forked from hashicorp/consul
/
leader.go
1582 lines (1380 loc) · 44.8 KB
/
leader.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
package consul
import (
"context"
"fmt"
"net"
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
"github.com/armon/go-metrics"
"github.com/hashicorp/consul/acl"
"github.com/hashicorp/consul/agent/connect"
ca "github.com/hashicorp/consul/agent/connect/ca"
"github.com/hashicorp/consul/agent/consul/autopilot"
"github.com/hashicorp/consul/agent/metadata"
"github.com/hashicorp/consul/agent/structs"
"github.com/hashicorp/consul/api"
"github.com/hashicorp/consul/lib"
"github.com/hashicorp/consul/types"
memdb "github.com/hashicorp/go-memdb"
uuid "github.com/hashicorp/go-uuid"
"github.com/hashicorp/go-version"
"github.com/hashicorp/raft"
"github.com/hashicorp/serf/serf"
"golang.org/x/time/rate"
)
const (
newLeaderEvent = "consul:new-leader"
barrierWriteTimeout = 2 * time.Minute
)
var (
// caRootPruneInterval is how often we check for stale CARoots to remove.
caRootPruneInterval = time.Hour
// minAutopilotVersion is the minimum Consul version in which Autopilot features
// are supported.
minAutopilotVersion = version.Must(version.NewVersion("0.8.0"))
)
// monitorLeadership is used to monitor if we acquire or lose our role
// as the leader in the Raft cluster. There is some work the leader is
// expected to do, so we must react to changes
func (s *Server) monitorLeadership() {
// We use the notify channel we configured Raft with, NOT Raft's
// leaderCh, which is only notified best-effort. Doing this ensures
// that we get all notifications in order, which is required for
// cleanup and to ensure we never run multiple leader loops.
raftNotifyCh := s.raftNotifyCh
aclModeCheckWait := aclModeCheckMinInterval
var aclUpgradeCh <-chan time.Time
if s.ACLsEnabled() {
aclUpgradeCh = time.After(aclModeCheckWait)
}
var weAreLeaderCh chan struct{}
var leaderLoop sync.WaitGroup
for {
select {
case isLeader := <-raftNotifyCh:
switch {
case isLeader:
if weAreLeaderCh != nil {
s.logger.Printf("[ERR] consul: attempted to start the leader loop while running")
continue
}
weAreLeaderCh = make(chan struct{})
leaderLoop.Add(1)
go func(ch chan struct{}) {
defer leaderLoop.Done()
s.leaderLoop(ch)
}(weAreLeaderCh)
s.logger.Printf("[INFO] consul: cluster leadership acquired")
default:
if weAreLeaderCh == nil {
s.logger.Printf("[ERR] consul: attempted to stop the leader loop while not running")
continue
}
s.logger.Printf("[DEBUG] consul: shutting down leader loop")
close(weAreLeaderCh)
leaderLoop.Wait()
weAreLeaderCh = nil
s.logger.Printf("[INFO] consul: cluster leadership lost")
}
case <-aclUpgradeCh:
if atomic.LoadInt32(&s.useNewACLs) == 0 {
aclModeCheckWait = aclModeCheckWait * 2
if aclModeCheckWait > aclModeCheckMaxInterval {
aclModeCheckWait = aclModeCheckMaxInterval
}
aclUpgradeCh = time.After(aclModeCheckWait)
if canUpgrade := s.canUpgradeToNewACLs(weAreLeaderCh != nil); canUpgrade {
if weAreLeaderCh != nil {
if err := s.initializeACLs(true); err != nil {
s.logger.Printf("[ERR] consul: error transitioning to using new ACLs: %v", err)
continue
}
}
s.logger.Printf("[DEBUG] acl: transitioning out of legacy ACL mode")
atomic.StoreInt32(&s.useNewACLs, 1)
s.updateACLAdvertisement()
// setting this to nil ensures that we will never hit this case again
aclUpgradeCh = nil
}
} else {
// establishLeadership probably transitioned us
aclUpgradeCh = nil
}
case <-s.shutdownCh:
return
}
}
}
// leaderLoop runs as long as we are the leader to run various
// maintenance activities
func (s *Server) leaderLoop(stopCh chan struct{}) {
// Fire a user event indicating a new leader
payload := []byte(s.config.NodeName)
for name, segment := range s.LANSegments() {
if err := segment.UserEvent(newLeaderEvent, payload, false); err != nil {
s.logger.Printf("[WARN] consul: failed to broadcast new leader event on segment %q: %v", name, err)
}
}
// Reconcile channel is only used once initial reconcile
// has succeeded
var reconcileCh chan serf.Member
establishedLeader := false
reassert := func() error {
if !establishedLeader {
return fmt.Errorf("leadership has not been established")
}
if err := s.revokeLeadership(); err != nil {
return err
}
if err := s.establishLeadership(); err != nil {
return err
}
return nil
}
RECONCILE:
// Setup a reconciliation timer
reconcileCh = nil
interval := time.After(s.config.ReconcileInterval)
// Apply a raft barrier to ensure our FSM is caught up
start := time.Now()
barrier := s.raft.Barrier(barrierWriteTimeout)
if err := barrier.Error(); err != nil {
s.logger.Printf("[ERR] consul: failed to wait for barrier: %v", err)
goto WAIT
}
metrics.MeasureSince([]string{"leader", "barrier"}, start)
// Check if we need to handle initial leadership actions
if !establishedLeader {
if err := s.establishLeadership(); err != nil {
s.logger.Printf("[ERR] consul: failed to establish leadership: %v", err)
// Immediately revoke leadership since we didn't successfully
// establish leadership.
if err := s.revokeLeadership(); err != nil {
s.logger.Printf("[ERR] consul: failed to revoke leadership: %v", err)
}
goto WAIT
}
establishedLeader = true
defer func() {
if err := s.revokeLeadership(); err != nil {
s.logger.Printf("[ERR] consul: failed to revoke leadership: %v", err)
}
}()
}
// Reconcile any missing data
if err := s.reconcile(); err != nil {
s.logger.Printf("[ERR] consul: failed to reconcile: %v", err)
goto WAIT
}
// Initial reconcile worked, now we can process the channel
// updates
reconcileCh = s.reconcileCh
WAIT:
// Poll the stop channel to give it priority so we don't waste time
// trying to perform the other operations if we have been asked to shut
// down.
select {
case <-stopCh:
return
default:
}
// Periodically reconcile as long as we are the leader,
// or when Serf events arrive
for {
select {
case <-stopCh:
return
case <-s.shutdownCh:
return
case <-interval:
goto RECONCILE
case member := <-reconcileCh:
s.reconcileMember(member)
case index := <-s.tombstoneGC.ExpireCh():
go s.reapTombstones(index)
case errCh := <-s.reassertLeaderCh:
errCh <- reassert()
}
}
}
// establishLeadership is invoked once we become leader and are able
// to invoke an initial barrier. The barrier is used to ensure any
// previously inflight transactions have been committed and that our
// state is up-to-date.
func (s *Server) establishLeadership() error {
// check for the upgrade here - this helps us transition to new ACLs much
// quicker if this is a new cluster or this is a test agent
if canUpgrade := s.canUpgradeToNewACLs(true); canUpgrade {
if err := s.initializeACLs(true); err != nil {
return err
}
atomic.StoreInt32(&s.useNewACLs, 1)
s.updateACLAdvertisement()
} else if err := s.initializeACLs(false); err != nil {
return err
}
// Hint the tombstone expiration timer. When we freshly establish leadership
// we become the authoritative timer, and so we need to start the clock
// on any pending GC events.
s.tombstoneGC.SetEnabled(true)
lastIndex := s.raft.LastIndex()
s.tombstoneGC.Hint(lastIndex)
// Setup the session timers. This is done both when starting up or when
// a leader fail over happens. Since the timers are maintained by the leader
// node along, effectively this means all the timers are renewed at the
// time of failover. The TTL contract is that the session will not be expired
// before the TTL, so expiring it later is allowable.
//
// This MUST be done after the initial barrier to ensure the latest Sessions
// are available to be initialized. Otherwise initialization may use stale
// data.
if err := s.initializeSessionTimers(); err != nil {
return err
}
s.getOrCreateAutopilotConfig()
s.autopilot.Start()
// todo(kyhavlov): start a goroutine here for handling periodic CA rotation
if err := s.initializeCA(); err != nil {
return err
}
s.startEnterpriseLeader()
s.startCARootPruning()
s.setConsistentReadReady()
return nil
}
// revokeLeadership is invoked once we step down as leader.
// This is used to cleanup any state that may be specific to a leader.
func (s *Server) revokeLeadership() error {
// Disable the tombstone GC, since it is only useful as a leader
s.tombstoneGC.SetEnabled(false)
// Clear the session timers on either shutdown or step down, since we
// are no longer responsible for session expirations.
if err := s.clearAllSessionTimers(); err != nil {
return err
}
s.stopEnterpriseLeader()
s.stopCARootPruning()
s.setCAProvider(nil, nil)
s.stopACLUpgrade()
s.resetConsistentReadReady()
s.autopilot.Stop()
return nil
}
// DEPRECATED (ACL-Legacy-Compat) - Remove once old ACL compatibility is removed
func (s *Server) initializeLegacyACL() error {
if !s.ACLsEnabled() {
return nil
}
authDC := s.config.ACLDatacenter
// Create anonymous token if missing.
state := s.fsm.State()
_, token, err := state.ACLTokenGetBySecret(nil, anonymousToken)
if err != nil {
return fmt.Errorf("failed to get anonymous token: %v", err)
}
if token == nil {
req := structs.ACLRequest{
Datacenter: authDC,
Op: structs.ACLSet,
ACL: structs.ACL{
ID: anonymousToken,
Name: "Anonymous Token",
Type: structs.ACLTokenTypeClient,
},
}
_, err := s.raftApply(structs.ACLRequestType, &req)
if err != nil {
return fmt.Errorf("failed to create anonymous token: %v", err)
}
s.logger.Printf("[INFO] acl: Created the anonymous token")
}
// Check for configured master token.
if master := s.config.ACLMasterToken; len(master) > 0 {
_, token, err = state.ACLTokenGetBySecret(nil, master)
if err != nil {
return fmt.Errorf("failed to get master token: %v", err)
}
if token == nil {
req := structs.ACLRequest{
Datacenter: authDC,
Op: structs.ACLSet,
ACL: structs.ACL{
ID: master,
Name: "Master Token",
Type: structs.ACLTokenTypeManagement,
},
}
_, err := s.raftApply(structs.ACLRequestType, &req)
if err != nil {
return fmt.Errorf("failed to create master token: %v", err)
}
s.logger.Printf("[INFO] consul: Created ACL master token from configuration")
}
}
// Check to see if we need to initialize the ACL bootstrap info. This
// needs a Consul version check since it introduces a new Raft operation
// that'll produce an error on older servers, and it also makes a piece
// of state in the state store that will cause problems with older
// servers consuming snapshots, so we have to wait to create it.
var minVersion = version.Must(version.NewVersion("0.9.1"))
if ServersMeetMinimumVersion(s.LANMembers(), minVersion) {
canBootstrap, _, err := state.CanBootstrapACLToken()
if err != nil {
return fmt.Errorf("failed looking for ACL bootstrap info: %v", err)
}
if canBootstrap {
req := structs.ACLRequest{
Datacenter: authDC,
Op: structs.ACLBootstrapInit,
}
resp, err := s.raftApply(structs.ACLRequestType, &req)
if err != nil {
return fmt.Errorf("failed to initialize ACL bootstrap: %v", err)
}
switch v := resp.(type) {
case error:
return fmt.Errorf("failed to initialize ACL bootstrap: %v", v)
case bool:
if v {
s.logger.Printf("[INFO] consul: ACL bootstrap enabled")
} else {
s.logger.Printf("[INFO] consul: ACL bootstrap disabled, existing management tokens found")
}
default:
return fmt.Errorf("unexpected response trying to initialize ACL bootstrap: %T", v)
}
}
} else {
s.logger.Printf("[WARN] consul: Can't initialize ACL bootstrap until all servers are >= %s", minVersion.String())
}
return nil
}
// initializeACLs is used to setup the ACLs if we are the leader
// and need to do this.
func (s *Server) initializeACLs(upgrade bool) error {
if !s.ACLsEnabled() {
return nil
}
// Purge the cache, since it could've changed while we were not the
// leader.
s.acls.cache.Purge()
if s.InACLDatacenter() {
if s.UseLegacyACLs() && !upgrade {
s.logger.Printf("[INFO] acl: initializing legacy acls")
return s.initializeLegacyACL()
}
s.logger.Printf("[INFO] acl: initializing acls")
// Create the builtin global-management policy
_, policy, err := s.fsm.State().ACLPolicyGetByID(nil, structs.ACLPolicyGlobalManagementID)
if err != nil {
return fmt.Errorf("failed to get the builtin global-management policy")
}
if policy == nil {
policy := structs.ACLPolicy{
ID: structs.ACLPolicyGlobalManagementID,
Name: "global-management",
Description: "Builtin Policy that grants unlimited access",
Rules: structs.ACLPolicyGlobalManagement,
Syntax: acl.SyntaxCurrent,
}
policy.SetHash(true)
req := structs.ACLPolicyBatchSetRequest{
Policies: structs.ACLPolicies{&policy},
}
_, err := s.raftApply(structs.ACLPolicySetRequestType, &req)
if err != nil {
return fmt.Errorf("failed to create global-management policy: %v", err)
}
s.logger.Printf("[INFO] consul: Created ACL 'global-management' policy")
}
// Check for configured master token.
if master := s.config.ACLMasterToken; len(master) > 0 {
state := s.fsm.State()
if _, err := uuid.ParseUUID(master); err != nil {
s.logger.Printf("[WARN] consul: Configuring a non-UUID master token is deprecated")
}
_, token, err := state.ACLTokenGetBySecret(nil, master)
if err != nil {
return fmt.Errorf("failed to get master token: %v", err)
}
if token == nil {
accessor, err := lib.GenerateUUID(s.checkTokenUUID)
if err != nil {
return fmt.Errorf("failed to generate the accessor ID for the master token: %v", err)
}
token := structs.ACLToken{
AccessorID: accessor,
SecretID: master,
Description: "Master Token",
Policies: []structs.ACLTokenPolicyLink{
{
ID: structs.ACLPolicyGlobalManagementID,
},
},
CreateTime: time.Now(),
Local: false,
// DEPRECATED (ACL-Legacy-Compat) - only needed for compatibility
Type: structs.ACLTokenTypeManagement,
}
token.SetHash(true)
done := false
if canBootstrap, _, err := state.CanBootstrapACLToken(); err == nil && canBootstrap {
req := structs.ACLTokenBootstrapRequest{
Token: token,
ResetIndex: 0,
}
if _, err := s.raftApply(structs.ACLBootstrapRequestType, &req); err == nil {
s.logger.Printf("[INFO] consul: Bootstrapped ACL master token from configuration")
done = true
} else {
if err.Error() != structs.ACLBootstrapNotAllowedErr.Error() &&
err.Error() != structs.ACLBootstrapInvalidResetIndexErr.Error() {
return fmt.Errorf("failed to bootstrap master token: %v", err)
}
}
}
if !done {
// either we didn't attempt to or setting the token with a bootstrap request failed.
req := structs.ACLTokenBatchSetRequest{
Tokens: structs.ACLTokens{&token},
CAS: false,
}
if _, err := s.raftApply(structs.ACLTokenSetRequestType, &req); err != nil {
return fmt.Errorf("failed to create master token: %v", err)
}
s.logger.Printf("[INFO] consul: Created ACL master token from configuration")
}
}
}
state := s.fsm.State()
_, token, err := state.ACLTokenGetBySecret(nil, structs.ACLTokenAnonymousID)
if err != nil {
return fmt.Errorf("failed to get anonymous token: %v", err)
}
if token == nil {
// DEPRECATED (ACL-Legacy-Compat) - Don't need to query for previous "anonymous" token
// check for legacy token that needs an upgrade
_, legacyToken, err := state.ACLTokenGetBySecret(nil, anonymousToken)
if err != nil {
return fmt.Errorf("failed to get anonymous token: %v", err)
}
// the token upgrade routine will take care of upgrading the token if a legacy version exists
if legacyToken == nil {
token = &structs.ACLToken{
AccessorID: structs.ACLTokenAnonymousID,
SecretID: anonymousToken,
Description: "Anonymous Token",
CreateTime: time.Now(),
}
token.SetHash(true)
req := structs.ACLTokenBatchSetRequest{
Tokens: structs.ACLTokens{token},
CAS: false,
}
_, err := s.raftApply(structs.ACLTokenSetRequestType, &req)
if err != nil {
return fmt.Errorf("failed to create anonymous token: %v", err)
}
s.logger.Printf("[INFO] consul: Created ACL anonymous token from configuration")
}
}
s.startACLUpgrade()
} else {
if s.UseLegacyACLs() && !upgrade {
if s.IsACLReplicationEnabled() {
s.startLegacyACLReplication()
}
}
if upgrade {
s.stopACLReplication()
}
// ACL replication is now mandatory
s.startACLReplication()
}
// launch the upgrade go routine to generate accessors for everything
return nil
}
func (s *Server) startACLUpgrade() {
s.aclUpgradeLock.Lock()
defer s.aclUpgradeLock.Unlock()
if s.aclUpgradeEnabled {
return
}
ctx, cancel := context.WithCancel(context.Background())
s.aclUpgradeCancel = cancel
go func() {
limiter := rate.NewLimiter(aclUpgradeRateLimit, int(aclUpgradeRateLimit))
for {
if err := limiter.Wait(ctx); err != nil {
return
}
// actually run the upgrade here
state := s.fsm.State()
tokens, waitCh, err := state.ACLTokenListUpgradeable(aclUpgradeBatchSize)
if err != nil {
s.logger.Printf("[WARN] acl: encountered an error while searching for tokens without accessor ids: %v", err)
}
if len(tokens) == 0 {
ws := memdb.NewWatchSet()
ws.Add(state.AbandonCh())
ws.Add(waitCh)
ws.Add(ctx.Done())
// wait for more tokens to need upgrading or the aclUpgradeCh to be closed
ws.Watch(nil)
continue
}
var newTokens structs.ACLTokens
for _, token := range tokens {
// This should be entirely unnecessary but is just a small safeguard against changing accessor IDs
if token.AccessorID != "" {
continue
}
newToken := *token
if token.SecretID == anonymousToken {
newToken.AccessorID = structs.ACLTokenAnonymousID
} else {
accessor, err := lib.GenerateUUID(s.checkTokenUUID)
if err != nil {
s.logger.Printf("[WARN] acl: failed to generate accessor during token auto-upgrade: %v", err)
continue
}
newToken.AccessorID = accessor
}
// Assign the global-management policy to legacy management tokens
if len(newToken.Policies) == 0 && newToken.Type == structs.ACLTokenTypeManagement {
newToken.Policies = append(newToken.Policies, structs.ACLTokenPolicyLink{ID: structs.ACLPolicyGlobalManagementID})
}
// need to copy these as we are going to do a CAS operation.
newToken.CreateIndex = token.CreateIndex
newToken.ModifyIndex = token.ModifyIndex
newToken.SetHash(true)
newTokens = append(newTokens, &newToken)
}
req := &structs.ACLTokenBatchSetRequest{Tokens: newTokens, CAS: true}
resp, err := s.raftApply(structs.ACLTokenSetRequestType, req)
if err != nil {
s.logger.Printf("[ERR] acl: failed to apply acl token upgrade batch: %v", err)
}
if err, ok := resp.(error); ok {
s.logger.Printf("[ERR] acl: failed to apply acl token upgrade batch: %v", err)
}
}
}()
s.aclUpgradeEnabled = true
}
func (s *Server) stopACLUpgrade() {
s.aclUpgradeLock.Lock()
defer s.aclUpgradeLock.Unlock()
if !s.aclUpgradeEnabled {
return
}
s.aclUpgradeCancel()
s.aclUpgradeCancel = nil
s.aclUpgradeEnabled = false
}
func (s *Server) startLegacyACLReplication() {
s.aclReplicationLock.Lock()
defer s.aclReplicationLock.Unlock()
if s.aclReplicationEnabled {
return
}
s.initReplicationStatus()
ctx, cancel := context.WithCancel(context.Background())
s.aclReplicationCancel = cancel
go func() {
var lastRemoteIndex uint64
limiter := rate.NewLimiter(rate.Limit(s.config.ACLReplicationRate), s.config.ACLReplicationBurst)
for {
if err := limiter.Wait(ctx); err != nil {
return
}
if s.tokens.ACLReplicationToken() == "" {
continue
}
index, exit, err := s.replicateLegacyACLs(lastRemoteIndex, ctx)
if exit {
return
}
if err != nil {
lastRemoteIndex = 0
s.updateACLReplicationStatusError()
s.logger.Printf("[WARN] consul: Legacy ACL replication error (will retry if still leader): %v", err)
} else {
lastRemoteIndex = index
s.updateACLReplicationStatusIndex(index)
s.logger.Printf("[DEBUG] consul: Legacy ACL replication completed through remote index %d", index)
}
}
}()
s.updateACLReplicationStatusRunning(structs.ACLReplicateLegacy)
s.aclReplicationEnabled = true
}
func (s *Server) startACLReplication() {
s.aclReplicationLock.Lock()
defer s.aclReplicationLock.Unlock()
if s.aclReplicationEnabled {
return
}
s.initReplicationStatus()
ctx, cancel := context.WithCancel(context.Background())
s.aclReplicationCancel = cancel
replicationType := structs.ACLReplicatePolicies
go func() {
var failedAttempts uint
limiter := rate.NewLimiter(rate.Limit(s.config.ACLReplicationRate), s.config.ACLReplicationBurst)
var lastRemoteIndex uint64
for {
if err := limiter.Wait(ctx); err != nil {
return
}
if s.tokens.ACLReplicationToken() == "" {
continue
}
index, exit, err := s.replicateACLPolicies(lastRemoteIndex, ctx)
if exit {
return
}
if err != nil {
lastRemoteIndex = 0
s.updateACLReplicationStatusError()
s.logger.Printf("[WARN] consul: ACL policy replication error (will retry if still leader): %v", err)
if (1 << failedAttempts) < aclReplicationMaxRetryBackoff {
failedAttempts++
}
select {
case <-ctx.Done():
return
case <-time.After((1 << failedAttempts) * time.Second):
// do nothing
}
} else {
lastRemoteIndex = index
s.updateACLReplicationStatusIndex(index)
s.logger.Printf("[DEBUG] consul: ACL policy replication completed through remote index %d", index)
failedAttempts = 0
}
}
}()
s.logger.Printf("[INFO] acl: started ACL Policy replication")
if s.config.ACLTokenReplication {
replicationType = structs.ACLReplicateTokens
go func() {
var failedAttempts uint
limiter := rate.NewLimiter(rate.Limit(s.config.ACLReplicationRate), s.config.ACLReplicationBurst)
var lastRemoteIndex uint64
for {
if err := limiter.Wait(ctx); err != nil {
return
}
if s.tokens.ACLReplicationToken() == "" {
continue
}
index, exit, err := s.replicateACLTokens(lastRemoteIndex, ctx)
if exit {
return
}
if err != nil {
lastRemoteIndex = 0
s.updateACLReplicationStatusError()
s.logger.Printf("[WARN] consul: ACL token replication error (will retry if still leader): %v", err)
if (1 << failedAttempts) < aclReplicationMaxRetryBackoff {
failedAttempts++
}
select {
case <-ctx.Done():
return
case <-time.After((1 << failedAttempts) * time.Second):
// do nothing
}
} else {
lastRemoteIndex = index
s.updateACLReplicationStatusTokenIndex(index)
s.logger.Printf("[DEBUG] consul: ACL token replication completed through remote index %d", index)
failedAttempts = 0
}
}
}()
s.logger.Printf("[INFO] acl: started ACL Token replication")
}
s.updateACLReplicationStatusRunning(replicationType)
s.aclReplicationEnabled = true
}
func (s *Server) stopACLReplication() {
s.aclReplicationLock.Lock()
defer s.aclReplicationLock.Unlock()
if !s.aclReplicationEnabled {
return
}
s.aclReplicationCancel()
s.aclReplicationCancel = nil
s.updateACLReplicationStatusStopped()
s.aclReplicationEnabled = false
}
// getOrCreateAutopilotConfig is used to get the autopilot config, initializing it if necessary
func (s *Server) getOrCreateAutopilotConfig() *autopilot.Config {
state := s.fsm.State()
_, config, err := state.AutopilotConfig()
if err != nil {
s.logger.Printf("[ERR] autopilot: failed to get config: %v", err)
return nil
}
if config != nil {
return config
}
if !ServersMeetMinimumVersion(s.LANMembers(), minAutopilotVersion) {
s.logger.Printf("[WARN] autopilot: can't initialize until all servers are >= %s", minAutopilotVersion.String())
return nil
}
config = s.config.AutopilotConfig
req := structs.AutopilotSetConfigRequest{Config: *config}
if _, err = s.raftApply(structs.AutopilotRequestType, req); err != nil {
s.logger.Printf("[ERR] autopilot: failed to initialize config: %v", err)
return nil
}
return config
}
// initializeCAConfig is used to initialize the CA config if necessary
// when setting up the CA during establishLeadership
func (s *Server) initializeCAConfig() (*structs.CAConfiguration, error) {
state := s.fsm.State()
_, config, err := state.CAConfig()
if err != nil {
return nil, err
}
if config != nil {
return config, nil
}
config = s.config.CAConfig
if config.ClusterID == "" {
id, err := uuid.GenerateUUID()
if err != nil {
return nil, err
}
config.ClusterID = id
}
req := structs.CARequest{
Op: structs.CAOpSetConfig,
Config: config,
}
if _, err = s.raftApply(structs.ConnectCARequestType, req); err != nil {
return nil, err
}
return config, nil
}
// initializeRootCA runs the initialization logic for a root CA.
func (s *Server) initializeRootCA(provider ca.Provider, conf *structs.CAConfiguration) error {
if err := provider.Configure(conf.ClusterID, true, conf.Config); err != nil {
return fmt.Errorf("error configuring provider: %v", err)
}
if err := provider.GenerateRoot(); err != nil {
return fmt.Errorf("error generating CA root certificate: %v", err)
}
// Get the active root cert from the CA
rootPEM, err := provider.ActiveRoot()
if err != nil {
return fmt.Errorf("error getting root cert: %v", err)
}
rootCA, err := parseCARoot(rootPEM, conf.Provider, conf.ClusterID)
if err != nil {
return err
}
// Check if the CA root is already initialized and exit if it is,
// adding on any existing intermediate certs since they aren't directly
// tied to the provider.
// Every change to the CA after this initial bootstrapping should
// be done through the rotation process.
state := s.fsm.State()
_, activeRoot, err := state.CARootActive(nil)
if err != nil {
return err
}
if activeRoot != nil {
// This state shouldn't be possible to get into because we update the root and
// CA config in the same FSM operation.
if activeRoot.ID != rootCA.ID {
return fmt.Errorf("stored CA root %q is not the active root (%s)", rootCA.ID, activeRoot.ID)
}
rootCA.IntermediateCerts = activeRoot.IntermediateCerts
s.setCAProvider(provider, rootCA)
return nil
}
// Get the highest index
idx, _, err := state.CARoots(nil)
if err != nil {
return err
}
// Store the root cert in raft
resp, err := s.raftApply(structs.ConnectCARequestType, &structs.CARequest{
Op: structs.CAOpSetRoots,
Index: idx,
Roots: []*structs.CARoot{rootCA},
})
if err != nil {
s.logger.Printf("[ERR] connect: Apply failed %v", err)
return err
}
if respErr, ok := resp.(error); ok {
return respErr
}
s.setCAProvider(provider, rootCA)
s.logger.Printf("[INFO] connect: initialized primary datacenter CA with provider %q", conf.Provider)
return nil
}
// parseCARoot returns a filled-in structs.CARoot from a raw PEM value.
func parseCARoot(pemValue, provider, clusterID string) (*structs.CARoot, error) {
id, err := connect.CalculateCertFingerprint(pemValue)
if err != nil {
return nil, fmt.Errorf("error parsing root fingerprint: %v", err)
}
rootCert, err := connect.ParseCert(pemValue)
if err != nil {
return nil, fmt.Errorf("error parsing root cert: %v", err)
}
return &structs.CARoot{
ID: id,
Name: fmt.Sprintf("%s CA Root Cert", strings.Title(provider)),
SerialNumber: rootCert.SerialNumber.Uint64(),
SigningKeyID: connect.HexString(rootCert.AuthorityKeyId),
ExternalTrustDomain: clusterID,
NotBefore: rootCert.NotBefore,
NotAfter: rootCert.NotAfter,
RootCert: pemValue,
Active: true,
}, nil
}
// createProvider returns a connect CA provider from the given config.
func (s *Server) createCAProvider(conf *structs.CAConfiguration) (ca.Provider, error) {
switch conf.Provider {
case structs.ConsulCAProvider:
return &ca.ConsulProvider{Delegate: &consulCADelegate{s}}, nil
case structs.VaultCAProvider:
return &ca.VaultProvider{}, nil
default:
return nil, fmt.Errorf("unknown CA provider %q", conf.Provider)
}
}
func (s *Server) getCAProvider() (ca.Provider, *structs.CARoot) {