forked from openshift/sdn
-
Notifications
You must be signed in to change notification settings - Fork 0
/
server_others.go
480 lines (420 loc) · 15.9 KB
/
server_others.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
//go:build !windows
// +build !windows
/*
Copyright 2014 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 app does all of the work necessary to configure and run a
// Kubernetes app process.
package app
import (
"context"
"errors"
"fmt"
goruntime "runtime"
"strings"
"time"
"github.com/google/cadvisor/machine"
"github.com/google/cadvisor/utils/sysfs"
"k8s.io/apimachinery/pkg/watch"
"k8s.io/apimachinery/pkg/runtime"
"k8s.io/client-go/tools/cache"
"k8s.io/apimachinery/pkg/fields"
v1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
clientset "k8s.io/client-go/kubernetes"
toolswatch "k8s.io/client-go/tools/watch"
utilsysctl "k8s.io/component-helpers/node/util/sysctl"
"k8s.io/kubernetes/pkg/proxy"
proxyconfigapi "k8s.io/kubernetes/pkg/proxy/apis/config"
"k8s.io/kubernetes/pkg/proxy/iptables"
"k8s.io/kubernetes/pkg/proxy/ipvs"
utilipset "k8s.io/kubernetes/pkg/proxy/ipvs/ipset"
utilipvs "k8s.io/kubernetes/pkg/proxy/ipvs/util"
proxymetrics "k8s.io/kubernetes/pkg/proxy/metrics"
proxyutil "k8s.io/kubernetes/pkg/proxy/util"
proxyutiliptables "k8s.io/kubernetes/pkg/proxy/util/iptables"
utiliptables "k8s.io/kubernetes/pkg/util/iptables"
"k8s.io/utils/exec"
"k8s.io/klog/v2"
)
// timeoutForNodePodCIDR is the time to wait for allocators to assign a PodCIDR to the
// node after it is registered.
var timeoutForNodePodCIDR = 5 * time.Minute
// platformApplyDefaults is called after parsing command-line flags and/or reading the
// config file, to apply platform-specific default values to config.
func (o *Options) platformApplyDefaults(config *proxyconfigapi.KubeProxyConfiguration) {
if config.Mode == "" {
klog.InfoS("Using iptables proxy")
config.Mode = proxyconfigapi.ProxyModeIPTables
}
if config.DetectLocalMode == "" {
klog.V(4).InfoS("Defaulting detect-local-mode", "localModeClusterCIDR", string(proxyconfigapi.LocalModeClusterCIDR))
config.DetectLocalMode = proxyconfigapi.LocalModeClusterCIDR
}
klog.V(2).InfoS("DetectLocalMode", "localMode", string(config.DetectLocalMode))
}
// platformSetup is called after setting up the ProxyServer, but before creating the
// Proxier. It should fill in any platform-specific fields and perform other
// platform-specific setup.
func (s *ProxyServer) platformSetup() error {
if s.Config.DetectLocalMode == proxyconfigapi.LocalModeNodeCIDR {
klog.InfoS("Watching for node, awaiting podCIDR allocation", "hostname", s.Hostname)
node, err := waitForPodCIDR(s.Client, s.Hostname)
if err != nil {
return err
}
s.podCIDRs = node.Spec.PodCIDRs
klog.InfoS("NodeInfo", "podCIDRs", node.Spec.PodCIDRs)
}
err := s.setupConntrack()
if err != nil {
return err
}
proxymetrics.RegisterMetrics()
return nil
}
// platformCheckSupported is called immediately before creating the Proxier, to check
// what IP families are supported (and whether the configuration is usable at all).
func (s *ProxyServer) platformCheckSupported() (ipv4Supported, ipv6Supported, dualStackSupported bool, err error) {
execer := exec.New()
ipt := utiliptables.New(execer, utiliptables.ProtocolIPv4)
ipv4Supported = ipt.Present()
ipt = utiliptables.New(execer, utiliptables.ProtocolIPv6)
ipv6Supported = ipt.Present()
// The Linux proxies can always support dual-stack if they can support both IPv4
// and IPv6.
dualStackSupported = ipv4Supported && ipv6Supported
if !ipv4Supported && !ipv6Supported {
err = fmt.Errorf("iptables is not available on this host")
} else if !ipv4Supported {
klog.InfoS("No iptables support for family", "ipFamily", v1.IPv4Protocol)
} else if !ipv6Supported {
klog.InfoS("No iptables support for family", "ipFamily", v1.IPv6Protocol)
}
return
}
// createProxier creates the proxy.Provider
func (s *ProxyServer) createProxier(config *proxyconfigapi.KubeProxyConfiguration, dualStack bool) (proxy.Provider, error) {
var proxier proxy.Provider
var err error
primaryProtocol := utiliptables.ProtocolIPv4
if s.PrimaryIPFamily == v1.IPv6Protocol {
primaryProtocol = utiliptables.ProtocolIPv6
}
execer := exec.New()
iptInterface := utiliptables.New(execer, primaryProtocol)
var ipt [2]utiliptables.Interface
// Create iptables handlers for both families, one is already created
// Always ordered as IPv4, IPv6
if primaryProtocol == utiliptables.ProtocolIPv4 {
ipt[0] = iptInterface
ipt[1] = utiliptables.New(execer, utiliptables.ProtocolIPv6)
} else {
ipt[0] = utiliptables.New(execer, utiliptables.ProtocolIPv4)
ipt[1] = iptInterface
}
if config.Mode == proxyconfigapi.ProxyModeIPTables {
klog.InfoS("Using iptables Proxier")
if dualStack {
// Always ordered to match []ipt
var localDetectors [2]proxyutiliptables.LocalTrafficDetector
localDetectors, err = getDualStackLocalDetectorTuple(config.DetectLocalMode, config, s.podCIDRs)
if err != nil {
return nil, fmt.Errorf("unable to create proxier: %v", err)
}
// TODO this has side effects that should only happen when Run() is invoked.
proxier, err = iptables.NewDualStackProxier(
ipt,
utilsysctl.New(),
execer,
config.IPTables.SyncPeriod.Duration,
config.IPTables.MinSyncPeriod.Duration,
config.IPTables.MasqueradeAll,
*config.IPTables.LocalhostNodePorts,
int(*config.IPTables.MasqueradeBit),
localDetectors,
s.Hostname,
s.NodeIPs,
s.Recorder,
s.HealthzServer,
config.NodePortAddresses,
)
} else {
// Create a single-stack proxier if and only if the node does not support dual-stack (i.e, no iptables support).
var localDetector proxyutiliptables.LocalTrafficDetector
localDetector, err = getLocalDetector(s.PrimaryIPFamily, config.DetectLocalMode, config, s.podCIDRs)
if err != nil {
return nil, fmt.Errorf("unable to create proxier: %v", err)
}
// TODO this has side effects that should only happen when Run() is invoked.
proxier, err = iptables.NewProxier(
s.PrimaryIPFamily,
iptInterface,
utilsysctl.New(),
execer,
config.IPTables.SyncPeriod.Duration,
config.IPTables.MinSyncPeriod.Duration,
config.IPTables.MasqueradeAll,
*config.IPTables.LocalhostNodePorts,
int(*config.IPTables.MasqueradeBit),
localDetector,
s.Hostname,
s.NodeIPs[s.PrimaryIPFamily],
s.Recorder,
s.HealthzServer,
config.NodePortAddresses,
)
}
if err != nil {
return nil, fmt.Errorf("unable to create proxier: %v", err)
}
} else if config.Mode == proxyconfigapi.ProxyModeIPVS {
kernelHandler := ipvs.NewLinuxKernelHandler()
ipsetInterface := utilipset.New(execer)
ipvsInterface := utilipvs.New()
if err := ipvs.CanUseIPVSProxier(ipvsInterface, ipsetInterface, config.IPVS.Scheduler); err != nil {
return nil, fmt.Errorf("can't use the IPVS proxier: %v", err)
}
klog.InfoS("Using ipvs Proxier")
if dualStack {
// Always ordered to match []ipt
var localDetectors [2]proxyutiliptables.LocalTrafficDetector
localDetectors, err = getDualStackLocalDetectorTuple(config.DetectLocalMode, config, s.podCIDRs)
if err != nil {
return nil, fmt.Errorf("unable to create proxier: %v", err)
}
proxier, err = ipvs.NewDualStackProxier(
ipt,
ipvsInterface,
ipsetInterface,
utilsysctl.New(),
execer,
config.IPVS.SyncPeriod.Duration,
config.IPVS.MinSyncPeriod.Duration,
config.IPVS.ExcludeCIDRs,
config.IPVS.StrictARP,
config.IPVS.TCPTimeout.Duration,
config.IPVS.TCPFinTimeout.Duration,
config.IPVS.UDPTimeout.Duration,
config.IPTables.MasqueradeAll,
int(*config.IPTables.MasqueradeBit),
localDetectors,
s.Hostname,
s.NodeIPs,
s.Recorder,
s.HealthzServer,
config.IPVS.Scheduler,
config.NodePortAddresses,
kernelHandler,
)
} else {
var localDetector proxyutiliptables.LocalTrafficDetector
localDetector, err = getLocalDetector(s.PrimaryIPFamily, config.DetectLocalMode, config, s.podCIDRs)
if err != nil {
return nil, fmt.Errorf("unable to create proxier: %v", err)
}
proxier, err = ipvs.NewProxier(
s.PrimaryIPFamily,
iptInterface,
ipvsInterface,
ipsetInterface,
utilsysctl.New(),
execer,
config.IPVS.SyncPeriod.Duration,
config.IPVS.MinSyncPeriod.Duration,
config.IPVS.ExcludeCIDRs,
config.IPVS.StrictARP,
config.IPVS.TCPTimeout.Duration,
config.IPVS.TCPFinTimeout.Duration,
config.IPVS.UDPTimeout.Duration,
config.IPTables.MasqueradeAll,
int(*config.IPTables.MasqueradeBit),
localDetector,
s.Hostname,
s.NodeIPs[s.PrimaryIPFamily],
s.Recorder,
s.HealthzServer,
config.IPVS.Scheduler,
config.NodePortAddresses,
kernelHandler,
)
}
if err != nil {
return nil, fmt.Errorf("unable to create proxier: %v", err)
}
}
return proxier, nil
}
func (s *ProxyServer) setupConntrack() error {
ct := &realConntracker{}
max, err := getConntrackMax(s.Config.Conntrack)
if err != nil {
return err
}
if max > 0 {
err := ct.SetMax(max)
if err != nil {
if err != errReadOnlySysFS {
return err
}
// errReadOnlySysFS is caused by a known docker issue (https://github.com/docker/docker/issues/24000),
// the only remediation we know is to restart the docker daemon.
// Here we'll send an node event with specific reason and message, the
// administrator should decide whether and how to handle this issue,
// whether to drain the node and restart docker. Occurs in other container runtimes
// as well.
// TODO(random-liu): Remove this when the docker bug is fixed.
const message = "CRI error: /sys is read-only: " +
"cannot modify conntrack limits, problems may arise later (If running Docker, see docker issue #24000)"
s.Recorder.Eventf(s.NodeRef, nil, v1.EventTypeWarning, err.Error(), "StartKubeProxy", message)
}
}
if s.Config.Conntrack.TCPEstablishedTimeout != nil && s.Config.Conntrack.TCPEstablishedTimeout.Duration > 0 {
timeout := int(s.Config.Conntrack.TCPEstablishedTimeout.Duration / time.Second)
if err := ct.SetTCPEstablishedTimeout(timeout); err != nil {
return err
}
}
if s.Config.Conntrack.TCPCloseWaitTimeout != nil && s.Config.Conntrack.TCPCloseWaitTimeout.Duration > 0 {
timeout := int(s.Config.Conntrack.TCPCloseWaitTimeout.Duration / time.Second)
if err := ct.SetTCPCloseWaitTimeout(timeout); err != nil {
return err
}
}
return nil
}
func getConntrackMax(config proxyconfigapi.KubeProxyConntrackConfiguration) (int, error) {
if config.MaxPerCore != nil && *config.MaxPerCore > 0 {
floor := 0
if config.Min != nil {
floor = int(*config.Min)
}
scaled := int(*config.MaxPerCore) * detectNumCPU()
if scaled > floor {
klog.V(3).InfoS("GetConntrackMax: using scaled conntrack-max-per-core")
return scaled, nil
}
klog.V(3).InfoS("GetConntrackMax: using conntrack-min")
return floor, nil
}
return 0, nil
}
func waitForPodCIDR(client clientset.Interface, nodeName string) (*v1.Node, error) {
// since allocators can assign the podCIDR after the node registers, we do a watch here to wait
// for podCIDR to be assigned, instead of assuming that the Get() on startup will have it.
ctx, cancelFunc := context.WithTimeout(context.TODO(), timeoutForNodePodCIDR)
defer cancelFunc()
fieldSelector := fields.OneTermEqualSelector("metadata.name", nodeName).String()
lw := &cache.ListWatch{
ListFunc: func(options metav1.ListOptions) (object runtime.Object, e error) {
options.FieldSelector = fieldSelector
return client.CoreV1().Nodes().List(ctx, options)
},
WatchFunc: func(options metav1.ListOptions) (i watch.Interface, e error) {
options.FieldSelector = fieldSelector
return client.CoreV1().Nodes().Watch(ctx, options)
},
}
condition := func(event watch.Event) (bool, error) {
// don't process delete events
if event.Type != watch.Modified && event.Type != watch.Added {
return false, nil
}
n, ok := event.Object.(*v1.Node)
if !ok {
return false, fmt.Errorf("event object not of type Node")
}
// don't consider the node if is going to be deleted and keep waiting
if !n.DeletionTimestamp.IsZero() {
return false, nil
}
return n.Spec.PodCIDR != "" && len(n.Spec.PodCIDRs) > 0, nil
}
evt, err := toolswatch.UntilWithSync(ctx, lw, &v1.Node{}, nil, condition)
if err != nil {
return nil, fmt.Errorf("timeout waiting for PodCIDR allocation to configure detect-local-mode %v: %v", proxyconfigapi.LocalModeNodeCIDR, err)
}
if n, ok := evt.Object.(*v1.Node); ok {
return n, nil
}
return nil, fmt.Errorf("event object not of type node")
}
func detectNumCPU() int {
// try get numCPU from /sys firstly due to a known issue (https://github.com/kubernetes/kubernetes/issues/99225)
_, numCPU, err := machine.GetTopology(sysfs.NewRealSysFs())
if err != nil || numCPU < 1 {
return goruntime.NumCPU()
}
return numCPU
}
func getLocalDetector(ipFamily v1.IPFamily, mode proxyconfigapi.LocalMode, config *proxyconfigapi.KubeProxyConfiguration, nodePodCIDRs []string) (proxyutiliptables.LocalTrafficDetector, error) {
switch mode {
case proxyconfigapi.LocalModeClusterCIDR:
// LocalModeClusterCIDR is the default if --detect-local-mode wasn't passed,
// but --cluster-cidr is optional.
clusterCIDRs := strings.TrimSpace(config.ClusterCIDR)
if len(clusterCIDRs) == 0 {
klog.InfoS("Detect-local-mode set to ClusterCIDR, but no cluster CIDR defined")
break
}
cidrsByFamily := proxyutil.MapCIDRsByIPFamily(strings.Split(clusterCIDRs, ","))
if len(cidrsByFamily[ipFamily]) != 0 {
return proxyutiliptables.NewDetectLocalByCIDR(cidrsByFamily[ipFamily][0])
}
klog.InfoS("Detect-local-mode set to ClusterCIDR, but no cluster CIDR for family", "ipFamily", ipFamily)
case proxyconfigapi.LocalModeNodeCIDR:
cidrsByFamily := proxyutil.MapCIDRsByIPFamily(nodePodCIDRs)
if len(cidrsByFamily[ipFamily]) != 0 {
return proxyutiliptables.NewDetectLocalByCIDR(cidrsByFamily[ipFamily][0])
}
klog.InfoS("Detect-local-mode set to NodeCIDR, but no PodCIDR defined at node for family", "ipFamily", ipFamily)
case proxyconfigapi.LocalModeBridgeInterface:
return proxyutiliptables.NewDetectLocalByBridgeInterface(config.DetectLocal.BridgeInterface)
case proxyconfigapi.LocalModeInterfaceNamePrefix:
return proxyutiliptables.NewDetectLocalByInterfaceNamePrefix(config.DetectLocal.InterfaceNamePrefix)
}
klog.InfoS("Defaulting to no-op detect-local")
return proxyutiliptables.NewNoOpLocalDetector(), nil
}
func getDualStackLocalDetectorTuple(mode proxyconfigapi.LocalMode, config *proxyconfigapi.KubeProxyConfiguration, nodePodCIDRs []string) ([2]proxyutiliptables.LocalTrafficDetector, error) {
var localDetectors [2]proxyutiliptables.LocalTrafficDetector
var err error
localDetectors[0], err = getLocalDetector(v1.IPv4Protocol, mode, config, nodePodCIDRs)
if err != nil {
return localDetectors, err
}
localDetectors[1], err = getLocalDetector(v1.IPv6Protocol, mode, config, nodePodCIDRs)
if err != nil {
return localDetectors, err
}
return localDetectors, nil
}
// cleanupAndExit remove iptables rules and ipset/ipvs rules
func cleanupAndExit() error {
execer := exec.New()
// cleanup IPv6 and IPv4 iptables rules, regardless of current configuration
ipts := []utiliptables.Interface{
utiliptables.New(execer, utiliptables.ProtocolIPv4),
utiliptables.New(execer, utiliptables.ProtocolIPv6),
}
ipsetInterface := utilipset.New(execer)
ipvsInterface := utilipvs.New()
var encounteredError bool
for _, ipt := range ipts {
encounteredError = iptables.CleanupLeftovers(ipt) || encounteredError
encounteredError = ipvs.CleanupLeftovers(ipvsInterface, ipt, ipsetInterface) || encounteredError
}
if encounteredError {
return errors.New("encountered an error while tearing down rules")
}
return nil
}