forked from kubernetes/kubernetes
-
Notifications
You must be signed in to change notification settings - Fork 0
/
proxier.go
1850 lines (1683 loc) · 68.1 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 2017 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 ipvs
//
// NOTE: this needs to be tested in e2e since it uses ipvs for everything.
//
import (
"bytes"
"fmt"
"net"
"reflect"
"strconv"
"strings"
"sync"
"sync/atomic"
"time"
"github.com/golang/glog"
clientv1 "k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/types"
utilruntime "k8s.io/apimachinery/pkg/util/runtime"
"k8s.io/apimachinery/pkg/util/sets"
"k8s.io/apimachinery/pkg/util/wait"
utilfeature "k8s.io/apiserver/pkg/util/feature"
"k8s.io/client-go/tools/record"
apiservice "k8s.io/kubernetes/pkg/api/service"
api "k8s.io/kubernetes/pkg/apis/core"
"k8s.io/kubernetes/pkg/apis/core/helper"
"k8s.io/kubernetes/pkg/features"
"k8s.io/kubernetes/pkg/proxy"
"k8s.io/kubernetes/pkg/proxy/healthcheck"
"k8s.io/kubernetes/pkg/proxy/metrics"
utilproxy "k8s.io/kubernetes/pkg/proxy/util"
"k8s.io/kubernetes/pkg/util/async"
utilipset "k8s.io/kubernetes/pkg/util/ipset"
utiliptables "k8s.io/kubernetes/pkg/util/iptables"
utilipvs "k8s.io/kubernetes/pkg/util/ipvs"
utilsysctl "k8s.io/kubernetes/pkg/util/sysctl"
utilexec "k8s.io/utils/exec"
)
const (
// kubeServicesChain is the services portal chain
kubeServicesChain utiliptables.Chain = "KUBE-SERVICES"
// KubeServiceIPSetsChain is the services access IP chain
KubeServiceIPSetsChain utiliptables.Chain = "KUBE-SVC-IPSETS"
// KubeFireWallChain is the kubernetes firewall chain.
KubeFireWallChain utiliptables.Chain = "KUBE-FIRE-WALL"
// kubePostroutingChain is the kubernetes postrouting chain
kubePostroutingChain utiliptables.Chain = "KUBE-POSTROUTING"
// KubeMarkMasqChain is the mark-for-masquerade chain
KubeMarkMasqChain utiliptables.Chain = "KUBE-MARK-MASQ"
// KubeMarkDropChain is the mark-for-drop chain
KubeMarkDropChain utiliptables.Chain = "KUBE-MARK-DROP"
// DefaultScheduler is the default ipvs scheduler algorithm - round robin.
DefaultScheduler = "rr"
// DefaultDummyDevice is the default dummy interface where ipvs service address will bind to it.
DefaultDummyDevice = "kube-ipvs0"
)
var ipvsModules = []string{
"ip_vs",
"ip_vs_rr",
"ip_vs_wrr",
"ip_vs_sh",
"nf_conntrack_ipv4",
}
// In IPVS proxy mode, the following flags need to be setted
const sysctlRouteLocalnet = "net/ipv4/conf/all/route_localnet"
const sysctlBridgeCallIPTables = "net/bridge/bridge-nf-call-iptables"
const sysctlVSConnTrack = "net/ipv4/vs/conntrack"
const sysctlForward = "net/ipv4/ip_forward"
// Proxier is an ipvs 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 last syncProxyRules call. 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 endpointsChangeMap
serviceChanges serviceChangeMap
mu sync.Mutex // protects the following fields
serviceMap proxyServiceMap
endpointsMap proxyEndpointsMap
portsMap map[utilproxy.LocalPort]utilproxy.Closeable
// endpointsSynced and servicesSynced are set to true when corresponding
// objects are synced after startup. This is used to avoid updating ipvs rules
// with some partial data after kube-proxy restart.
endpointsSynced bool
servicesSynced bool
initialized int32
syncRunner *async.BoundedFrequencyRunner // governs calls to syncProxyRules
// These are effectively const and do not need the mutex to be held.
syncPeriod time.Duration
minSyncPeriod time.Duration
iptables utiliptables.Interface
ipvs utilipvs.Interface
ipset utilipset.Interface
exec utilexec.Interface
masqueradeAll bool
masqueradeMark string
clusterCIDR string
hostname string
nodeIP net.IP
portMapper utilproxy.PortOpener
recorder record.EventRecorder
healthChecker healthcheck.Server
healthzServer healthcheck.HealthzUpdater
ipvsScheduler string
// Added as a member to the struct to allow injection for testing.
ipGetter IPGetter
// The following buffers are used to reuse memory and avoid allocations
// that are significantly impacting performance.
iptablesData *bytes.Buffer
natChains *bytes.Buffer
natRules *bytes.Buffer
// Added as a member to the struct to allow injection for testing.
netlinkHandle NetLinkHandle
// loopbackSet is the ipset where stores all endpoints IP:Port,IP for solving hairpin mode purpose.
loopbackSet *IPSet
// clusterIPSet is the ipset where stores all service ClusterIP:Port
clusterIPSet *IPSet
// nodePortSetTCP is the bitmap:port type ipset where stores all TCP node port
nodePortSetTCP *IPSet
// nodePortSetTCP is the bitmap:port type ipset where stores all UDP node port
nodePortSetUDP *IPSet
// externalIPSet is the hash:ip,port type ipset where stores all service ExternalIP:Port
externalIPSet *IPSet
// lbIngressSet is the hash:ip,port type ipset where stores all service load balancer ingress IP:Port.
lbIngressSet *IPSet
// lbMasqSet is the hash:ip,port type ipset where stores all service load balancer ingress IP:Port which needs masquerade.
lbMasqSet *IPSet
// lbWhiteListIPSet is the hash:ip,port,ip type ipset where stores all service load balancer ingress IP:Port,sourceIP pair, any packets
// with the source IP visit ingress IP:Port can pass through.
lbWhiteListIPSet *IPSet
// lbWhiteListIPSet is the hash:ip,port,net type ipset where stores all service load balancer ingress IP:Port,sourceCIDR pair, any packets
// from the source CIDR visit ingress IP:Port can pass through.
lbWhiteListCIDRSet *IPSet
}
// IPGetter helps get node network interface IP
type IPGetter interface {
NodeIPs() ([]net.IP, error)
}
type realIPGetter struct{}
func (r *realIPGetter) NodeIPs() (ips []net.IP, err error) {
interfaces, err := net.Interfaces()
if err != nil {
return nil, err
}
for i := range interfaces {
name := interfaces[i].Name
// We assume node ip bind to eth{x}
if !strings.HasPrefix(name, "eth") {
continue
}
intf, err := net.InterfaceByName(name)
if err != nil {
utilruntime.HandleError(fmt.Errorf("Failed to get interface by name: %s, error: %v", name, err))
continue
}
addrs, err := intf.Addrs()
if err != nil {
utilruntime.HandleError(fmt.Errorf("Failed to get addresses from interface: %s, error: %v", name, err))
continue
}
for _, a := range addrs {
if ipnet, ok := a.(*net.IPNet); ok {
ips = append(ips, ipnet.IP)
}
}
}
return
}
// Proxier implements ProxyProvider
var _ proxy.ProxyProvider = &Proxier{}
// NewProxier returns a new Proxier given an iptables and ipvs Interface instance.
// Because of the iptables and ipvs logic, it is assumed that there is only a single Proxier active on a machine.
// An error will be returned if it fails to update or acquire the initial lock.
// Once a proxier is created, it will keep iptables and ipvs rules up to date in the background and
// will not terminate if a particular iptables or ipvs call fails.
func NewProxier(ipt utiliptables.Interface,
ipvs utilipvs.Interface,
ipset utilipset.Interface,
sysctl utilsysctl.Interface,
exec utilexec.Interface,
syncPeriod time.Duration,
minSyncPeriod time.Duration,
masqueradeAll bool,
masqueradeBit int,
clusterCIDR string,
hostname string,
nodeIP net.IP,
recorder record.EventRecorder,
healthzServer healthcheck.HealthzUpdater,
scheduler string,
) (*Proxier, error) {
// Set the route_localnet sysctl we need for
if err := sysctl.SetSysctl(sysctlRouteLocalnet, 1); err != nil {
return nil, fmt.Errorf("can't set sysctl %s: %v", sysctlRouteLocalnet, 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 {
glog.Infof("missing br-netfilter module or unset sysctl br-nf-call-iptables; proxy may not work as intended")
}
// Set the conntrack sysctl we need for
if err := sysctl.SetSysctl(sysctlVSConnTrack, 1); err != nil {
return nil, fmt.Errorf("can't set sysctl %s: %v", sysctlVSConnTrack, err)
}
// Set the ip_forward sysctl we need for
if err := sysctl.SetSysctl(sysctlForward, 1); err != nil {
return nil, fmt.Errorf("can't set sysctl %s: %v", sysctlForward, err)
}
// Generate the masquerade mark to use for SNAT rules.
masqueradeValue := 1 << uint(masqueradeBit)
masqueradeMark := fmt.Sprintf("%#08x/%#08x", masqueradeValue, masqueradeValue)
if nodeIP == nil {
glog.Warningf("invalid nodeIP, initializing kube-proxy with 127.0.0.1 as nodeIP")
nodeIP = net.ParseIP("127.0.0.1")
}
if len(clusterCIDR) == 0 {
glog.Warningf("clusterCIDR not specified, unable to distinguish between internal and external traffic")
}
if len(scheduler) == 0 {
glog.Warningf("IPVS scheduler not specified, use %s by default", DefaultScheduler)
scheduler = DefaultScheduler
}
healthChecker := healthcheck.NewServer(hostname, recorder, nil, nil) // use default implementations of deps
isIPv6 := utilproxy.IsIPv6(nodeIP)
glog.V(2).Infof("nodeIP: %v, isIPv6: %v", nodeIP, isIPv6)
proxier := &Proxier{
portsMap: make(map[utilproxy.LocalPort]utilproxy.Closeable),
serviceMap: make(proxyServiceMap),
serviceChanges: newServiceChangeMap(),
endpointsMap: make(proxyEndpointsMap),
endpointsChanges: newEndpointsChangeMap(hostname),
syncPeriod: syncPeriod,
minSyncPeriod: minSyncPeriod,
iptables: ipt,
masqueradeAll: masqueradeAll,
masqueradeMark: masqueradeMark,
exec: exec,
clusterCIDR: clusterCIDR,
hostname: hostname,
nodeIP: nodeIP,
portMapper: &listenPortOpener{},
recorder: recorder,
healthChecker: healthChecker,
healthzServer: healthzServer,
ipvs: ipvs,
ipvsScheduler: scheduler,
ipGetter: &realIPGetter{},
iptablesData: bytes.NewBuffer(nil),
natChains: bytes.NewBuffer(nil),
natRules: bytes.NewBuffer(nil),
netlinkHandle: NewNetLinkHandle(),
ipset: ipset,
loopbackSet: NewIPSet(ipset, KubeLoopBackIPSet, utilipset.HashIPPortIP, isIPv6),
clusterIPSet: NewIPSet(ipset, KubeClusterIPSet, utilipset.HashIPPort, isIPv6),
externalIPSet: NewIPSet(ipset, KubeExternalIPSet, utilipset.HashIPPort, isIPv6),
lbIngressSet: NewIPSet(ipset, KubeLoadBalancerSet, utilipset.HashIPPort, isIPv6),
lbMasqSet: NewIPSet(ipset, KubeLoadBalancerMasqSet, utilipset.HashIPPort, isIPv6),
lbWhiteListIPSet: NewIPSet(ipset, KubeLoadBalancerSourceIPSet, utilipset.HashIPPortIP, isIPv6),
lbWhiteListCIDRSet: NewIPSet(ipset, KubeLoadBalancerSourceCIDRSet, utilipset.HashIPPortNet, isIPv6),
nodePortSetTCP: NewIPSet(ipset, KubeNodePortSetTCP, utilipset.BitmapPort, false),
nodePortSetUDP: NewIPSet(ipset, KubeNodePortSetUDP, utilipset.BitmapPort, false),
}
burstSyncs := 2
glog.V(3).Infof("minSyncPeriod: %v, syncPeriod: %v, burstSyncs: %d", minSyncPeriod, syncPeriod, burstSyncs)
proxier.syncRunner = async.NewBoundedFrequencyRunner("sync-runner", proxier.syncProxyRules, minSyncPeriod, syncPeriod, burstSyncs)
return proxier, nil
}
type proxyServiceMap map[proxy.ServicePortName]*serviceInfo
// internal struct for string service information
type serviceInfo struct {
clusterIP net.IP
port int
protocol api.Protocol
nodePort int
loadBalancerStatus api.LoadBalancerStatus
sessionAffinityType api.ServiceAffinity
stickyMaxAgeSeconds int
externalIPs []string
loadBalancerSourceRanges []string
onlyNodeLocalEndpoints bool
healthCheckNodePort int
// The following fields are computed and stored for performance reasons.
serviceNameString string
}
// <serviceMap> is updated by this function (based on the given changes).
// <changes> map is cleared after applying them.
func updateServiceMap(
serviceMap proxyServiceMap,
changes *serviceChangeMap) (result updateServiceMapResult) {
result.staleServices = sets.NewString()
func() {
changes.lock.Lock()
defer changes.lock.Unlock()
for _, change := range changes.items {
existingPorts := serviceMap.merge(change.current)
serviceMap.unmerge(change.previous, existingPorts, result.staleServices)
}
changes.items = make(map[types.NamespacedName]*serviceChange)
}()
// TODO: If this will appear to be computationally expensive, consider
// computing this incrementally similarly to serviceMap.
result.hcServices = make(map[types.NamespacedName]uint16)
for svcPortName, info := range serviceMap {
if info.healthCheckNodePort != 0 {
result.hcServices[svcPortName.NamespacedName] = uint16(info.healthCheckNodePort)
}
}
return result
}
// returns a new serviceInfo struct
func newServiceInfo(svcPortName proxy.ServicePortName, port *api.ServicePort, service *api.Service) *serviceInfo {
onlyNodeLocalEndpoints := false
if utilfeature.DefaultFeatureGate.Enabled(features.ExternalTrafficLocalOnly) &&
apiservice.RequestsOnlyLocalTraffic(service) {
onlyNodeLocalEndpoints = true
}
var stickyMaxAgeSeconds int
if service.Spec.SessionAffinity == api.ServiceAffinityClientIP {
stickyMaxAgeSeconds = int(*service.Spec.SessionAffinityConfig.ClientIP.TimeoutSeconds)
}
info := &serviceInfo{
clusterIP: net.ParseIP(service.Spec.ClusterIP),
port: int(port.Port),
protocol: port.Protocol,
nodePort: int(port.NodePort),
// Deep-copy in case the service instance changes
loadBalancerStatus: *helper.LoadBalancerStatusDeepCopy(&service.Status.LoadBalancer),
sessionAffinityType: service.Spec.SessionAffinity,
stickyMaxAgeSeconds: stickyMaxAgeSeconds,
externalIPs: make([]string, len(service.Spec.ExternalIPs)),
loadBalancerSourceRanges: make([]string, len(service.Spec.LoadBalancerSourceRanges)),
onlyNodeLocalEndpoints: onlyNodeLocalEndpoints,
}
copy(info.loadBalancerSourceRanges, service.Spec.LoadBalancerSourceRanges)
copy(info.externalIPs, service.Spec.ExternalIPs)
if apiservice.NeedsHealthCheck(service) {
p := service.Spec.HealthCheckNodePort
if p == 0 {
glog.Errorf("Service %q has no healthcheck nodeport", svcPortName.NamespacedName.String())
} else {
info.healthCheckNodePort = int(p)
}
}
// Store the following for performance reasons.
info.serviceNameString = svcPortName.String()
return info
}
func (sm *proxyServiceMap) merge(other proxyServiceMap) sets.String {
existingPorts := sets.NewString()
for svcPortName, info := range other {
existingPorts.Insert(svcPortName.Port)
_, exists := (*sm)[svcPortName]
if !exists {
glog.V(1).Infof("Adding new service port %q at %s:%d/%s", svcPortName, info.clusterIP, info.port, info.protocol)
} else {
glog.V(1).Infof("Updating existing service port %q at %s:%d/%s", svcPortName, info.clusterIP, info.port, info.protocol)
}
(*sm)[svcPortName] = info
}
return existingPorts
}
func (sm *proxyServiceMap) unmerge(other proxyServiceMap, existingPorts, staleServices sets.String) {
for svcPortName := range other {
if existingPorts.Has(svcPortName.Port) {
continue
}
info, exists := (*sm)[svcPortName]
if exists {
glog.V(1).Infof("Removing service port %q", svcPortName)
if info.protocol == api.ProtocolUDP {
staleServices.Insert(info.clusterIP.String())
}
delete(*sm, svcPortName)
} else {
glog.Errorf("Service port %q removed, but doesn't exists", svcPortName)
}
}
}
type serviceChangeMap struct {
lock sync.Mutex
items map[types.NamespacedName]*serviceChange
}
type serviceChange struct {
previous proxyServiceMap
current proxyServiceMap
}
type updateEndpointMapResult struct {
hcEndpoints map[types.NamespacedName]int
staleEndpoints map[endpointServicePair]bool
staleServiceNames map[proxy.ServicePortName]bool
}
type updateServiceMapResult struct {
hcServices map[types.NamespacedName]uint16
staleServices sets.String
}
func newServiceChangeMap() serviceChangeMap {
return serviceChangeMap{
items: make(map[types.NamespacedName]*serviceChange),
}
}
func (scm *serviceChangeMap) update(namespacedName *types.NamespacedName, previous, current *api.Service) bool {
scm.lock.Lock()
defer scm.lock.Unlock()
change, exists := scm.items[*namespacedName]
if !exists {
change = &serviceChange{}
change.previous = serviceToServiceMap(previous)
scm.items[*namespacedName] = change
}
change.current = serviceToServiceMap(current)
if reflect.DeepEqual(change.previous, change.current) {
delete(scm.items, *namespacedName)
}
return len(scm.items) > 0
}
// Translates single Service object to proxyServiceMap.
//
// NOTE: service object should NOT be modified.
func serviceToServiceMap(service *api.Service) proxyServiceMap {
if service == nil {
return nil
}
svcName := types.NamespacedName{Namespace: service.Namespace, Name: service.Name}
if utilproxy.ShouldSkipService(svcName, service) {
return nil
}
serviceMap := make(proxyServiceMap)
for i := range service.Spec.Ports {
servicePort := &service.Spec.Ports[i]
svcPortName := proxy.ServicePortName{NamespacedName: svcName, Port: servicePort.Name}
serviceMap[svcPortName] = newServiceInfo(svcPortName, servicePort, service)
}
return serviceMap
}
// internal struct for endpoints information
type endpointsInfo struct {
endpoint string // TODO: should be an endpointString type
isLocal bool
}
func (e *endpointsInfo) String() string {
return fmt.Sprintf("%v", *e)
}
// IPPart returns just the IP part of the endpoint.
func (e *endpointsInfo) IPPart() string {
return utilproxy.IPPart(e.endpoint)
}
// PortPart returns just the Port part of the endpoint.
func (e *endpointsInfo) PortPart() (int, error) {
return utilproxy.PortPart(e.endpoint)
}
type endpointServicePair struct {
endpoint string
servicePortName proxy.ServicePortName
}
type proxyEndpointsMap map[proxy.ServicePortName][]*endpointsInfo
type endpointsChange struct {
previous proxyEndpointsMap
current proxyEndpointsMap
}
type endpointsChangeMap struct {
lock sync.Mutex
hostname string
items map[types.NamespacedName]*endpointsChange
}
// <staleEndpoints> and <staleServices> are modified by this function with detected stale connections.
func detectStaleConnections(oldEndpointsMap, newEndpointsMap proxyEndpointsMap, staleEndpoints map[endpointServicePair]bool, staleServiceNames map[proxy.ServicePortName]bool) {
for svcPortName, epList := range oldEndpointsMap {
for _, ep := range epList {
stale := true
for i := range newEndpointsMap[svcPortName] {
if *newEndpointsMap[svcPortName][i] == *ep {
stale = false
break
}
}
if stale {
glog.V(4).Infof("Stale endpoint %v -> %v", svcPortName, ep.endpoint)
staleEndpoints[endpointServicePair{endpoint: ep.endpoint, servicePortName: svcPortName}] = true
}
}
}
for svcPortName, epList := range newEndpointsMap {
// For udp service, if its backend changes from 0 to non-0. There may exist a conntrack entry that could blackhole traffic to the service.
if len(epList) > 0 && len(oldEndpointsMap[svcPortName]) == 0 {
staleServiceNames[svcPortName] = true
}
}
}
// <endpointsMap> is updated by this function (based on the given changes).
// <changes> map is cleared after applying them.
func updateEndpointsMap(
endpointsMap proxyEndpointsMap,
changes *endpointsChangeMap,
hostname string) (result updateEndpointMapResult) {
result.staleEndpoints = make(map[endpointServicePair]bool)
result.staleServiceNames = make(map[proxy.ServicePortName]bool)
func() {
changes.lock.Lock()
defer changes.lock.Unlock()
for _, change := range changes.items {
endpointsMap.unmerge(change.previous)
endpointsMap.merge(change.current)
detectStaleConnections(change.previous, change.current, result.staleEndpoints, result.staleServiceNames)
}
changes.items = make(map[types.NamespacedName]*endpointsChange)
}()
if !utilfeature.DefaultFeatureGate.Enabled(features.ExternalTrafficLocalOnly) {
return
}
// TODO: If this will appear to be computationally expensive, consider
// computing this incrementally similarly to endpointsMap.
result.hcEndpoints = make(map[types.NamespacedName]int)
localIPs := getLocalIPs(endpointsMap)
for nsn, ips := range localIPs {
result.hcEndpoints[nsn] = len(ips)
}
return result
}
// Translates single Endpoints object to proxyEndpointsMap.
// This function is used for incremental updated of endpointsMap.
//
// NOTE: endpoints object should NOT be modified.
func endpointsToEndpointsMap(endpoints *api.Endpoints, hostname string) proxyEndpointsMap {
if endpoints == nil {
return nil
}
endpointsMap := make(proxyEndpointsMap)
// We need to build a map of portname -> all ip:ports for that
// portname. Explode Endpoints.Subsets[*] into this structure.
for i := range endpoints.Subsets {
ss := &endpoints.Subsets[i]
for i := range ss.Ports {
port := &ss.Ports[i]
if port.Port == 0 {
glog.Warningf("ignoring invalid endpoint port %s", port.Name)
continue
}
svcPort := proxy.ServicePortName{
NamespacedName: types.NamespacedName{Namespace: endpoints.Namespace, Name: endpoints.Name},
Port: port.Name,
}
for i := range ss.Addresses {
addr := &ss.Addresses[i]
if addr.IP == "" {
glog.Warningf("ignoring invalid endpoint port %s with empty host", port.Name)
continue
}
epInfo := &endpointsInfo{
endpoint: net.JoinHostPort(addr.IP, strconv.Itoa(int(port.Port))),
isLocal: addr.NodeName != nil && *addr.NodeName == hostname,
}
endpointsMap[svcPort] = append(endpointsMap[svcPort], epInfo)
}
if glog.V(3) {
newEPList := []string{}
for _, ep := range endpointsMap[svcPort] {
newEPList = append(newEPList, ep.endpoint)
}
glog.Infof("Setting endpoints for %q to %+v", svcPort, newEPList)
}
}
}
return endpointsMap
}
func newEndpointsChangeMap(hostname string) endpointsChangeMap {
return endpointsChangeMap{
hostname: hostname,
items: make(map[types.NamespacedName]*endpointsChange),
}
}
func (ecm *endpointsChangeMap) update(namespacedName *types.NamespacedName, previous, current *api.Endpoints) bool {
ecm.lock.Lock()
defer ecm.lock.Unlock()
change, exists := ecm.items[*namespacedName]
if !exists {
change = &endpointsChange{}
change.previous = endpointsToEndpointsMap(previous, ecm.hostname)
ecm.items[*namespacedName] = change
}
change.current = endpointsToEndpointsMap(current, ecm.hostname)
if reflect.DeepEqual(change.previous, change.current) {
delete(ecm.items, *namespacedName)
}
return len(ecm.items) > 0
}
func (em proxyEndpointsMap) merge(other proxyEndpointsMap) {
for svcPort := range other {
em[svcPort] = other[svcPort]
}
}
func (em proxyEndpointsMap) unmerge(other proxyEndpointsMap) {
for svcPort := range other {
delete(em, svcPort)
}
}
// CanUseIPVSProxier returns true if we can use the ipvs Proxier.
// This is determined by checking if all the required kernel modules can be loaded. It may
// return an error if it fails to get the kernel modules information without error, in which
// case it will also return false.
func CanUseIPVSProxier(ipsetver IPSetVersioner) (bool, error) {
// Try to load IPVS required kernel modules using modprobe
for _, kmod := range ipvsModules {
err := utilexec.New().Command("modprobe", "--", kmod).Run()
if err != nil {
glog.Warningf("Failed to load kernel module %v with modprobe. "+
"You can ignore this message when kube-proxy is running inside container without mounting /lib/modules", kmod)
}
}
// Find out loaded kernel modules
out, err := utilexec.New().Command("cut", "-f1", "-d", " ", "/proc/modules").CombinedOutput()
if err != nil {
return false, err
}
mods := strings.Split(string(out), "\n")
wantModules := sets.NewString()
loadModules := sets.NewString()
wantModules.Insert(ipvsModules...)
loadModules.Insert(mods...)
modules := wantModules.Difference(loadModules).UnsortedList()
if len(modules) != 0 {
return false, fmt.Errorf("IPVS proxier will not be used because the following required kernel modules are not loaded: %v", modules)
}
// Check ipset version
versionString, err := ipsetver.GetVersion()
if err != nil {
return false, fmt.Errorf("error getting ipset version, error: %v", err)
}
if !checkMinVersion(versionString) {
return false, fmt.Errorf("ipset version: %s is less than min required version: %s", versionString, MinIPSetCheckVersion)
}
return true, nil
}
// TODO: make it simpler.
// CleanupIptablesLeftovers removes all iptables rules and chains created by the Proxier
// It returns true if an error was encountered. Errors are logged.
func cleanupIptablesLeftovers(ipt utiliptables.Interface) (encounteredError bool) {
// Unlink the services chain.
args := []string{
"-m", "comment", "--comment", "kubernetes service portals",
"-j", string(kubeServicesChain),
}
tableChainsWithJumpServices := []struct {
table utiliptables.Table
chain utiliptables.Chain
}{
{utiliptables.TableNAT, utiliptables.ChainOutput},
{utiliptables.TableNAT, utiliptables.ChainPrerouting},
}
for _, tc := range tableChainsWithJumpServices {
if err := ipt.DeleteRule(tc.table, tc.chain, args...); err != nil {
if !utiliptables.IsNotFoundError(err) {
glog.Errorf("Error removing pure-iptables proxy rule: %v", err)
encounteredError = true
}
}
}
// Unlink the postrouting chain.
args = []string{
"-m", "comment", "--comment", "kubernetes postrouting rules",
"-j", string(kubePostroutingChain),
}
if err := ipt.DeleteRule(utiliptables.TableNAT, utiliptables.ChainPostrouting, args...); err != nil {
if !utiliptables.IsNotFoundError(err) {
glog.Errorf("Error removing ipvs Proxier iptables rule: %v", err)
encounteredError = true
}
}
// Flush and remove all of our chains.
iptablesData := bytes.NewBuffer(nil)
if err := ipt.SaveInto(utiliptables.TableNAT, iptablesData); err != nil {
glog.Errorf("Failed to execute iptables-save for %s: %v", utiliptables.TableNAT, err)
encounteredError = true
} else {
existingNATChains := utiliptables.GetChainLines(utiliptables.TableNAT, iptablesData.Bytes())
natChains := bytes.NewBuffer(nil)
natRules := bytes.NewBuffer(nil)
writeLine(natChains, "*nat")
// Start with chains we know we need to remove.
for _, chain := range []utiliptables.Chain{kubeServicesChain, kubePostroutingChain, KubeMarkMasqChain, KubeServiceIPSetsChain} {
if _, found := existingNATChains[chain]; found {
chainString := string(chain)
writeLine(natChains, existingNATChains[chain]) // flush
writeLine(natRules, "-X", chainString) // delete
}
}
writeLine(natRules, "COMMIT")
natLines := append(natChains.Bytes(), natRules.Bytes()...)
// Write it.
err = ipt.Restore(utiliptables.TableNAT, natLines, utiliptables.NoFlushTables, utiliptables.RestoreCounters)
if err != nil {
glog.Errorf("Failed to execute iptables-restore for %s: %v", utiliptables.TableNAT, err)
encounteredError = true
}
}
return encounteredError
}
// CleanupLeftovers clean up all ipvs and iptables rules created by ipvs Proxier.
func CleanupLeftovers(ipvs utilipvs.Interface, ipt utiliptables.Interface, ipset utilipset.Interface, cleanupIPVS bool) (encounteredError bool) {
if cleanupIPVS {
// Return immediately when ipvs interface is nil - Probably initialization failed in somewhere.
if ipvs == nil {
return true
}
encounteredError = false
err := ipvs.Flush()
if err != nil {
encounteredError = true
}
}
// Delete dummy interface created by ipvs Proxier.
nl := NewNetLinkHandle()
err := nl.DeleteDummyDevice(DefaultDummyDevice)
if err != nil {
encounteredError = true
}
// Clear iptables created by ipvs Proxier.
encounteredError = cleanupIptablesLeftovers(ipt) || encounteredError
// Destroy ip sets created by ipvs Proxier. We should call it after cleaning up
// iptables since we can NOT delete ip set which is still referenced by iptables.
ipSetsToDestroy := []string{KubeLoopBackIPSet, KubeClusterIPSet, KubeLoadBalancerSet, KubeNodePortSetTCP, KubeNodePortSetUDP,
KubeExternalIPSet, KubeLoadBalancerSourceIPSet, KubeLoadBalancerSourceCIDRSet, KubeLoadBalancerMasqSet}
for _, set := range ipSetsToDestroy {
err = ipset.DestroySet(set)
if err != nil {
encounteredError = true
}
}
return encounteredError
}
// Sync is called to synchronize the proxier state to iptables and ipvs as soon as possible.
func (proxier *Proxier) Sync() {
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.UpdateTimestamp()
}
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 *api.Service) {
namespacedName := types.NamespacedName{Namespace: service.Namespace, Name: service.Name}
if proxier.serviceChanges.update(&namespacedName, nil, service) && proxier.isInitialized() {
proxier.syncRunner.Run()
}
}
// OnServiceUpdate is called whenever modification of an existing service object is observed.
func (proxier *Proxier) OnServiceUpdate(oldService, service *api.Service) {
namespacedName := types.NamespacedName{Namespace: service.Namespace, Name: service.Name}
if proxier.serviceChanges.update(&namespacedName, oldService, service) && proxier.isInitialized() {
proxier.syncRunner.Run()
}
}
// OnServiceDelete is called whenever deletion of an existing service object is observed.
func (proxier *Proxier) OnServiceDelete(service *api.Service) {
namespacedName := types.NamespacedName{Namespace: service.Namespace, Name: service.Name}
if proxier.serviceChanges.update(&namespacedName, service, nil) && proxier.isInitialized() {
proxier.syncRunner.Run()
}
}
// OnServiceSynced is called once all the initial even 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.servicesSynced && proxier.endpointsSynced)
proxier.mu.Unlock()
// Sync unconditionally - this is called once per lifetime.
proxier.syncProxyRules()
}
// OnEndpointsAdd is called whenever creation of new endpoints object is observed.
func (proxier *Proxier) OnEndpointsAdd(endpoints *api.Endpoints) {
namespacedName := types.NamespacedName{Namespace: endpoints.Namespace, Name: endpoints.Name}
if proxier.endpointsChanges.update(&namespacedName, nil, endpoints) && proxier.isInitialized() {
proxier.syncRunner.Run()
}
}
// OnEndpointsUpdate is called whenever modification of an existing endpoints object is observed.
func (proxier *Proxier) OnEndpointsUpdate(oldEndpoints, endpoints *api.Endpoints) {
namespacedName := types.NamespacedName{Namespace: endpoints.Namespace, Name: endpoints.Name}
if proxier.endpointsChanges.update(&namespacedName, oldEndpoints, endpoints) && proxier.isInitialized() {
proxier.syncRunner.Run()
}
}
// OnEndpointsDelete is called whenever deletion of an existing endpoints object is observed.
func (proxier *Proxier) OnEndpointsDelete(endpoints *api.Endpoints) {
namespacedName := types.NamespacedName{Namespace: endpoints.Namespace, Name: endpoints.Name}
if proxier.endpointsChanges.update(&namespacedName, endpoints, nil) && proxier.isInitialized() {
proxier.syncRunner.Run()
}
}
// OnEndpointsSynced is called once all the initial event handlers were called and the state is fully propagated to local cache.
func (proxier *Proxier) OnEndpointsSynced() {
proxier.mu.Lock()
proxier.endpointsSynced = true
proxier.mu.Unlock()
proxier.syncProxyRules()
}
// This is where all of the ipvs calls happen.
// assumes proxier.mu is held
func (proxier *Proxier) syncProxyRules() {
proxier.mu.Lock()
defer proxier.mu.Unlock()
start := time.Now()
defer func() {
metrics.SyncProxyRulesLatency.Observe(metrics.SinceInMicroseconds(start))
glog.V(4).Infof("syncProxyRules took %v", time.Since(start))
}()
// don't sync rules till we've received services and endpoints
if !proxier.endpointsSynced || !proxier.servicesSynced {
glog.V(2).Info("Not syncing ipvs rules until Services and Endpoints have been received from master")
return
}
// We assume that if this was called, we really want to sync them,
// even if nothing changed in the meantime. In other words, callers are
// responsible for detecting no-op changes and not calling this function.
serviceUpdateResult := updateServiceMap(
proxier.serviceMap, &proxier.serviceChanges)
endpointUpdateResult := updateEndpointsMap(
proxier.endpointsMap, &proxier.endpointsChanges, proxier.hostname)
staleServices := serviceUpdateResult.staleServices
// merge stale services gathered from updateEndpointsMap
for svcPortName := range endpointUpdateResult.staleServiceNames {
if svcInfo, ok := proxier.serviceMap[svcPortName]; ok && svcInfo != nil && svcInfo.protocol == api.ProtocolUDP {
glog.V(2).Infof("Stale udp service %v -> %s", svcPortName, svcInfo.clusterIP.String())
staleServices.Insert(svcInfo.clusterIP.String())
}
}
glog.V(3).Infof("Syncing ipvs Proxier rules")
// TODO: UT output result
// Begin install iptables
// Get iptables-save output so we can check for existing chains and rules.
// This will be a map of chain name to chain with rules as stored in iptables-save/iptables-restore
existingNATChains := make(map[utiliptables.Chain]string)
proxier.iptablesData.Reset()
err := proxier.iptables.SaveInto(utiliptables.TableNAT, proxier.iptablesData)
if err != nil { // if we failed to get any rules
glog.Errorf("Failed to execute iptables-save, syncing all rules: %v", err)
} else { // otherwise parse the output
existingNATChains = utiliptables.GetChainLines(utiliptables.TableNAT, proxier.iptablesData.Bytes())
}
// Reset all buffers used later.
// This is to avoid memory reallocations and thus improve performance.
proxier.natChains.Reset()
proxier.natRules.Reset()
// Write table headers.
writeLine(proxier.natChains, "*nat")
// Make sure we keep stats for the top-level chains, if they existed
// (which most should have because we created them above).
if chain, ok := existingNATChains[kubePostroutingChain]; ok {
writeLine(proxier.natChains, chain)
} else {
writeLine(proxier.natChains, utiliptables.MakeChainLine(kubePostroutingChain))
}
// 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.
writeLine(proxier.natRules, []string{
"-A", string(kubePostroutingChain),
"-m", "comment", "--comment", `"kubernetes service traffic requiring SNAT"`,
"-m", "mark", "--mark", proxier.masqueradeMark,
"-j", "MASQUERADE",
}...)
if chain, ok := existingNATChains[KubeMarkMasqChain]; ok {
writeLine(proxier.natChains, chain)
} else {