] = peer
// TODO(armen): Send new peer event to logger, if any
}
return peer, nil
}
// Remove a peer from our data structures.
func (n *node) removePeer(peer *peer) {
if peer == nil {
return
}
// Tell the calling application the peer has gone
select {
case n.events <- &Event{eventType: EventExit, sender: peer.identity, name: peer.name, time: time.Now()}:
default:
if n.verbose {
log.Printf("[%s] Dropping event: %s", n.name, EventExit)
}
}
// TODO(armen): Send a log event
// Remove peer from any groups we've got it in
for _, group := range n.peerGroups {
group.leave(peer)
}
// It's really important to disconnect from the peer before
// deleting it, unless we'd end up difficulties to reconnect
// to the same endpoint
peer.disconnect()
delete(n.peers, peer.identity)
}
// requirePeerGroup finds or creates group via its name
func (n *node) requirePeerGroup(name string) *group {
group, ok := n.peerGroups[name]
if !ok {
group = newGroup(name)
n.peerGroups[name] = group
}
return group
}
// joinPeerGroup joins the peer to a group
func (n *node) joinPeerGroup(peer *peer, name string) *group {
group := n.requirePeerGroup(name)
group.join(peer)
// Now tell the caller about the peer joined group
select {
case n.events <- &Event{eventType: EventJoin, sender: peer.identity, name: peer.name, group: name, time: time.Now()}:
default:
if n.verbose {
log.Printf("[%s] Dropping event: %s", n.name, EventJoin)
}
}
return group
}
// leavePeerGroup leaves the peer to a group
func (n *node) leavePeerGroup(peer *peer, name string) *group {
group := n.requirePeerGroup(name)
group.leave(peer)
// Now tell the caller about the peer left group
select {
case n.events <- &Event{eventType: EventLeave, sender: peer.identity, name: peer.name, group: name, time: time.Now()}:
default:
if n.verbose {
log.Printf("[%s] Dropping event: %s", n.name, EventLeave)
}
}
return group
}
// recvFromPeer handles messages coming from other peers
func (n *node) recvFromPeer(transit msg.Transit) {
if transit == nil {
// Invalid transit
return
}
routingID := transit.RoutingID()
if len(routingID) < 1 {
// Invalid routing id, ignore the peer
return
}
// Router socket tells us the identity of this peer
// Identity must be [1] followed by 16-byte UUID, ignore the [1]
identity := fmt.Sprintf("%X", routingID[1:])
peer := n.peers[identity]
if n.verbose {
for i, str := range strings.Split(transit.String(), "\n") {
if len(str) <= 0 {
continue
}
if i == 0 && peer != nil {
log.Printf("[%s] %s %s", n.name, peer.name, str)
} else {
log.Printf("[%s] %s", n.name, str)
}
}
}
switch m := transit.(type) {
case *msg.Hello:
// On HELLO we may create the peer if it's unknown
// On other cmds the peer must already exist
if peer != nil {
// Remove fake peers
if peer.ready {
n.removePeer(peer)
} else if n.endpoint == peer.endpoint {
// We ignore HELLO, if peer has same endpoint as current node
return
}
}
var err error
peer, err = n.requirePeer(identity, m.Endpoint)
if err == nil {
peer.ready = true
} else if n.verbose {
log.Printf("[%s] %s", n.name, err)
}
}
// Ignore command if peer isn't ready
if peer == nil || !peer.ready {
if peer != nil {
n.removePeer(peer)
}
return
}
if !peer.checkMessage(transit) {
log.Printf("[%s] lost messages from %s", n.name, identity)
return
}
// Now process each command
switch m := transit.(type) {
case *msg.Hello:
// Store properties from HELLO command into peer
peer.name = m.Name
event := &Event{
eventType: EventEnter,
sender: peer.identity,
name: peer.name,
address: strings.SplitN(strings.TrimPrefix(m.Endpoint, "tcp://"), ":", 2)[0],
headers: make(map[string]string),
time: time.Now(),
}
// Store peer headers for future reference
for key, val := range m.Headers {
peer.headers[key] = val
event.headers[key] = val
}
select {
case n.events <- event:
default:
if n.verbose {
log.Printf("[%s] Dropping event: %s", n.name, EventEnter)
}
}
// Join peer to listed groups
for _, group := range m.Groups {
n.joinPeerGroup(peer, group)
}
// Now take peer's status from HELLO, after joining groups
peer.status = m.Status
// TODO(armen): If peer is a ZRE/LOG collector, connect to it
case *msg.Whisper:
// Pass up to caller API as WHISPER event
select {
case n.events <- &Event{eventType: EventWhisper, sender: identity, name: peer.name, msg: m.Content, time: time.Now()}:
default:
if n.verbose {
log.Printf("[%s] Dropping event: %s", n.name, EventWhisper)
}
}
case *msg.Shout:
// Pass up to caller as SHOUT event
select {
case n.events <- &Event{eventType: EventShout, sender: identity, name: peer.name, group: m.Group, msg: m.Content, time: time.Now()}:
default:
if n.verbose {
log.Printf("[%s] Dropping event: %s", n.name, EventShout)
}
}
case *msg.Ping:
ping := msg.NewPingOk()
peer.send(ping)
case *msg.Join:
n.joinPeerGroup(peer, m.Group)
if m.Status != peer.status {
panic(fmt.Sprintf("[%X] message status isn't equal to peer status, %d != %d", n.uuid, m.Status, peer.status))
}
case *msg.Leave:
n.leavePeerGroup(peer, m.Group)
if m.Status != peer.status {
panic(fmt.Sprintf("[%X] message status isn't equal to peer status, %d != %d", n.uuid, m.Status, peer.status))
}
}
// Activity from peer resets peer timers
peer.refresh()
}
// recvFromBeacon handles a new signal received from beacon
func (n *node) recvFromBeacon(s *beacon.Signal) {
b := &aBeacon{}
buffer := bytes.NewBuffer(s.Transmit)
binary.Read(buffer, binary.BigEndian, &b.Protocol)
binary.Read(buffer, binary.BigEndian, &b.Version)
uuid := make([]byte, 16)
binary.Read(buffer, binary.BigEndian, uuid)
b.UUID = append(b.UUID, uuid...)
binary.Read(buffer, binary.BigEndian, &b.Port)
// Ignore anything that isn't a valid beacon
if b.Version == beaconVersion {
// Check that the peer, identified by its UUID, exists
identity := fmt.Sprintf("%X", b.UUID)
if b.Port != 0 {
var endpoint string
// s.Addr is IP address of peer beacon
ip := net.ParseIP(s.Addr)
if ip.To4() == nil {
endpoint = fmt.Sprintf("tcp://[%s]:%d", ip.String(), b.Port)
} else {
endpoint = fmt.Sprintf("tcp://%s:%d", ip.String(), b.Port)
}
peer, err := n.requirePeer(identity, endpoint)
if err == nil {
peer.refresh()
} else if n.verbose {
log.Printf("[%s] %s", n.name, err)
}
} else {
// Zero port means peer is going away; remove it if
// we had any knowledge of it already
peer := n.peers[identity]
n.removePeer(peer)
}
} else if n.verbose {
log.Printf("[%s] Received a beacon with invalid version number %d", n.name, b.Version)
}
}
// recvFromGossip handles a new response received from gossip
func (n *node) recvFromGossip(r interface{}) {
resp := r.(*zgossip.Resp)
if n.verbose {
log.Printf("[%s] recvFromGossip: %#v", n.name, resp.Payload.(map[string]string))
}
for identity, endpoint := range resp.Payload.(map[string]string) {
if endpoint != n.endpoint {
peer, err := n.requirePeer(identity, endpoint)
if err == nil {
peer.refresh()
} else if n.verbose {
log.Printf("[%s] %s", n.name, err)
}
}
}
}
// We do this once a second:
// - if peer has gone quiet, send TCP ping
// - if peer has disappeared, expire it
func (n *node) pingPeer(peer *peer) {
if time.Now().Unix() >= peer.expiredAt.Unix() {
n.removePeer(peer)
} else if time.Now().Unix() >= peer.evasiveAt.Unix() {
// If peer is being evasive, force a TCP ping.
// TODO(armen): do this only once for a peer in this state;
// it would be nicer to use a proper state machine
// for peer management.
m := msg.NewPing()
peer.send(m)
}
}
// Terminate leaves all the groups and the closes all the connections to the peers
func (n *node) terminate() {
// Disconnect from all peers
for peerID, peer := range n.peers {
// It's really important to disconnect from the peer before
// deleting it, unless we'd end up difficulties to reconnect
// to the same endpoint
peer.disconnect()
delete(n.peers, peerID)
}
// Now it's safe to close the socket
n.inbox.Unbind(fmt.Sprintf("tcp://*:%d", n.port))
n.inbox.Close()
}
func (n *node) actor() {
defer func() {
n.wg.Done()
}()
// Handle terminate signal
n.reactor.AddChannel(n.terminated, 1, func(interface{}) error {
// Quiting
n.stop()
n.terminate()
return errors.New("terminate")
})
// Received a command from the caller/API
n.reactor.AddChannel(n.cmds, 1, func(c interface{}) error {
n.recvFromAPI(c.(*cmd))
return nil
})
// Handle the inbox
n.reactor.AddSocket(n.inbox, zmq.POLLIN, func(s zmq.State) error {
transit, err := msg.Recv(n.inbox)
if err != nil {
if n.verbose {
log.Printf("[%s] %s", n.name, err)
}
return nil
}
n.recvFromPeer(transit)
return nil
})
n.reactor.Run(10 * time.Millisecond)
}
func (n *node) ping() {
if n.verbose && len(n.peers) == 0 {
log.Printf("[%s] There is no peer to ping", n.name)
}
for _, peer := range n.peers {
n.pingPeer(peer)
}
}
func bind(sock *zmq.Socket, endpoint string) (string, uint16, error) {
var port uint16
e, err := url.Parse(endpoint)
if err != nil {
return endpoint, 0, err
}
if e.Scheme == "inproc" {
err = sock.Bind(endpoint)
return endpoint, 0, err
}
ip, p, err := net.SplitHostPort(e.Host)
if err != nil {
return endpoint, 0, err
}
if p == "*" {
for i := dynPortFrom; i <= dynPortTo; i++ {
rand.Seed(time.Now().UTC().UnixNano())
port = uint16(rand.Intn(int(dynPortTo-dynPortFrom))) + dynPortFrom
endpoint = fmt.Sprintf("%s://%s:%d", e.Scheme, ip, port)
err = sock.Bind(endpoint)
if err == nil {
break
} else if err.Error() == "no sock.ch device" {
port = 0
err = fmt.Errorf("no sock.ch device: %s", endpoint)
break
} else if i-dynPortFrom > 100 {
err = errors.New("Unable to bind to an ephemeral port")
break
}
}
return endpoint, port, err
}
pp, err := strconv.ParseUint(p, 10, 16)
if err != nil {
return endpoint, 0, err
}
port = uint16(pp)
err = sock.Bind(endpoint)
return endpoint, port, err
}