/
proxier.go
1643 lines (1470 loc) · 65.9 KB
/
proxier.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 2015 The Kubernetes 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 iptables
//
// NOTE: this needs to be tested in e2e since it uses iptables for everything.
//
import (
"bytes"
"crypto/sha256"
"encoding/base32"
"fmt"
"net"
"reflect"
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
v1 "k8s.io/api/core/v1"
discovery "k8s.io/api/discovery/v1"
"k8s.io/apimachinery/pkg/types"
"k8s.io/apimachinery/pkg/util/sets"
"k8s.io/apimachinery/pkg/util/wait"
utilfeature "k8s.io/apiserver/pkg/util/feature"
"k8s.io/client-go/tools/events"
utilsysctl "k8s.io/component-helpers/node/util/sysctl"
"k8s.io/klog/v2"
"k8s.io/kubernetes/pkg/features"
"k8s.io/kubernetes/pkg/proxy"
"k8s.io/kubernetes/pkg/proxy/healthcheck"
"k8s.io/kubernetes/pkg/proxy/metaproxier"
"k8s.io/kubernetes/pkg/proxy/metrics"
utilproxy "k8s.io/kubernetes/pkg/proxy/util"
proxyutiliptables "k8s.io/kubernetes/pkg/proxy/util/iptables"
"k8s.io/kubernetes/pkg/util/async"
"k8s.io/kubernetes/pkg/util/conntrack"
utiliptables "k8s.io/kubernetes/pkg/util/iptables"
utilexec "k8s.io/utils/exec"
netutils "k8s.io/utils/net"
)
const (
// the services chain
kubeServicesChain utiliptables.Chain = "KUBE-SERVICES"
// the external services chain
kubeExternalServicesChain utiliptables.Chain = "KUBE-EXTERNAL-SERVICES"
// the nodeports chain
kubeNodePortsChain utiliptables.Chain = "KUBE-NODEPORTS"
// the kubernetes postrouting chain
kubePostroutingChain utiliptables.Chain = "KUBE-POSTROUTING"
// kubeMarkMasqChain is the mark-for-masquerade chain
kubeMarkMasqChain utiliptables.Chain = "KUBE-MARK-MASQ"
// the kubernetes forward chain
kubeForwardChain utiliptables.Chain = "KUBE-FORWARD"
// kubeProxyFirewallChain is the kube-proxy firewall chain
kubeProxyFirewallChain utiliptables.Chain = "KUBE-PROXY-FIREWALL"
// kube proxy canary chain is used for monitoring rule reload
kubeProxyCanaryChain utiliptables.Chain = "KUBE-PROXY-CANARY"
// largeClusterEndpointsThreshold is the number of endpoints at which
// we switch into "large cluster mode" and optimize for iptables
// performance over iptables debuggability
largeClusterEndpointsThreshold = 1000
)
const sysctlRouteLocalnet = "net/ipv4/conf/all/route_localnet"
const sysctlBridgeCallIPTables = "net/bridge/bridge-nf-call-iptables"
// internal struct for string service information
type servicePortInfo struct {
*proxy.BaseServicePortInfo
// The following fields are computed and stored for performance reasons.
nameString string
clusterPolicyChainName utiliptables.Chain
localPolicyChainName utiliptables.Chain
firewallChainName utiliptables.Chain
externalChainName utiliptables.Chain
}
// returns a new proxy.ServicePort which abstracts a serviceInfo
func newServiceInfo(port *v1.ServicePort, service *v1.Service, bsvcPortInfo *proxy.BaseServicePortInfo) proxy.ServicePort {
svcPort := &servicePortInfo{BaseServicePortInfo: bsvcPortInfo}
// Store the following for performance reasons.
svcName := types.NamespacedName{Namespace: service.Namespace, Name: service.Name}
svcPortName := proxy.ServicePortName{NamespacedName: svcName, Port: port.Name}
protocol := strings.ToLower(string(svcPort.Protocol()))
svcPort.nameString = svcPortName.String()
svcPort.clusterPolicyChainName = servicePortPolicyClusterChain(svcPort.nameString, protocol)
svcPort.localPolicyChainName = servicePortPolicyLocalChainName(svcPort.nameString, protocol)
svcPort.firewallChainName = serviceFirewallChainName(svcPort.nameString, protocol)
svcPort.externalChainName = serviceExternalChainName(svcPort.nameString, protocol)
return svcPort
}
// internal struct for endpoints information
type endpointsInfo struct {
*proxy.BaseEndpointInfo
ChainName utiliptables.Chain
}
// returns a new proxy.Endpoint which abstracts a endpointsInfo
func newEndpointInfo(baseInfo *proxy.BaseEndpointInfo, svcPortName *proxy.ServicePortName) proxy.Endpoint {
return &endpointsInfo{
BaseEndpointInfo: baseInfo,
ChainName: servicePortEndpointChainName(svcPortName.String(), strings.ToLower(string(svcPortName.Protocol)), baseInfo.Endpoint),
}
}
// Equal overrides the Equal() function implemented by proxy.BaseEndpointInfo.
func (e *endpointsInfo) Equal(other proxy.Endpoint) bool {
o, ok := other.(*endpointsInfo)
if !ok {
klog.ErrorS(nil, "Failed to cast endpointsInfo")
return false
}
return e.Endpoint == o.Endpoint &&
e.IsLocal == o.IsLocal &&
e.ChainName == o.ChainName &&
e.Ready == o.Ready
}
// Proxier is an iptables based proxy for connections between a localhost:lport
// and services that provide the actual backends.
type Proxier struct {
// endpointsChanges and serviceChanges contains all changes to endpoints and
// services that happened since iptables was synced. For a single object,
// changes are accumulated, i.e. previous is state from before all of them,
// current is state after applying all of those.
endpointsChanges *proxy.EndpointChangeTracker
serviceChanges *proxy.ServiceChangeTracker
mu sync.Mutex // protects the following fields
svcPortMap proxy.ServicePortMap
endpointsMap proxy.EndpointsMap
nodeLabels map[string]string
// endpointSlicesSynced, and servicesSynced are set to true
// when corresponding objects are synced after startup. This is used to avoid
// updating iptables with some partial data after kube-proxy restart.
endpointSlicesSynced bool
servicesSynced bool
needFullSync bool
initialized int32
syncRunner *async.BoundedFrequencyRunner // governs calls to syncProxyRules
syncPeriod time.Duration
lastIPTablesCleanup time.Time
// These are effectively const and do not need the mutex to be held.
iptables utiliptables.Interface
masqueradeAll bool
masqueradeMark string
exec utilexec.Interface
localDetector proxyutiliptables.LocalTrafficDetector
hostname string
nodeIP net.IP
recorder events.EventRecorder
serviceHealthServer healthcheck.ServiceHealthServer
healthzServer healthcheck.ProxierHealthUpdater
// Since converting probabilities (floats) to strings is expensive
// and we are using only probabilities in the format of 1/n, we are
// precomputing some number of those and cache for future reuse.
precomputedProbabilities []string
// The following buffers are used to reuse memory and avoid allocations
// that are significantly impacting performance.
iptablesData *bytes.Buffer
existingFilterChainsData *bytes.Buffer
filterChains utilproxy.LineBuffer
filterRules utilproxy.LineBuffer
natChains utilproxy.LineBuffer
natRules utilproxy.LineBuffer
// largeClusterMode is set at the beginning of syncProxyRules if we are
// going to end up outputting "lots" of iptables rules and so we need to
// optimize for performance over debuggability.
largeClusterMode bool
// localhostNodePorts indicates whether to generate iptables rules that
// disable NodePort services to be accessed via localhost.
localhostNodePorts bool
// Values are as a parameter to select the interfaces where nodePort works.
nodePortAddresses []string
// networkInterfacer defines an interface for several net library functions.
// Inject for test purpose.
networkInterfacer utilproxy.NetworkInterfacer
}
// Proxier implements proxy.Provider
var _ proxy.Provider = &Proxier{}
// NewProxier returns a new Proxier given an iptables Interface instance.
// Because of the iptables logic, it is assumed that there is only a single Proxier active on a machine.
// An error will be returned if iptables fails to update or acquire the initial lock.
// Once a proxier is created, it will keep iptables up to date in the background and
// will not terminate if a particular iptables call fails.
func NewProxier(ipt utiliptables.Interface,
sysctl utilsysctl.Interface,
exec utilexec.Interface,
syncPeriod time.Duration,
minSyncPeriod time.Duration,
masqueradeAll bool,
localhostNodePorts bool,
masqueradeBit int,
localDetector proxyutiliptables.LocalTrafficDetector,
hostname string,
nodeIP net.IP,
recorder events.EventRecorder,
healthzServer healthcheck.ProxierHealthUpdater,
nodePortAddresses []string,
) (*Proxier, error) {
if localhostNodePorts && utilproxy.ContainsIPv4Loopback(nodePortAddresses) {
// Set the route_localnet sysctl we need for exposing NodePorts on loopback addresses
// Refer to https://issues.k8s.io/90259
klog.InfoS("Setting route_localnet=1 to allow node-ports on localhost; to change this either disable iptables.localhostNodePorts (--iptables-localhost-nodeports) or set nodePortAddresses (--nodeport-addresses) to filter loopback addresses")
if err := utilproxy.EnsureSysctl(sysctl, sysctlRouteLocalnet, 1); err != nil {
return nil, err
}
}
// Proxy needs br_netfilter and bridge-nf-call-iptables=1 when containers
// are connected to a Linux bridge (but not SDN bridges). Until most
// plugins handle this, log when config is missing
if val, err := sysctl.GetSysctl(sysctlBridgeCallIPTables); err == nil && val != 1 {
klog.InfoS("Missing br-netfilter module or unset sysctl br-nf-call-iptables, proxy may not work as intended")
}
// Generate the masquerade mark to use for SNAT rules.
masqueradeValue := 1 << uint(masqueradeBit)
masqueradeMark := fmt.Sprintf("%#08x", masqueradeValue)
klog.V(2).InfoS("Using iptables mark for masquerade", "ipFamily", ipt.Protocol(), "mark", masqueradeMark)
serviceHealthServer := healthcheck.NewServiceHealthServer(hostname, recorder, nodePortAddresses)
ipFamily := v1.IPv4Protocol
if ipt.IsIPv6() {
ipFamily = v1.IPv6Protocol
}
ipFamilyMap := utilproxy.MapCIDRsByIPFamily(nodePortAddresses)
nodePortAddresses = ipFamilyMap[ipFamily]
// Log the IPs not matching the ipFamily
if ips, ok := ipFamilyMap[utilproxy.OtherIPFamily(ipFamily)]; ok && len(ips) > 0 {
klog.InfoS("Found node IPs of the wrong family", "ipFamily", ipFamily, "IPs", strings.Join(ips, ","))
}
proxier := &Proxier{
svcPortMap: make(proxy.ServicePortMap),
serviceChanges: proxy.NewServiceChangeTracker(newServiceInfo, ipFamily, recorder, nil),
endpointsMap: make(proxy.EndpointsMap),
endpointsChanges: proxy.NewEndpointChangeTracker(hostname, newEndpointInfo, ipFamily, recorder, nil),
syncPeriod: syncPeriod,
iptables: ipt,
masqueradeAll: masqueradeAll,
masqueradeMark: masqueradeMark,
exec: exec,
localDetector: localDetector,
hostname: hostname,
nodeIP: nodeIP,
recorder: recorder,
serviceHealthServer: serviceHealthServer,
healthzServer: healthzServer,
precomputedProbabilities: make([]string, 0, 1001),
iptablesData: bytes.NewBuffer(nil),
existingFilterChainsData: bytes.NewBuffer(nil),
filterChains: utilproxy.LineBuffer{},
filterRules: utilproxy.LineBuffer{},
natChains: utilproxy.LineBuffer{},
natRules: utilproxy.LineBuffer{},
localhostNodePorts: localhostNodePorts,
nodePortAddresses: nodePortAddresses,
networkInterfacer: utilproxy.RealNetwork{},
}
burstSyncs := 2
klog.V(2).InfoS("Iptables sync params", "ipFamily", ipt.Protocol(), "minSyncPeriod", minSyncPeriod, "syncPeriod", syncPeriod, "burstSyncs", burstSyncs)
// We pass syncPeriod to ipt.Monitor, which will call us only if it needs to.
// We need to pass *some* maxInterval to NewBoundedFrequencyRunner anyway though.
// time.Hour is arbitrary.
proxier.syncRunner = async.NewBoundedFrequencyRunner("sync-runner", proxier.syncProxyRules, minSyncPeriod, time.Hour, burstSyncs)
go ipt.Monitor(kubeProxyCanaryChain, []utiliptables.Table{utiliptables.TableMangle, utiliptables.TableNAT, utiliptables.TableFilter},
proxier.forceSyncProxyRules, syncPeriod, wait.NeverStop)
if ipt.HasRandomFully() {
klog.V(2).InfoS("Iptables supports --random-fully", "ipFamily", ipt.Protocol())
} else {
klog.V(2).InfoS("Iptables does not support --random-fully", "ipFamily", ipt.Protocol())
}
return proxier, nil
}
// NewDualStackProxier creates a MetaProxier instance, with IPv4 and IPv6 proxies.
func NewDualStackProxier(
ipt [2]utiliptables.Interface,
sysctl utilsysctl.Interface,
exec utilexec.Interface,
syncPeriod time.Duration,
minSyncPeriod time.Duration,
masqueradeAll bool,
localhostNodePorts bool,
masqueradeBit int,
localDetectors [2]proxyutiliptables.LocalTrafficDetector,
hostname string,
nodeIP [2]net.IP,
recorder events.EventRecorder,
healthzServer healthcheck.ProxierHealthUpdater,
nodePortAddresses []string,
) (proxy.Provider, error) {
// Create an ipv4 instance of the single-stack proxier
ipFamilyMap := utilproxy.MapCIDRsByIPFamily(nodePortAddresses)
ipv4Proxier, err := NewProxier(ipt[0], sysctl,
exec, syncPeriod, minSyncPeriod, masqueradeAll, localhostNodePorts, masqueradeBit, localDetectors[0], hostname,
nodeIP[0], recorder, healthzServer, ipFamilyMap[v1.IPv4Protocol])
if err != nil {
return nil, fmt.Errorf("unable to create ipv4 proxier: %v", err)
}
ipv6Proxier, err := NewProxier(ipt[1], sysctl,
exec, syncPeriod, minSyncPeriod, masqueradeAll, false, masqueradeBit, localDetectors[1], hostname,
nodeIP[1], recorder, healthzServer, ipFamilyMap[v1.IPv6Protocol])
if err != nil {
return nil, fmt.Errorf("unable to create ipv6 proxier: %v", err)
}
return metaproxier.NewMetaProxier(ipv4Proxier, ipv6Proxier), nil
}
type iptablesJumpChain struct {
table utiliptables.Table
dstChain utiliptables.Chain
srcChain utiliptables.Chain
comment string
extraArgs []string
}
var iptablesJumpChains = []iptablesJumpChain{
{utiliptables.TableFilter, kubeExternalServicesChain, utiliptables.ChainInput, "kubernetes externally-visible service portals", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableFilter, kubeExternalServicesChain, utiliptables.ChainForward, "kubernetes externally-visible service portals", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableFilter, kubeNodePortsChain, utiliptables.ChainInput, "kubernetes health check service ports", nil},
{utiliptables.TableFilter, kubeServicesChain, utiliptables.ChainForward, "kubernetes service portals", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableFilter, kubeServicesChain, utiliptables.ChainOutput, "kubernetes service portals", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableFilter, kubeForwardChain, utiliptables.ChainForward, "kubernetes forwarding rules", nil},
{utiliptables.TableFilter, kubeProxyFirewallChain, utiliptables.ChainInput, "kubernetes load balancer firewall", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableFilter, kubeProxyFirewallChain, utiliptables.ChainOutput, "kubernetes load balancer firewall", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableFilter, kubeProxyFirewallChain, utiliptables.ChainForward, "kubernetes load balancer firewall", []string{"-m", "conntrack", "--ctstate", "NEW"}},
{utiliptables.TableNAT, kubeServicesChain, utiliptables.ChainOutput, "kubernetes service portals", nil},
{utiliptables.TableNAT, kubeServicesChain, utiliptables.ChainPrerouting, "kubernetes service portals", nil},
{utiliptables.TableNAT, kubePostroutingChain, utiliptables.ChainPostrouting, "kubernetes postrouting rules", nil},
}
var iptablesCleanupOnlyChains = []iptablesJumpChain{
// Present in kube 1.13 - 1.19. Removed by #95252 in favor of adding reject rules for incoming/forwarding packets to kubeExternalServicesChain
{utiliptables.TableFilter, kubeServicesChain, utiliptables.ChainInput, "kubernetes service portals", []string{"-m", "conntrack", "--ctstate", "NEW"}},
}
// CleanupLeftovers removes all iptables rules and chains created by the Proxier
// It returns true if an error was encountered. Errors are logged.
func CleanupLeftovers(ipt utiliptables.Interface) (encounteredError bool) {
// Unlink our chains
for _, jump := range append(iptablesJumpChains, iptablesCleanupOnlyChains...) {
args := append(jump.extraArgs,
"-m", "comment", "--comment", jump.comment,
"-j", string(jump.dstChain),
)
if err := ipt.DeleteRule(jump.table, jump.srcChain, args...); err != nil {
if !utiliptables.IsNotFoundError(err) {
klog.ErrorS(err, "Error removing pure-iptables proxy rule")
encounteredError = true
}
}
}
// Flush and remove all of our "-t nat" chains.
iptablesData := bytes.NewBuffer(nil)
if err := ipt.SaveInto(utiliptables.TableNAT, iptablesData); err != nil {
klog.ErrorS(err, "Failed to execute iptables-save", "table", utiliptables.TableNAT)
encounteredError = true
} else {
existingNATChains := utiliptables.GetChainsFromTable(iptablesData.Bytes())
natChains := &utilproxy.LineBuffer{}
natRules := &utilproxy.LineBuffer{}
natChains.Write("*nat")
// Start with chains we know we need to remove.
for _, chain := range []utiliptables.Chain{kubeServicesChain, kubeNodePortsChain, kubePostroutingChain} {
if _, found := existingNATChains[chain]; found {
chainString := string(chain)
natChains.Write(utiliptables.MakeChainLine(chain)) // flush
natRules.Write("-X", chainString) // delete
}
}
// Hunt for service and endpoint chains.
for chain := range existingNATChains {
chainString := string(chain)
if isServiceChainName(chainString) {
natChains.Write(utiliptables.MakeChainLine(chain)) // flush
natRules.Write("-X", chainString) // delete
}
}
natRules.Write("COMMIT")
natLines := append(natChains.Bytes(), natRules.Bytes()...)
// Write it.
err = ipt.Restore(utiliptables.TableNAT, natLines, utiliptables.NoFlushTables, utiliptables.RestoreCounters)
if err != nil {
klog.ErrorS(err, "Failed to execute iptables-restore", "table", utiliptables.TableNAT)
metrics.IptablesRestoreFailuresTotal.Inc()
encounteredError = true
}
}
// Flush and remove all of our "-t filter" chains.
iptablesData.Reset()
if err := ipt.SaveInto(utiliptables.TableFilter, iptablesData); err != nil {
klog.ErrorS(err, "Failed to execute iptables-save", "table", utiliptables.TableFilter)
encounteredError = true
} else {
existingFilterChains := utiliptables.GetChainsFromTable(iptablesData.Bytes())
filterChains := &utilproxy.LineBuffer{}
filterRules := &utilproxy.LineBuffer{}
filterChains.Write("*filter")
for _, chain := range []utiliptables.Chain{kubeServicesChain, kubeExternalServicesChain, kubeForwardChain, kubeNodePortsChain} {
if _, found := existingFilterChains[chain]; found {
chainString := string(chain)
filterChains.Write(utiliptables.MakeChainLine(chain))
filterRules.Write("-X", chainString)
}
}
filterRules.Write("COMMIT")
filterLines := append(filterChains.Bytes(), filterRules.Bytes()...)
// Write it.
if err := ipt.Restore(utiliptables.TableFilter, filterLines, utiliptables.NoFlushTables, utiliptables.RestoreCounters); err != nil {
klog.ErrorS(err, "Failed to execute iptables-restore", "table", utiliptables.TableFilter)
metrics.IptablesRestoreFailuresTotal.Inc()
encounteredError = true
}
}
return encounteredError
}
func computeProbability(n int) string {
return fmt.Sprintf("%0.10f", 1.0/float64(n))
}
// This assumes proxier.mu is held
func (proxier *Proxier) precomputeProbabilities(numberOfPrecomputed int) {
if len(proxier.precomputedProbabilities) == 0 {
proxier.precomputedProbabilities = append(proxier.precomputedProbabilities, "<bad value>")
}
for i := len(proxier.precomputedProbabilities); i <= numberOfPrecomputed; i++ {
proxier.precomputedProbabilities = append(proxier.precomputedProbabilities, computeProbability(i))
}
}
// This assumes proxier.mu is held
func (proxier *Proxier) probability(n int) string {
if n >= len(proxier.precomputedProbabilities) {
proxier.precomputeProbabilities(n)
}
return proxier.precomputedProbabilities[n]
}
// Sync is called to synchronize the proxier state to iptables as soon as possible.
func (proxier *Proxier) Sync() {
if proxier.healthzServer != nil {
proxier.healthzServer.QueuedUpdate()
}
metrics.SyncProxyRulesLastQueuedTimestamp.SetToCurrentTime()
proxier.syncRunner.Run()
}
// SyncLoop runs periodic work. This is expected to run as a goroutine or as the main loop of the app. It does not return.
func (proxier *Proxier) SyncLoop() {
// Update healthz timestamp at beginning in case Sync() never succeeds.
if proxier.healthzServer != nil {
proxier.healthzServer.Updated()
}
// synthesize "last change queued" time as the informers are syncing.
metrics.SyncProxyRulesLastQueuedTimestamp.SetToCurrentTime()
proxier.syncRunner.Loop(wait.NeverStop)
}
func (proxier *Proxier) setInitialized(value bool) {
var initialized int32
if value {
initialized = 1
}
atomic.StoreInt32(&proxier.initialized, initialized)
}
func (proxier *Proxier) isInitialized() bool {
return atomic.LoadInt32(&proxier.initialized) > 0
}
// OnServiceAdd is called whenever creation of new service object
// is observed.
func (proxier *Proxier) OnServiceAdd(service *v1.Service) {
proxier.OnServiceUpdate(nil, service)
}
// OnServiceUpdate is called whenever modification of an existing
// service object is observed.
func (proxier *Proxier) OnServiceUpdate(oldService, service *v1.Service) {
if proxier.serviceChanges.Update(oldService, service) && proxier.isInitialized() {
proxier.Sync()
}
}
// OnServiceDelete is called whenever deletion of an existing service
// object is observed.
func (proxier *Proxier) OnServiceDelete(service *v1.Service) {
proxier.OnServiceUpdate(service, nil)
}
// OnServiceSynced is called once all the initial event handlers were
// called and the state is fully propagated to local cache.
func (proxier *Proxier) OnServiceSynced() {
proxier.mu.Lock()
proxier.servicesSynced = true
proxier.setInitialized(proxier.endpointSlicesSynced)
proxier.mu.Unlock()
// Sync unconditionally - this is called once per lifetime.
proxier.forceSyncProxyRules()
}
// OnEndpointSliceAdd is called whenever creation of a new endpoint slice object
// is observed.
func (proxier *Proxier) OnEndpointSliceAdd(endpointSlice *discovery.EndpointSlice) {
if proxier.endpointsChanges.EndpointSliceUpdate(endpointSlice, false) && proxier.isInitialized() {
proxier.Sync()
}
}
// OnEndpointSliceUpdate is called whenever modification of an existing endpoint
// slice object is observed.
func (proxier *Proxier) OnEndpointSliceUpdate(_, endpointSlice *discovery.EndpointSlice) {
if proxier.endpointsChanges.EndpointSliceUpdate(endpointSlice, false) && proxier.isInitialized() {
proxier.Sync()
}
}
// OnEndpointSliceDelete is called whenever deletion of an existing endpoint slice
// object is observed.
func (proxier *Proxier) OnEndpointSliceDelete(endpointSlice *discovery.EndpointSlice) {
if proxier.endpointsChanges.EndpointSliceUpdate(endpointSlice, true) && proxier.isInitialized() {
proxier.Sync()
}
}
// OnEndpointSlicesSynced is called once all the initial event handlers were
// called and the state is fully propagated to local cache.
func (proxier *Proxier) OnEndpointSlicesSynced() {
proxier.mu.Lock()
proxier.endpointSlicesSynced = true
proxier.setInitialized(proxier.servicesSynced)
proxier.mu.Unlock()
// Sync unconditionally - this is called once per lifetime.
proxier.forceSyncProxyRules()
}
// OnNodeAdd is called whenever creation of new node object
// is observed.
func (proxier *Proxier) OnNodeAdd(node *v1.Node) {
if node.Name != proxier.hostname {
klog.ErrorS(nil, "Received a watch event for a node that doesn't match the current node",
"eventNode", node.Name, "currentNode", proxier.hostname)
return
}
if reflect.DeepEqual(proxier.nodeLabels, node.Labels) {
return
}
proxier.mu.Lock()
proxier.nodeLabels = map[string]string{}
for k, v := range node.Labels {
proxier.nodeLabels[k] = v
}
proxier.needFullSync = true
proxier.mu.Unlock()
klog.V(4).InfoS("Updated proxier node labels", "labels", node.Labels)
proxier.Sync()
}
// OnNodeUpdate is called whenever modification of an existing
// node object is observed.
func (proxier *Proxier) OnNodeUpdate(oldNode, node *v1.Node) {
if node.Name != proxier.hostname {
klog.ErrorS(nil, "Received a watch event for a node that doesn't match the current node",
"eventNode", node.Name, "currentNode", proxier.hostname)
return
}
if reflect.DeepEqual(proxier.nodeLabels, node.Labels) {
return
}
proxier.mu.Lock()
proxier.nodeLabels = map[string]string{}
for k, v := range node.Labels {
proxier.nodeLabels[k] = v
}
proxier.needFullSync = true
proxier.mu.Unlock()
klog.V(4).InfoS("Updated proxier node labels", "labels", node.Labels)
proxier.Sync()
}
// OnNodeDelete is called whenever deletion of an existing node
// object is observed.
func (proxier *Proxier) OnNodeDelete(node *v1.Node) {
if node.Name != proxier.hostname {
klog.ErrorS(nil, "Received a watch event for a node that doesn't match the current node",
"eventNode", node.Name, "currentNode", proxier.hostname)
return
}
proxier.mu.Lock()
proxier.nodeLabels = nil
proxier.needFullSync = true
proxier.mu.Unlock()
proxier.Sync()
}
// OnNodeSynced is called once all the initial event handlers were
// called and the state is fully propagated to local cache.
func (proxier *Proxier) OnNodeSynced() {
}
// portProtoHash takes the ServicePortName and protocol for a service
// returns the associated 16 character hash. This is computed by hashing (sha256)
// then encoding to base32 and truncating to 16 chars. We do this because IPTables
// Chain Names must be <= 28 chars long, and the longer they are the harder they are to read.
func portProtoHash(servicePortName string, protocol string) string {
hash := sha256.Sum256([]byte(servicePortName + protocol))
encoded := base32.StdEncoding.EncodeToString(hash[:])
return encoded[:16]
}
const (
servicePortPolicyClusterChainNamePrefix = "KUBE-SVC-"
servicePortPolicyLocalChainNamePrefix = "KUBE-SVL-"
serviceFirewallChainNamePrefix = "KUBE-FW-"
serviceExternalChainNamePrefix = "KUBE-EXT-"
servicePortEndpointChainNamePrefix = "KUBE-SEP-"
// For cleanup. This can be removed after 1.26 is released.
deprecatedServiceLBChainNamePrefix = "KUBE-XLB-"
)
// servicePortPolicyClusterChain returns the name of the KUBE-SVC-XXXX chain for a service, which is the
// main iptables chain for that service, used for dispatching to endpoints when using `Cluster`
// traffic policy.
func servicePortPolicyClusterChain(servicePortName string, protocol string) utiliptables.Chain {
return utiliptables.Chain(servicePortPolicyClusterChainNamePrefix + portProtoHash(servicePortName, protocol))
}
// servicePortPolicyLocalChainName returns the name of the KUBE-SVL-XXXX chain for a service, which
// handles dispatching to local endpoints when using `Local` traffic policy. This chain only
// exists if the service has `Local` internal or external traffic policy.
func servicePortPolicyLocalChainName(servicePortName string, protocol string) utiliptables.Chain {
return utiliptables.Chain(servicePortPolicyLocalChainNamePrefix + portProtoHash(servicePortName, protocol))
}
// serviceFirewallChainName returns the name of the KUBE-FW-XXXX chain for a service, which
// is used to implement the filtering for the LoadBalancerSourceRanges feature.
func serviceFirewallChainName(servicePortName string, protocol string) utiliptables.Chain {
return utiliptables.Chain(serviceFirewallChainNamePrefix + portProtoHash(servicePortName, protocol))
}
// serviceExternalChainName returns the name of the KUBE-EXT-XXXX chain for a service, which
// implements "short-circuiting" for internally-originated external-destination traffic when using
// `Local` external traffic policy. It forwards traffic from local sources to the KUBE-SVC-XXXX
// chain and traffic from external sources to the KUBE-SVL-XXXX chain.
func serviceExternalChainName(servicePortName string, protocol string) utiliptables.Chain {
return utiliptables.Chain(serviceExternalChainNamePrefix + portProtoHash(servicePortName, protocol))
}
// servicePortEndpointChainName returns the name of the KUBE-SEP-XXXX chain for a particular
// service endpoint.
func servicePortEndpointChainName(servicePortName string, protocol string, endpoint string) utiliptables.Chain {
hash := sha256.Sum256([]byte(servicePortName + protocol + endpoint))
encoded := base32.StdEncoding.EncodeToString(hash[:])
return utiliptables.Chain(servicePortEndpointChainNamePrefix + encoded[:16])
}
func isServiceChainName(chainString string) bool {
prefixes := []string{
servicePortPolicyClusterChainNamePrefix,
servicePortPolicyLocalChainNamePrefix,
servicePortEndpointChainNamePrefix,
serviceFirewallChainNamePrefix,
serviceExternalChainNamePrefix,
deprecatedServiceLBChainNamePrefix,
}
for _, p := range prefixes {
if strings.HasPrefix(chainString, p) {
return true
}
}
return false
}
// After a UDP or SCTP endpoint has been removed, we must flush any pending conntrack entries to it, or else we
// risk sending more traffic to it, all of which will be lost.
// This assumes the proxier mutex is held
// TODO: move it to util
func (proxier *Proxier) deleteEndpointConnections(connectionMap []proxy.ServiceEndpoint) {
for _, epSvcPair := range connectionMap {
if svcInfo, ok := proxier.svcPortMap[epSvcPair.ServicePortName]; ok && conntrack.IsClearConntrackNeeded(svcInfo.Protocol()) {
endpointIP := utilproxy.IPPart(epSvcPair.Endpoint)
nodePort := svcInfo.NodePort()
svcProto := svcInfo.Protocol()
var err error
if nodePort != 0 {
err = conntrack.ClearEntriesForPortNAT(proxier.exec, endpointIP, nodePort, svcProto)
if err != nil {
klog.ErrorS(err, "Failed to delete nodeport-related endpoint connections", "servicePortName", epSvcPair.ServicePortName)
}
}
err = conntrack.ClearEntriesForNAT(proxier.exec, svcInfo.ClusterIP().String(), endpointIP, svcProto)
if err != nil {
klog.ErrorS(err, "Failed to delete endpoint connections", "servicePortName", epSvcPair.ServicePortName)
}
for _, extIP := range svcInfo.ExternalIPStrings() {
err := conntrack.ClearEntriesForNAT(proxier.exec, extIP, endpointIP, svcProto)
if err != nil {
klog.ErrorS(err, "Failed to delete endpoint connections for externalIP", "servicePortName", epSvcPair.ServicePortName, "externalIP", extIP)
}
}
for _, lbIP := range svcInfo.LoadBalancerIPStrings() {
err := conntrack.ClearEntriesForNAT(proxier.exec, lbIP, endpointIP, svcProto)
if err != nil {
klog.ErrorS(err, "Failed to delete endpoint connections for LoadBalancerIP", "servicePortName", epSvcPair.ServicePortName, "loadBalancerIP", lbIP)
}
}
}
}
}
// Assumes proxier.mu is held.
func (proxier *Proxier) appendServiceCommentLocked(args []string, svcName string) []string {
// Not printing these comments, can reduce size of iptables (in case of large
// number of endpoints) even by 40%+. So if total number of endpoint chains
// is large enough, we simply drop those comments.
if proxier.largeClusterMode {
return args
}
return append(args, "-m", "comment", "--comment", svcName)
}
// Called by the iptables.Monitor, and in response to topology changes; this calls
// syncProxyRules() and tells it to resync all services, regardless of whether the
// Service or Endpoints/EndpointSlice objects themselves have changed
func (proxier *Proxier) forceSyncProxyRules() {
proxier.mu.Lock()
proxier.needFullSync = true
proxier.mu.Unlock()
proxier.syncProxyRules()
}
// This is where all of the iptables-save/restore calls happen.
// The only other iptables rules are those that are setup in iptablesInit()
// This assumes proxier.mu is NOT held
func (proxier *Proxier) syncProxyRules() {
proxier.mu.Lock()
defer proxier.mu.Unlock()
// don't sync rules till we've received services and endpoints
if !proxier.isInitialized() {
klog.V(2).InfoS("Not syncing iptables until Services and Endpoints have been received from master")
return
}
// Keep track of how long syncs take.
start := time.Now()
defer func() {
metrics.SyncProxyRulesLatency.Observe(metrics.SinceInSeconds(start))
klog.V(2).InfoS("SyncProxyRules complete", "elapsed", time.Since(start))
}()
tryPartialSync := !proxier.needFullSync && utilfeature.DefaultFeatureGate.Enabled(features.MinimizeIPTablesRestore)
var serviceChanged, endpointsChanged sets.String
if tryPartialSync {
serviceChanged = proxier.serviceChanges.PendingChanges()
endpointsChanged = proxier.endpointsChanges.PendingChanges()
}
serviceUpdateResult := proxier.svcPortMap.Update(proxier.serviceChanges)
endpointUpdateResult := proxier.endpointsMap.Update(proxier.endpointsChanges)
// We need to detect stale connections to UDP Services so we
// can clean dangling conntrack entries that can blackhole traffic.
conntrackCleanupServiceIPs := serviceUpdateResult.UDPStaleClusterIP
conntrackCleanupServiceNodePorts := sets.NewInt()
// merge stale services gathered from updateEndpointsMap
// an UDP service that changes from 0 to non-0 endpoints is considered stale.
for _, svcPortName := range endpointUpdateResult.StaleServiceNames {
if svcInfo, ok := proxier.svcPortMap[svcPortName]; ok && svcInfo != nil && conntrack.IsClearConntrackNeeded(svcInfo.Protocol()) {
klog.V(2).InfoS("Stale service", "protocol", strings.ToLower(string(svcInfo.Protocol())), "servicePortName", svcPortName, "clusterIP", svcInfo.ClusterIP())
conntrackCleanupServiceIPs.Insert(svcInfo.ClusterIP().String())
for _, extIP := range svcInfo.ExternalIPStrings() {
conntrackCleanupServiceIPs.Insert(extIP)
}
for _, lbIP := range svcInfo.LoadBalancerIPStrings() {
conntrackCleanupServiceIPs.Insert(lbIP)
}
nodePort := svcInfo.NodePort()
if svcInfo.Protocol() == v1.ProtocolUDP && nodePort != 0 {
klog.V(2).InfoS("Stale service", "protocol", strings.ToLower(string(svcInfo.Protocol())), "servicePortName", svcPortName, "nodePort", nodePort)
conntrackCleanupServiceNodePorts.Insert(nodePort)
}
}
}
klog.V(2).InfoS("Syncing iptables rules")
success := false
defer func() {
if !success {
klog.InfoS("Sync failed", "retryingTime", proxier.syncPeriod)
proxier.syncRunner.RetryAfter(proxier.syncPeriod)
if tryPartialSync {
metrics.IptablesPartialRestoreFailuresTotal.Inc()
}
// proxier.serviceChanges and proxier.endpointChanges have already
// been flushed, so we've lost the state needed to be able to do
// a partial sync.
proxier.needFullSync = true
}
}()
// Create and link the kube chains.
for _, jump := range iptablesJumpChains {
if _, err := proxier.iptables.EnsureChain(jump.table, jump.dstChain); err != nil {
klog.ErrorS(err, "Failed to ensure chain exists", "table", jump.table, "chain", jump.dstChain)
return
}
args := append(jump.extraArgs,
"-m", "comment", "--comment", jump.comment,
"-j", string(jump.dstChain),
)
if _, err := proxier.iptables.EnsureRule(utiliptables.Prepend, jump.table, jump.srcChain, args...); err != nil {
klog.ErrorS(err, "Failed to ensure chain jumps", "table", jump.table, "srcChain", jump.srcChain, "dstChain", jump.dstChain)
return
}
}
//
// Below this point we will not return until we try to write the iptables rules.
//
// Reset all buffers used later.
// This is to avoid memory reallocations and thus improve performance.
proxier.filterChains.Reset()
proxier.filterRules.Reset()
proxier.natChains.Reset()
proxier.natRules.Reset()
// Write chain lines for all the "top-level" chains we'll be filling in
for _, chainName := range []utiliptables.Chain{kubeServicesChain, kubeExternalServicesChain, kubeForwardChain, kubeNodePortsChain, kubeProxyFirewallChain} {
proxier.filterChains.Write(utiliptables.MakeChainLine(chainName))
}
for _, chainName := range []utiliptables.Chain{kubeServicesChain, kubeNodePortsChain, kubePostroutingChain, kubeMarkMasqChain} {
proxier.natChains.Write(utiliptables.MakeChainLine(chainName))
}
// Install the kubernetes-specific postrouting rules. We use a whole chain for
// this so that it is easier to flush and change, for example if the mark
// value should ever change.
// NB: THIS MUST MATCH the corresponding code in the kubelet
proxier.natRules.Write(
"-A", string(kubePostroutingChain),
"-m", "mark", "!", "--mark", fmt.Sprintf("%s/%s", proxier.masqueradeMark, proxier.masqueradeMark),
"-j", "RETURN",
)
// Clear the mark to avoid re-masquerading if the packet re-traverses the network stack.
proxier.natRules.Write(
"-A", string(kubePostroutingChain),
"-j", "MARK", "--xor-mark", proxier.masqueradeMark,
)
masqRule := []string{
"-A", string(kubePostroutingChain),
"-m", "comment", "--comment", `"kubernetes service traffic requiring SNAT"`,
"-j", "MASQUERADE",
}
if proxier.iptables.HasRandomFully() {
masqRule = append(masqRule, "--random-fully")
}
proxier.natRules.Write(masqRule)
// Install the kubernetes-specific masquerade mark rule. We use a whole chain for
// this so that it is easier to flush and change, for example if the mark
// value should ever change.
proxier.natRules.Write(
"-A", string(kubeMarkMasqChain),
"-j", "MARK", "--or-mark", proxier.masqueradeMark,
)
// Accumulate NAT chains to keep.
activeNATChains := map[utiliptables.Chain]bool{} // use a map as a set
// To avoid growing this slice, we arbitrarily set its size to 64,
// there is never more than that many arguments for a single line.
// Note that even if we go over 64, it will still be correct - it
// is just for efficiency, not correctness.
args := make([]string, 64)
// Compute total number of endpoint chains across all services to get
// a sense of how big the cluster is.
totalEndpoints := 0
for svcName := range proxier.svcPortMap {
totalEndpoints += len(proxier.endpointsMap[svcName])
}
proxier.largeClusterMode = (totalEndpoints > largeClusterEndpointsThreshold)
nodeAddresses, err := utilproxy.GetNodeAddresses(proxier.nodePortAddresses, proxier.networkInterfacer)
if err != nil {
klog.ErrorS(err, "Failed to get node ip address matching nodeport cidrs, services with nodeport may not work as intended", "CIDRs", proxier.nodePortAddresses)
}
// nodeAddresses may contain dual-stack zero-CIDRs if proxier.nodePortAddresses is empty.
// Ensure nodeAddresses only contains the addresses for this proxier's IP family.
isIPv6 := proxier.iptables.IsIPv6()
for addr := range nodeAddresses {
if utilproxy.IsZeroCIDR(addr) && isIPv6 == netutils.IsIPv6CIDRString(addr) {
// if any of the addresses is zero cidr of this IP family, non-zero IPs can be excluded.
nodeAddresses = sets.NewString(addr)
break
}
}
// These two variables are used to publish the sync_proxy_rules_no_endpoints_total
// metric.
serviceNoLocalEndpointsTotalInternal := 0
serviceNoLocalEndpointsTotalExternal := 0
// Build rules for each service-port.
for svcName, svc := range proxier.svcPortMap {
svcInfo, ok := svc.(*servicePortInfo)
if !ok {
klog.ErrorS(nil, "Failed to cast serviceInfo", "serviceName", svcName)
continue
}
protocol := strings.ToLower(string(svcInfo.Protocol()))
svcPortNameString := svcInfo.nameString
// Figure out the endpoints for Cluster and Local traffic policy.
// allLocallyReachableEndpoints is the set of all endpoints that can be routed to
// from this node, given the service's traffic policies. hasEndpoints is true
// if the service has any usable endpoints on any node, not just this one.
allEndpoints := proxier.endpointsMap[svcName]
clusterEndpoints, localEndpoints, allLocallyReachableEndpoints, hasEndpoints := proxy.CategorizeEndpoints(allEndpoints, svcInfo, proxier.nodeLabels)
// Note the endpoint chains that will be used
for _, ep := range allLocallyReachableEndpoints {
if epInfo, ok := ep.(*endpointsInfo); ok {
activeNATChains[epInfo.ChainName] = true
}
}
// clusterPolicyChain contains the endpoints used with "Cluster" traffic policy
clusterPolicyChain := svcInfo.clusterPolicyChainName
usesClusterPolicyChain := len(clusterEndpoints) > 0 && svcInfo.UsesClusterEndpoints()
if usesClusterPolicyChain {
activeNATChains[clusterPolicyChain] = true
}