/
events.go
1940 lines (1776 loc) · 55.5 KB
/
events.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
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Copyright 2018-2021 The NATS Authors
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package server
import (
"bytes"
"encoding/json"
"errors"
"fmt"
"math/rand"
"net/http"
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
"github.com/nats-io/jwt/v2"
"github.com/nats-io/nats-server/v2/server/pse"
)
const (
accLookupReqTokens = 6
accLookupReqSubj = "$SYS.REQ.ACCOUNT.%s.CLAIMS.LOOKUP"
accPackReqSubj = "$SYS.REQ.CLAIMS.PACK"
accListReqSubj = "$SYS.REQ.CLAIMS.LIST"
accClaimsReqSubj = "$SYS.REQ.CLAIMS.UPDATE"
accDeleteReqSubj = "$SYS.REQ.CLAIMS.DELETE"
connectEventSubj = "$SYS.ACCOUNT.%s.CONNECT"
disconnectEventSubj = "$SYS.ACCOUNT.%s.DISCONNECT"
accReqSubj = "$SYS.REQ.ACCOUNT.%s.%s"
// kept for backward compatibility when using http resolver
// this overlaps with the names for events but you'd have to have the operator private key in order to succeed.
accUpdateEventSubjOld = "$SYS.ACCOUNT.%s.CLAIMS.UPDATE"
accUpdateEventSubjNew = "$SYS.REQ.ACCOUNT.%s.CLAIMS.UPDATE"
connsRespSubj = "$SYS._INBOX_.%s"
accConnsEventSubjNew = "$SYS.ACCOUNT.%s.SERVER.CONNS"
accConnsEventSubjOld = "$SYS.SERVER.ACCOUNT.%s.CONNS" // kept for backward compatibility
shutdownEventSubj = "$SYS.SERVER.%s.SHUTDOWN"
authErrorEventSubj = "$SYS.SERVER.%s.CLIENT.AUTH.ERR"
serverStatsSubj = "$SYS.SERVER.%s.STATSZ"
serverDirectReqSubj = "$SYS.REQ.SERVER.%s.%s"
serverPingReqSubj = "$SYS.REQ.SERVER.PING.%s"
serverStatsPingReqSubj = "$SYS.REQ.SERVER.PING" // use $SYS.REQ.SERVER.PING.STATSZ instead
leafNodeConnectEventSubj = "$SYS.ACCOUNT.%s.LEAFNODE.CONNECT"
remoteLatencyEventSubj = "$SYS.LATENCY.M2.%s"
inboxRespSubj = "$SYS._INBOX.%s.%s"
// FIXME(dlc) - Should account scope, even with wc for now, but later on
// we can then shard as needed.
accNumSubsReqSubj = "$SYS.REQ.ACCOUNT.NSUBS"
// These are for exported debug services. These are local to this server only.
accSubsSubj = "$SYS.DEBUG.SUBSCRIBERS"
shutdownEventTokens = 4
serverSubjectIndex = 2
accUpdateTokensNew = 6
accUpdateTokensOld = 5
accUpdateAccIdxOld = 2
accReqTokens = 5
accReqAccIndex = 3
)
// FIXME(dlc) - make configurable.
var eventsHBInterval = 30 * time.Second
// Used to send and receive messages from inside the server.
type internal struct {
account *Account
client *client
seq uint64
sid int
servers map[string]*serverUpdate
sweeper *time.Timer
stmr *time.Timer
replies map[string]msgHandler
sendq chan *pubMsg
resetCh chan struct{}
wg sync.WaitGroup
sq *sendq
orphMax time.Duration
chkOrph time.Duration
statsz time.Duration
cstatsz time.Duration
shash string
inboxPre string
}
// ServerStatsMsg is sent periodically with stats updates.
type ServerStatsMsg struct {
Server ServerInfo `json:"server"`
Stats ServerStats `json:"statsz"`
}
// ConnectEventMsg is sent when a new connection is made that is part of an account.
type ConnectEventMsg struct {
TypedEvent
Server ServerInfo `json:"server"`
Client ClientInfo `json:"client"`
}
// ConnectEventMsgType is the schema type for ConnectEventMsg
const ConnectEventMsgType = "io.nats.server.advisory.v1.client_connect"
// DisconnectEventMsg is sent when a new connection previously defined from a
// ConnectEventMsg is closed.
type DisconnectEventMsg struct {
TypedEvent
Server ServerInfo `json:"server"`
Client ClientInfo `json:"client"`
Sent DataStats `json:"sent"`
Received DataStats `json:"received"`
Reason string `json:"reason"`
}
// DisconnectEventMsgType is the schema type for DisconnectEventMsg
const DisconnectEventMsgType = "io.nats.server.advisory.v1.client_disconnect"
// AccountNumConns is an event that will be sent from a server that is tracking
// a given account when the number of connections changes. It will also HB
// updates in the absence of any changes.
type AccountNumConns struct {
TypedEvent
Server ServerInfo `json:"server"`
Account string `json:"acc"`
Conns int `json:"conns"`
LeafNodes int `json:"leafnodes"`
TotalConns int `json:"total_conns"`
}
const AccountNumConnsMsgType = "io.nats.server.advisory.v1.account_connections"
// accNumConnsReq is sent when we are starting to track an account for the first
// time. We will request others send info to us about their local state.
type accNumConnsReq struct {
Server ServerInfo `json:"server"`
Account string `json:"acc"`
}
// ServerInfo identifies remote servers.
type ServerInfo struct {
Name string `json:"name"`
Host string `json:"host"`
ID string `json:"id"`
Cluster string `json:"cluster,omitempty"`
Version string `json:"ver"`
Seq uint64 `json:"seq"`
JetStream bool `json:"jetstream"`
Time time.Time `json:"time"`
}
// ClientInfo is detailed information about the client forming a connection.
type ClientInfo struct {
Start *time.Time `json:"start,omitempty"`
Host string `json:"host,omitempty"`
ID uint64 `json:"id,omitempty"`
Account string `json:"acc"`
Service string `json:"svc,omitempty"`
User string `json:"user,omitempty"`
Name string `json:"name,omitempty"`
Lang string `json:"lang,omitempty"`
Version string `json:"ver,omitempty"`
RTT time.Duration `json:"rtt,omitempty"`
Server string `json:"server,omitempty"`
Cluster string `json:"cluster,omitempty"`
Stop *time.Time `json:"stop,omitempty"`
Jwt string `json:"jwt,omitempty"`
IssuerKey string `json:"issuer_key,omitempty"`
NameTag string `json:"name_tag,omitempty"`
Tags jwt.TagList `json:"tags,omitempty"`
}
// ServerStats hold various statistics that we will periodically send out.
type ServerStats struct {
Start time.Time `json:"start"`
Mem int64 `json:"mem"`
Cores int `json:"cores"`
CPU float64 `json:"cpu"`
Connections int `json:"connections"`
TotalConnections uint64 `json:"total_connections"`
ActiveAccounts int `json:"active_accounts"`
NumSubs uint32 `json:"subscriptions"`
Sent DataStats `json:"sent"`
Received DataStats `json:"received"`
SlowConsumers int64 `json:"slow_consumers"`
Routes []*RouteStat `json:"routes,omitempty"`
Gateways []*GatewayStat `json:"gateways,omitempty"`
}
// RouteStat holds route statistics.
type RouteStat struct {
ID uint64 `json:"rid"`
Name string `json:"name,omitempty"`
Sent DataStats `json:"sent"`
Received DataStats `json:"received"`
Pending int `json:"pending"`
}
// GatewayStat holds gateway statistics.
type GatewayStat struct {
ID uint64 `json:"gwid"`
Name string `json:"name"`
Sent DataStats `json:"sent"`
Received DataStats `json:"received"`
NumInbound int `json:"inbound_connections"`
}
// DataStats reports how may msg and bytes. Applicable for both sent and received.
type DataStats struct {
Msgs int64 `json:"msgs"`
Bytes int64 `json:"bytes"`
}
// Used for internally queueing up messages that the server wants to send.
type pubMsg struct {
c *client
sub string
rply string
si *ServerInfo
msg interface{}
last bool
}
// Used to track server updates.
type serverUpdate struct {
seq uint64
ltime time.Time
}
// TypedEvent is a event or advisory sent by the server that has nats type hints
// typically used for events that might be consumed by 3rd party event systems
type TypedEvent struct {
Type string `json:"type"`
ID string `json:"id"`
Time time.Time `json:"timestamp"`
}
// internalSendLoop will be responsible for serializing all messages that
// a server wants to send.
func (s *Server) internalSendLoop(wg *sync.WaitGroup) {
defer wg.Done()
RESET:
s.mu.Lock()
if s.sys == nil || s.sys.sendq == nil {
s.mu.Unlock()
return
}
sysc := s.sys.client
resetCh := s.sys.resetCh
sendq := s.sys.sendq
id := s.info.ID
host := s.info.Host
servername := s.info.Name
seqp := &s.sys.seq
js := s.info.JetStream
cluster := s.info.Cluster
if s.gateway.enabled {
cluster = s.getGatewayName()
}
s.mu.Unlock()
// Warn when internal send queue is backed up past 75%
warnThresh := 3 * internalSendQLen / 4
warnFreq := time.Second
last := time.Now().Add(-warnFreq)
for s.eventsRunning() {
// Setup information for next message
if len(sendq) > warnThresh && time.Since(last) >= warnFreq {
s.Warnf("Internal system send queue > 75%%")
last = time.Now()
}
select {
case pm := <-sendq:
if pm.si != nil {
pm.si.Name = servername
pm.si.Host = host
pm.si.Cluster = cluster
pm.si.ID = id
pm.si.Seq = atomic.AddUint64(seqp, 1)
pm.si.Version = VERSION
pm.si.Time = time.Now().UTC()
pm.si.JetStream = js
}
var b []byte
if pm.msg != nil {
switch v := pm.msg.(type) {
case string:
b = []byte(v)
case []byte:
b = v
default:
b, _ = json.Marshal(pm.msg)
}
}
// Setup our client. If the user wants to use a non-system account use our internal
// account scoped here so that we are not changing out accounts for the system client.
var c *client
if pm.c != nil {
c = pm.c
} else {
c = sysc
}
// Grab client lock.
c.mu.Lock()
// Prep internal structures needed to send message.
c.pa.subject = []byte(pm.sub)
c.pa.size = len(b)
c.pa.szb = []byte(strconv.FormatInt(int64(len(b)), 10))
c.pa.reply = []byte(pm.rply)
trace := c.trace
c.mu.Unlock()
// Add in NL
b = append(b, _CRLF_...)
if trace {
c.traceInOp(fmt.Sprintf("PUB %s %s %d", c.pa.subject, c.pa.reply, c.pa.size), nil)
c.traceMsg(b)
}
// Process like a normal inbound msg.
c.processInboundClientMsg(b)
// See if we are doing graceful shutdown.
if !pm.last {
c.flushClients(0) // Never spend time in place.
} else {
// For the Shutdown event, we need to send in place otherwise
// there is a chance that the process will exit before the
// writeLoop has a chance to send it.
c.flushClients(time.Second)
return
}
case <-resetCh:
goto RESET
case <-s.quitCh:
return
}
}
}
// Will send a shutdown message.
func (s *Server) sendShutdownEvent() {
s.mu.Lock()
if s.sys == nil || s.sys.sendq == nil {
s.mu.Unlock()
return
}
subj := fmt.Sprintf(shutdownEventSubj, s.info.ID)
sendq := s.sys.sendq
// Stop any more messages from queueing up.
s.sys.sendq = nil
// Unhook all msgHandlers. Normal client cleanup will deal with subs, etc.
s.sys.replies = nil
s.mu.Unlock()
// Send to the internal queue and mark as last.
si := &ServerInfo{}
sendq <- &pubMsg{nil, subj, _EMPTY_, si, si, true}
}
// Used to send an internal message to an arbitrary account.
func (s *Server) sendInternalAccountMsg(a *Account, subject string, msg interface{}) error {
s.mu.Lock()
if s.sys == nil || s.sys.sendq == nil {
s.mu.Unlock()
return ErrNoSysAccount
}
sendq := s.sys.sendq
// Don't hold lock while placing on the channel.
c := s.sys.client
s.mu.Unlock()
// Replace our client with the account's internal client.
if a != nil {
a.mu.Lock()
c = a.internalClient()
a.mu.Unlock()
}
sendq <- &pubMsg{c, subject, _EMPTY_, nil, msg, false}
return nil
}
// This will queue up a message to be sent.
// Lock should not be held.
func (s *Server) sendInternalMsgLocked(sub, rply string, si *ServerInfo, msg interface{}) {
s.mu.Lock()
s.sendInternalMsg(sub, rply, si, msg)
s.mu.Unlock()
}
// This will queue up a message to be sent.
// Assumes lock is held on entry.
func (s *Server) sendInternalMsg(sub, rply string, si *ServerInfo, msg interface{}) {
if s.sys == nil || s.sys.sendq == nil {
return
}
sendq := s.sys.sendq
// Don't hold lock while placing on the channel.
s.mu.Unlock()
sendq <- &pubMsg{nil, sub, rply, si, msg, false}
s.mu.Lock()
}
// Used to send internal messages from other system clients to avoid no echo issues.
func (c *client) sendInternalMsg(sub, rply string, si *ServerInfo, msg interface{}) {
if c == nil {
return
}
s := c.srv
if s == nil {
return
}
s.mu.Lock()
if s.sys == nil || s.sys.sendq == nil {
return
}
sendq := s.sys.sendq
// Don't hold lock while placing on the channel.
s.mu.Unlock()
sendq <- &pubMsg{c, sub, rply, si, msg, false}
}
// Locked version of checking if events system running. Also checks server.
func (s *Server) eventsRunning() bool {
if s == nil {
return false
}
s.mu.Lock()
er := s.running && s.eventsEnabled()
s.mu.Unlock()
return er
}
// EventsEnabled will report if the server has internal events enabled via
// a defined system account.
func (s *Server) EventsEnabled() bool {
s.mu.Lock()
ee := s.eventsEnabled()
s.mu.Unlock()
return ee
}
// eventsEnabled will report if events are enabled.
// Lock should be held.
func (s *Server) eventsEnabled() bool {
return s.sys != nil && s.sys.client != nil && s.sys.account != nil
}
// TrackedRemoteServers returns how many remote servers we are tracking
// from a system events perspective.
func (s *Server) TrackedRemoteServers() int {
s.mu.Lock()
if !s.running || !s.eventsEnabled() {
return -1
}
ns := len(s.sys.servers)
s.mu.Unlock()
return ns
}
// Check for orphan servers who may have gone away without notification.
// This should be wrapChk() to setup common locking.
func (s *Server) checkRemoteServers() {
now := time.Now()
for sid, su := range s.sys.servers {
if now.Sub(su.ltime) > s.sys.orphMax {
s.Debugf("Detected orphan remote server: %q", sid)
// Simulate it going away.
s.processRemoteServerShutdown(sid)
delete(s.sys.servers, sid)
}
}
if s.sys.sweeper != nil {
s.sys.sweeper.Reset(s.sys.chkOrph)
}
}
// Grab RSS and PCPU
// Server lock will be held but released.
func (s *Server) updateServerUsage(v *ServerStats) {
s.mu.Unlock()
defer s.mu.Lock()
var vss int64
pse.ProcUsage(&v.CPU, &v.Mem, &vss)
v.Cores = numCores
}
// Generate a route stat for our statz update.
func routeStat(r *client) *RouteStat {
if r == nil {
return nil
}
r.mu.Lock()
rs := &RouteStat{
ID: r.cid,
Sent: DataStats{
Msgs: atomic.LoadInt64(&r.outMsgs),
Bytes: atomic.LoadInt64(&r.outBytes),
},
Received: DataStats{
Msgs: atomic.LoadInt64(&r.inMsgs),
Bytes: atomic.LoadInt64(&r.inBytes),
},
Pending: int(r.out.pb),
}
if r.route != nil {
rs.Name = r.route.remoteName
}
r.mu.Unlock()
return rs
}
// Actual send method for statz updates.
// Lock should be held.
func (s *Server) sendStatsz(subj string) {
m := ServerStatsMsg{}
s.updateServerUsage(&m.Stats)
m.Stats.Start = s.start
m.Stats.Connections = len(s.clients)
m.Stats.TotalConnections = s.totalClients
m.Stats.ActiveAccounts = int(atomic.LoadInt32(&s.activeAccounts))
m.Stats.Received.Msgs = atomic.LoadInt64(&s.inMsgs)
m.Stats.Received.Bytes = atomic.LoadInt64(&s.inBytes)
m.Stats.Sent.Msgs = atomic.LoadInt64(&s.outMsgs)
m.Stats.Sent.Bytes = atomic.LoadInt64(&s.outBytes)
m.Stats.SlowConsumers = atomic.LoadInt64(&s.slowConsumers)
m.Stats.NumSubs = s.numSubscriptions()
for _, r := range s.routes {
m.Stats.Routes = append(m.Stats.Routes, routeStat(r))
}
if s.gateway.enabled {
gw := s.gateway
gw.RLock()
for name, c := range gw.out {
gs := &GatewayStat{Name: name}
c.mu.Lock()
gs.ID = c.cid
gs.Sent = DataStats{
Msgs: atomic.LoadInt64(&c.outMsgs),
Bytes: atomic.LoadInt64(&c.outBytes),
}
c.mu.Unlock()
// Gather matching inbound connections
gs.Received = DataStats{}
for _, c := range gw.in {
c.mu.Lock()
if c.gw.name == name {
gs.Received.Msgs += atomic.LoadInt64(&c.inMsgs)
gs.Received.Bytes += atomic.LoadInt64(&c.inBytes)
gs.NumInbound++
}
c.mu.Unlock()
}
m.Stats.Gateways = append(m.Stats.Gateways, gs)
}
gw.RUnlock()
}
s.sendInternalMsg(subj, _EMPTY_, &m.Server, &m)
}
// Send out our statz update.
// This should be wrapChk() to setup common locking.
func (s *Server) heartbeatStatsz() {
if s.sys.stmr != nil {
// Increase after startup to our max.
s.sys.cstatsz *= 4
if s.sys.cstatsz > s.sys.statsz {
s.sys.cstatsz = s.sys.statsz
}
s.sys.stmr.Reset(s.sys.cstatsz)
}
s.sendStatsz(fmt.Sprintf(serverStatsSubj, s.info.ID))
}
// This should be wrapChk() to setup common locking.
func (s *Server) startStatszTimer() {
// We will start by sending out more of these and trail off to the statsz being the max.
s.sys.cstatsz = time.Second
// Send out the first one only after a second.
s.sys.stmr = time.AfterFunc(s.sys.cstatsz, s.wrapChk(s.heartbeatStatsz))
}
// Start a ticker that will fire periodically and check for orphaned servers.
// This should be wrapChk() to setup common locking.
func (s *Server) startRemoteServerSweepTimer() {
s.sys.sweeper = time.AfterFunc(s.sys.chkOrph, s.wrapChk(s.checkRemoteServers))
}
// Length of our system hash used for server targeted messages.
const sysHashLen = 8
// Computes a hash of 8 characters for the name.
func getHash(name string) []byte {
return getHashSize(name, sysHashLen)
}
// Returns the node name for this server which is a hash of the server name.
func (s *Server) Node() string {
s.mu.Lock()
defer s.mu.Unlock()
if s.sys != nil {
return s.sys.shash
}
return _EMPTY_
}
// This will setup our system wide tracking subs.
// For now we will setup one wildcard subscription to
// monitor all accounts for changes in number of connections.
// We can make this on a per account tracking basis if needed.
// Tradeoff is subscription and interest graph events vs connect and
// disconnect events, etc.
func (s *Server) initEventTracking() {
if !s.EventsEnabled() {
return
}
// Create a system hash which we use for other servers to target us specifically.
s.sys.shash = string(getHash(s.info.Name))
// This will be for all inbox responses.
subject := fmt.Sprintf(inboxRespSubj, s.sys.shash, "*")
if _, err := s.sysSubscribe(subject, s.inboxReply); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
s.sys.inboxPre = subject
// This is for remote updates for connection accounting.
subject = fmt.Sprintf(accConnsEventSubjOld, "*")
if _, err := s.sysSubscribe(subject, s.remoteConnsUpdate); err != nil {
s.Errorf("Error setting up internal tracking for %s: %v", subject, err)
}
// This will be for responses for account info that we send out.
subject = fmt.Sprintf(connsRespSubj, s.info.ID)
if _, err := s.sysSubscribe(subject, s.remoteConnsUpdate); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
// Listen for broad requests to respond with number of subscriptions for a given subject.
if _, err := s.sysSubscribe(accNumSubsReqSubj, s.nsubsRequest); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
// Listen for statsz from others.
subject = fmt.Sprintf(serverStatsSubj, "*")
if _, err := s.sysSubscribe(subject, s.remoteServerUpdate); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
// Listen for all server shutdowns.
subject = fmt.Sprintf(shutdownEventSubj, "*")
if _, err := s.sysSubscribe(subject, s.remoteServerShutdown); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
// Listen for account claims updates.
subscribeToUpdate := true
if s.accResolver != nil {
subscribeToUpdate = !s.accResolver.IsTrackingUpdate()
}
if subscribeToUpdate {
for _, sub := range []string{accUpdateEventSubjOld, accUpdateEventSubjNew} {
if _, err := s.sysSubscribe(fmt.Sprintf(sub, "*"), s.accountClaimUpdate); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
}
}
// Listen for ping messages that will be sent to all servers for statsz.
// This subscription is kept for backwards compatibility. Got replaced by ...PING.STATZ from below
if _, err := s.sysSubscribe(serverStatsPingReqSubj, s.statszReq); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
monSrvc := map[string]msgHandler{
"STATSZ": s.statszReq,
"VARZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &VarzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Varz(&optz.VarzOptions) })
},
"SUBSZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &SubszEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Subsz(&optz.SubszOptions) })
},
"CONNZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &ConnzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Connz(&optz.ConnzOptions) })
},
"ROUTEZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &RoutezEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Routez(&optz.RoutezOptions) })
},
"GATEWAYZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &GatewayzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Gatewayz(&optz.GatewayzOptions) })
},
"LEAFZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &LeafzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Leafz(&optz.LeafzOptions) })
},
"ACCOUNTZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &AccountzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Accountz(&optz.AccountzOptions) })
},
"JSZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &JszEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) { return s.Jsz(&optz.JSzOptions) })
},
}
for name, req := range monSrvc {
subject = fmt.Sprintf(serverDirectReqSubj, s.info.ID, name)
if _, err := s.sysSubscribe(subject, req); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
subject = fmt.Sprintf(serverPingReqSubj, name)
if _, err := s.sysSubscribe(subject, req); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
}
extractAccount := func(subject string) (string, error) {
if tk := strings.Split(subject, tsep); len(tk) != accReqTokens {
return "", fmt.Errorf("subject %q is malformed", subject)
} else {
return tk[accReqAccIndex], nil
}
}
monAccSrvc := map[string]msgHandler{
"SUBSZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &SubszEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) {
if acc, err := extractAccount(subject); err != nil {
return nil, err
} else {
optz.SubszOptions.Subscriptions = true
optz.SubszOptions.Account = acc
return s.Subsz(&optz.SubszOptions)
}
})
},
"CONNZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &ConnzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) {
if acc, err := extractAccount(subject); err != nil {
return nil, err
} else {
optz.ConnzOptions.Account = acc
return s.Connz(&optz.ConnzOptions)
}
})
},
"LEAFZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &LeafzEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) {
if acc, err := extractAccount(subject); err != nil {
return nil, err
} else {
optz.LeafzOptions.Account = acc
return s.Leafz(&optz.LeafzOptions)
}
})
},
"JSZ": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &JszEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) {
if acc, err := extractAccount(subject); err != nil {
return nil, err
} else {
optz.Account = acc
return s.JszAccount(&optz.JSzOptions)
}
})
},
"INFO": func(sub *subscription, _ *client, subject, reply string, msg []byte) {
optz := &AccInfoEventOptions{}
s.zReq(reply, msg, &optz.EventFilterOptions, optz, func() (interface{}, error) {
if acc, err := extractAccount(subject); err != nil {
return nil, err
} else {
return s.accountInfo(acc)
}
})
},
"CONNS": s.connsRequest,
}
for name, req := range monAccSrvc {
if _, err := s.sysSubscribe(fmt.Sprintf(accReqSubj, "*", name), req); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
}
// Listen for updates when leaf nodes connect for a given account. This will
// force any gateway connections to move to `modeInterestOnly`
subject = fmt.Sprintf(leafNodeConnectEventSubj, "*")
if _, err := s.sysSubscribe(subject, s.leafNodeConnected); err != nil {
s.Errorf("Error setting up internal tracking: %v", err)
}
// For tracking remote latency measurements.
subject = fmt.Sprintf(remoteLatencyEventSubj, s.sys.shash)
if _, err := s.sysSubscribe(subject, s.remoteLatencyUpdate); err != nil {
s.Errorf("Error setting up internal latency tracking: %v", err)
}
// This is for simple debugging of number of subscribers that exist in the system.
if _, err := s.sysSubscribeInternal(accSubsSubj, s.debugSubscribers); err != nil {
s.Errorf("Error setting up internal debug service for subscribers: %v", err)
}
}
// add all exports a system account will need
func (s *Server) addSystemAccountExports(sacc *Account) {
if !s.EventsEnabled() {
return
}
if err := sacc.AddServiceExport(accSubsSubj, nil); err != nil {
s.Errorf("Error adding system service export for %q: %v", accSubsSubj, err)
}
if s.JetStreamEnabled() {
s.checkJetStreamExports()
}
}
// accountClaimUpdate will receive claim updates for accounts.
func (s *Server) accountClaimUpdate(sub *subscription, _ *client, subject, resp string, msg []byte) {
if !s.EventsEnabled() {
return
}
pubKey := ""
toks := strings.Split(subject, tsep)
if len(toks) == accUpdateTokensNew {
pubKey = toks[accReqAccIndex]
} else if len(toks) == accUpdateTokensOld {
pubKey = toks[accUpdateAccIdxOld]
} else {
s.Debugf("Received account claims update on bad subject %q", subject)
return
}
if claim, err := jwt.DecodeAccountClaims(string(msg)); err != nil {
respondToUpdate(s, resp, pubKey, "jwt update resulted in error", err)
} else if claim.Subject != pubKey {
err := errors.New("subject does not match jwt content")
respondToUpdate(s, resp, pubKey, "jwt update resulted in error", err)
} else if v, ok := s.accounts.Load(pubKey); !ok {
respondToUpdate(s, resp, pubKey, "jwt update skipped", nil)
} else if err := s.updateAccountWithClaimJWT(v.(*Account), string(msg)); err != nil {
respondToUpdate(s, resp, pubKey, "jwt update resulted in error", err)
} else {
respondToUpdate(s, resp, pubKey, "jwt updated", nil)
}
}
// processRemoteServerShutdown will update any affected accounts.
// Will update the remote count for clients.
// Lock assume held.
func (s *Server) processRemoteServerShutdown(sid string) {
s.accounts.Range(func(k, v interface{}) bool {
v.(*Account).removeRemoteServer(sid)
return true
})
// Update any state in nodeInfo.
s.nodeToInfo.Range(func(k, v interface{}) bool {
si := v.(nodeInfo)
if si.id == sid {
si.offline = true
s.nodeToInfo.Store(k, si)
return false
}
return true
})
}
// remoteServerShutdownEvent is called when we get an event from another server shutting down.
func (s *Server) remoteServerShutdown(sub *subscription, _ *client, subject, reply string, msg []byte) {
s.mu.Lock()
defer s.mu.Unlock()
if !s.eventsEnabled() {
return
}
toks := strings.Split(subject, tsep)
if len(toks) < shutdownEventTokens {
s.Debugf("Received remote server shutdown on bad subject %q", subject)
return
}
sid := toks[serverSubjectIndex]
if su := s.sys.servers[sid]; su != nil {
s.processRemoteServerShutdown(sid)
}
if len(msg) == 0 {
return
}
// We have an optional serverInfo here, remove from nodeToX lookups.
var si ServerInfo
if err := json.Unmarshal(msg, &si); err != nil {
s.Debugf("Received bad server info for remote server shutdown")
return
}
// Additional processing here.
node := string(getHash(si.Name))
s.nodeToInfo.Store(node, nodeInfo{si.Name, si.Cluster, si.ID, true, true})
}
// remoteServerUpdate listens for statsz updates from other servers.
func (s *Server) remoteServerUpdate(sub *subscription, _ *client, subject, reply string, msg []byte) {
var ssm ServerStatsMsg
if err := json.Unmarshal(msg, &ssm); err != nil {
s.Debugf("Received bad server info for remote server update")
return
}
si := ssm.Server
node := string(getHash(si.Name))
if _, ok := s.nodeToInfo.Load(node); !ok {
// Since we have not seen this one they probably have not seen us so send out our update.
s.mu.Lock()
s.sendStatsz(fmt.Sprintf(serverStatsSubj, s.info.ID))
s.mu.Unlock()
}
s.nodeToInfo.Store(node, nodeInfo{si.Name, si.Cluster, si.ID, false, si.JetStream})
}
// updateRemoteServer is called when we have an update from a remote server.
// This allows us to track remote servers, respond to shutdown messages properly,
// make sure that messages are ordered, and allow us to prune dead servers.
// Lock should be held upon entry.
func (s *Server) updateRemoteServer(ms *ServerInfo) {
su := s.sys.servers[ms.ID]
if su == nil {
s.sys.servers[ms.ID] = &serverUpdate{ms.Seq, time.Now()}
s.processNewServer(ms)
} else {
// Should always be going up.
if ms.Seq <= su.seq {
s.Errorf("Received out of order remote server update from: %q", ms.ID)
return
}
su.seq = ms.Seq
su.ltime = time.Now()
}
}
// processNewServer will hold any logic we want to use when we discover a new server.
// Lock should be held upon entry.
func (s *Server) processNewServer(ms *ServerInfo) {
// Right now we only check if we have leafnode servers and if so send another
// connect update to make sure they switch this account to interest only mode.
s.ensureGWsInterestOnlyForLeafNodes()
// Add to our nodeToName
node := string(getHash(ms.Name))
s.nodeToInfo.Store(node, nodeInfo{ms.Name, ms.Cluster, ms.ID, false, ms.JetStream})
}
// If GW is enabled on this server and there are any leaf node connections,
// this function will send a LeafNode connect system event to the super cluster
// to ensure that the GWs are in interest-only mode for this account.
// Lock should be held upon entry.
// TODO(dlc) - this will cause this account to be loaded on all servers. Need a better
// way with GW2.
func (s *Server) ensureGWsInterestOnlyForLeafNodes() {
if !s.gateway.enabled || len(s.leafs) == 0 {
return
}
sent := make(map[*Account]bool, len(s.leafs))
for _, c := range s.leafs {
if !sent[c.acc] {
s.sendLeafNodeConnectMsg(c.acc.Name)
sent[c.acc] = true
}
}
}
// shutdownEventing will clean up all eventing state.
func (s *Server) shutdownEventing() {
if !s.eventsRunning() {
return
}
s.mu.Lock()
clearTimer(&s.sys.sweeper)
clearTimer(&s.sys.stmr)
sys := s.sys
s.mu.Unlock()
// We will queue up a shutdown event and wait for the
// internal send loop to exit.
s.sendShutdownEvent()