forked from cisco-app-networking/networkservicemesh
-
Notifications
You must be signed in to change notification settings - Fork 0
/
nsm_heal_processor.go
486 lines (409 loc) · 15.4 KB
/
nsm_heal_processor.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
package nsm
import (
"context"
"fmt"
"sync"
"time"
"github.com/networkservicemesh/networkservicemesh/controlplane/pkg/properties"
"github.com/networkservicemesh/networkservicemesh/controlplane/pkg/api/nsm"
"github.com/pkg/errors"
"github.com/networkservicemesh/networkservicemesh/pkg/tools/spanhelper"
"github.com/networkservicemesh/networkservicemesh/controlplane/api/connection"
"github.com/networkservicemesh/networkservicemesh/controlplane/api/networkservice"
"github.com/networkservicemesh/networkservicemesh/controlplane/pkg/common"
"github.com/sirupsen/logrus"
"github.com/networkservicemesh/networkservicemesh/controlplane/api/registry"
"github.com/networkservicemesh/networkservicemesh/controlplane/pkg/model"
"github.com/networkservicemesh/networkservicemesh/controlplane/pkg/serviceregistry"
)
type healProcessor struct {
serviceRegistry serviceregistry.ServiceRegistry
model model.Model
props *properties.Properties
healCancellers map[string]func()
healCancellersMutex sync.Mutex
manager nsm.NetworkServiceRequestManager
nseManager nsm.NetworkServiceEndpointManager
eventCh chan healEvent
}
type healEvent struct {
healID string
cc *model.ClientConnection
healState nsm.HealState
ctx context.Context
}
func newNetworkServiceHealProcessor(
serviceRegistry serviceregistry.ServiceRegistry,
model model.Model,
properties *properties.Properties,
manager nsm.NetworkServiceRequestManager,
nseManager nsm.NetworkServiceEndpointManager) nsm.NetworkServiceHealProcessor {
p := &healProcessor{
serviceRegistry: serviceRegistry,
model: model,
props: properties,
manager: manager,
nseManager: nseManager,
eventCh: make(chan healEvent, 1),
healCancellers: make(map[string]func()),
}
go p.serve()
return p
}
func (p *healProcessor) Heal(ctx context.Context, clientConnection nsm.ClientConnection, healState nsm.HealState) {
cc := clientConnection.(*model.ClientConnection)
opName := "Heal"
if clientConnection.GetConnectionSource().IsRemote() {
opName = "RemoteHeal"
}
// We need to create new context, since we need to control a lifetime.
span := spanhelper.CopySpan(context.Background(), spanhelper.GetSpanHelper(ctx), opName)
defer span.Finish()
ctx = span.Context()
logger := span.Logger()
ctx = common.WithLog(ctx, logger)
ctx, cancelFunc := context.WithCancel(ctx)
p.healCancellersMutex.Lock()
p.healCancellers[clientConnection.GetID()] = cancelFunc
p.healCancellersMutex.Unlock()
healID := create_logid()
logger.Infof("NSM_Heal(%v) %v", healID, cc)
if !p.props.HealEnabled {
logger.Infof("NSM_Heal(%v) Is Disabled/Closing connection %v", healID, cc)
_ = p.CloseConnection(ctx, cc)
return
}
if modelCC := p.model.GetClientConnection(cc.GetID()); modelCC == nil {
logger.Errorf("NSM_Heal(%v) Trying to heal not existing connection", healID)
return
} else if modelCC.ConnectionState != model.ClientConnectionReady {
healErr := errors.Errorf("NSM_Heal(%v) Trying to heal connection in bad state", healID)
span.LogError(healErr)
return
}
cc = p.model.ApplyClientConnectionChanges(ctx, cc.GetID(), func(modelCC *model.ClientConnection) {
modelCC.ConnectionState = model.ClientConnectionHealingBegin
})
p.eventCh <- healEvent{
healID: healID,
cc: cc,
healState: healState,
ctx: ctx,
}
}
func (p *healProcessor) CloseConnection(ctx context.Context, conn nsm.ClientConnection) error {
var err error
// Cancel context after closing.
defer func() {
p.healCancellersMutex.Lock()
contextCancelFunc, isHealing := p.healCancellers[conn.GetID()]
if isHealing {
logrus.Infof("Closing client connection while healing, cancel client context")
delete(p.healCancellers, conn.GetID())
contextCancelFunc()
}
p.healCancellersMutex.Unlock()
}()
if conn.GetConnectionSource().IsRemote() {
_, err = p.manager.RemoteManager().Close(ctx, conn.GetConnectionSource())
} else {
_, err = p.manager.LocalManager(conn).Close(ctx, conn.GetConnectionSource())
}
if err != nil {
logrus.Errorf("NSM_Heal Error in Close: %v", err)
}
return err
}
func (p *healProcessor) serve() {
for {
e, ok := <-p.eventCh
if !ok {
return
}
go func() {
span := spanhelper.FromContext(e.ctx, "heal")
defer span.Finish()
ctx := span.Context()
logger := span.Logger()
defer func() {
logger.Infof("NSM_Heal(%v) Connection %v healing state is finished...", e.healID, e.cc.GetID())
}()
healed := false
ctx = common.WithModelConnection(ctx, e.cc)
switch e.healState {
case nsm.HealStateDstDown:
healed = p.healDstDown(ctx, e.cc)
case nsm.HealStateForwarderDown:
healed = p.healForwarderDown(ctx, e.cc)
case nsm.HealStateDstUpdate:
healed = p.healDstUpdate(ctx, e.cc)
case nsm.HealStateDstNmgrDown:
healed = p.healDstMgrDown(ctx, e.cc)
}
if healed {
span.LogValue("status", "healed")
logger.Infof("NSM_Heal(%v) Heal: Connection recovered: %v", e.healID, e.cc)
p.healCancellersMutex.Lock()
delete(p.healCancellers, e.cc.GetID())
p.healCancellersMutex.Unlock()
} else {
span.LogValue("status", "closing")
_ = p.CloseConnection(ctx, e.cc)
}
}()
}
}
func (p *healProcessor) healDstDown(ctx context.Context, cc *model.ClientConnection) bool {
span := spanhelper.FromContext(ctx, "healDstDown")
defer span.Finish()
logger := span.Logger()
// Update context
ctx = span.Context()
logger.Infof("NSM_Heal(1.1.1) Checking if DST die is NSMD/DST die...")
// Check if this is a really HealStateDstDown or HealStateDstNmgrDown
if !p.nseManager.IsLocalEndpoint(cc.Endpoint) {
waitCtx, waitCancel := context.WithTimeout(ctx, p.props.HealTimeout*3)
defer waitCancel()
remoteNsmClient, err := p.nseManager.CreateNSEClient(waitCtx, cc.Endpoint)
if remoteNsmClient != nil {
_ = remoteNsmClient.Cleanup()
}
if err != nil {
// This is NSMD die case.
logger.Infof("NSM_Heal(1.1.2) Connection healing state is %v...", nsm.HealStateDstNmgrDown)
span.LogError(err)
return p.healDstMgrDown(ctx, cc)
}
}
logger.Infof("NSM_Heal(1.1.2) Connection healing state is %v...", nsm.HealStateDstDown)
// Destination is down, we need to find it again.
if cc.Xcon.GetRemoteSource() != nil {
// NSMd id remote one, we just need to close and return.
logger.Infof("NSM_Heal(2.1) Remote NSE heal is done on source side")
return false
}
logger.Infof("NSM_Heal(2.2) Starting DST Heal...")
// We are client NSMd, we need to try recover our connection srv.
// Wait for NSE not equal to down one, since we know it will be re-registered with new endpoint name.
ctx = p.waitForNSEUpdateContext(ctx, cc.Endpoint, cc)
// Fallback to heal with choose of new NSE.
for attempt := 0; attempt < p.props.HealRetryCount; attempt++ {
// If client context is cancelled, we need to stop attempts.
if ctx.Err() != nil {
logger.Info("Client context is broken, stopping heal attempts")
break
}
attemptSpan := spanhelper.FromContext(ctx, fmt.Sprintf("healing-attempt-%v", attempt))
requestCtx, requestCancel := context.WithTimeout(attemptSpan.Context(), p.props.HealRequestTimeout)
defer requestCancel()
defer attemptSpan.Finish()
logger.Infof("NSM_Heal(2.3.0) Starting Heal by calling request: %v", cc.Request)
_, err := p.manager.LocalManager(cc).Request(requestCtx, cc.Request)
span.LogError(err)
if err == nil {
return true
}
logger.Errorf("NSM_Heal(2.3.1) Failed to heal connection: %v. Delaying: %v", err, p.props.HealRetryDelay)
if attempt+1 < p.props.HealRetryCount {
attemptSpan.Finish()
<-time.After(p.props.HealRetryDelay)
continue
}
}
return false
}
func (p *healProcessor) healForwarderDown(ctx context.Context, cc *model.ClientConnection) bool {
var cancel context.CancelFunc
ctx, cancel = context.WithTimeout(ctx, p.props.HealTimeout)
defer cancel()
span := spanhelper.FromContext(ctx, "healForwarderDown")
defer span.Finish()
logger := span.Logger()
// Forwarder is down, we only need to re-programm forwarder.
// 1. Wait for forwarder to appear.
logger.Infof("NSM_Heal(3.1) Waiting for Forwarder to recovery...")
if err := p.serviceRegistry.WaitForForwarderAvailable(span.Context(), p.model, p.props.HealForwarderTimeout); err != nil {
err = errors.Errorf("NSM_Heal(3.1) Forwarder is not available on recovery for timeout %v: %v", p.props.HealForwarderTimeout, err)
span.LogError(err)
return false
}
logger.Infof("NSM_Heal(3.2) Forwarder is now available...")
// 3.3. Set source connection down
p.model.ApplyClientConnectionChanges(span.Context(), cc.GetID(), func(modelCC *model.ClientConnection) {
modelCC.Xcon.Source.State = connection.State_DOWN
})
if cc.Xcon.GetRemoteSource() != nil {
// NSMd id remote one, we just need to close and return.
// Recovery will be performed by NSM client side.
logger.Infof("NSM_Heal(3.4) Healing will be continued on source side...")
return true
}
request := cc.Request.Clone()
request.SetRequestConnection(cc.GetConnectionSource())
if _, err := p.manager.LocalManager(cc).Request(span.Context(), cc.Request); err != nil {
logger.Errorf("NSM_Heal(3.5) Failed to heal connection: %v", err)
return false
}
return true
}
func (p *healProcessor) healDstUpdate(ctx context.Context, cc *model.ClientConnection) bool {
span := spanhelper.FromContext(ctx, "healDstUpdate")
defer span.Finish()
ctx = span.Context()
var cancel context.CancelFunc
ctx, cancel = context.WithTimeout(ctx, p.props.HealTimeout)
defer cancel()
logger := span.Logger()
// Destination is updated.
// Update request to contain a proper connection object from previous attempt.
logger.Infof("NSM_Heal(5.1-%v) Healing Src Update... %v", cc)
if cc.Request == nil {
return false
}
request := cc.Request.Clone()
request.SetRequestConnection(cc.GetConnectionSource())
err := p.performRequest(ctx, request, cc)
if err != nil {
span.LogError(err)
logger.Errorf("NSM_Heal(5.2) Failed to heal connection: %v", err)
return false
}
return true
}
func (p *healProcessor) performRequest(ctx context.Context, request *networkservice.NetworkServiceRequest, cc nsm.ClientConnection) error {
span := spanhelper.FromContext(ctx, "performRequest")
defer span.Finish()
if request.GetConnection().IsRemote() {
resp, err := p.manager.RemoteManager().Request(span.Context(), request)
span.LogObject("response", resp)
return err
}
resp, err := p.manager.LocalManager(cc).Request(span.Context(), request)
span.LogObject("response", resp)
return err
}
func (p *healProcessor) healDstMgrDown(ctx context.Context, cc *model.ClientConnection) bool {
span := spanhelper.FromContext(ctx, "healDstNsmgrDown")
defer span.Finish()
ctx = span.Context()
logger := span.Logger()
logger.Infof("NSM_Heal(6.1) Starting DST + NSMGR Heal...")
var endpointName string
// Wait for exact same NSE to be available with NSMD connection alive.
if cc.Endpoint != nil {
endpointName = cc.Endpoint.GetNetworkServiceEndpoint().GetName()
waitCtx, waitCancel := context.WithTimeout(ctx, p.props.HealTimeout*3)
defer waitCancel()
if !p.waitNSE(waitCtx, endpointName, cc.GetNetworkService(), p.nseIsSameAndAvailable) {
span.LogValue("waitNSE", "failed to find endpoint by name with timeout")
ctx = common.WithIgnoredEndpoints(ctx, map[registry.EndpointNSMName]*registry.NSERegistration{
cc.Endpoint.GetEndpointNSMName(): cc.Endpoint,
})
}
}
for attempt := 0; attempt < p.props.HealRetryCount; attempt++ {
attemptSpan := spanhelper.FromContext(ctx, fmt.Sprintf("healing-attempt-%v", attempt))
requestCtx, requestCancel := context.WithTimeout(attemptSpan.Context(), p.props.HealRequestTimeout)
defer requestCancel()
defer attemptSpan.Finish()
err := p.performRequest(requestCtx, cc.Request, cc)
if err == nil {
attemptSpan.LogObject("state", "healed")
return true
}
err = errors.Errorf("heal(6.2.3) Failed to heal connection: %v. Delaying: %v", err, p.props.HealRetryDelay)
span.LogError(err)
attemptSpan.Finish()
if attempt+1 < p.props.HealRetryCount {
attemptSpan.Finish()
<-time.After(p.props.HealRetryDelay)
continue
}
}
return true
}
type nseValidator func(ctx context.Context, endpoint string, reg *registry.NSERegistration) bool
func (p *healProcessor) nseIsNewAndAvailable(ctx context.Context, endpointName string, reg *registry.NSERegistration) bool {
if endpointName != "" && reg.GetNetworkServiceEndpoint().GetName() == endpointName {
// Skip ignored endpoint
return false
}
// Check remote is accessible.
if p.nseManager.CheckUpdateNSE(ctx, reg) {
logrus.Infof("NSE is available and Remote NSMD is accessible. %s.", reg.NetworkServiceManager.Url)
// We are able to connect to NSM with required NSE
return true
}
return false
}
func (p *healProcessor) nseIsSameAndAvailable(ctx context.Context, endpointName string, reg *registry.NSERegistration) bool {
if reg.GetNetworkServiceEndpoint().GetName() != endpointName {
return false
}
// Our endpoint, we need to check if it is remote one and NSM is accessible.
// Check remote is accessible.
if p.nseManager.CheckUpdateNSE(ctx, reg) {
logrus.Infof("NSE is available and Remote NSMD is accessible. %s.", reg.NetworkServiceManager.Url)
// We are able to connect to NSM with required NSE
return true
}
return false
}
func (p *healProcessor) waitNSE(ctx context.Context, endpointName, networkService string, nseValidator nseValidator) bool {
span := spanhelper.FromContext(ctx, "waitNSE")
defer span.Finish()
ctx = span.Context()
span.LogObject("endpointName", endpointName)
span.LogObject("networkService", networkService)
logger := span.Logger()
discoveryClient, err := p.serviceRegistry.DiscoveryClient(span.Context())
if err != nil {
span.LogError(err)
return false
}
st := time.Now()
nseRequest := ®istry.FindNetworkServiceRequest{
NetworkServiceName: networkService,
}
defer func() {
logger.Infof("Complete Waiting for Remote NSE/NSMD with network service %s. Since elapsed: %v", networkService, time.Since(st))
}()
for {
logger.Infof("NSM: RemoteNSE: Waiting for NSE with network service %s. Since elapsed: %v", networkService, time.Since(st))
// If client context was cancelled, we need to stop waiting.
if ctx.Err() != nil {
logger.Infof("Client context is cancelled, stop waiting for network service %s", networkService)
return false
}
endpointResponse, err := discoveryClient.FindNetworkService(ctx, nseRequest)
if err == nil {
for _, ep := range endpointResponse.NetworkServiceEndpoints {
reg := ®istry.NSERegistration{
NetworkServiceManager: endpointResponse.GetNetworkServiceManagers()[ep.GetNetworkServiceManagerName()],
NetworkServiceEndpoint: ep,
NetworkService: endpointResponse.GetNetworkService(),
}
if nseValidator(ctx, endpointName, reg) {
return true
}
}
}
if time.Since(st) > p.props.HealDSTNSEWaitTimeout {
span.LogError(errors.Errorf("timeout waiting for NetworkService: %v timeout: %v", networkService, time.Since(st)))
return false
}
// Wait a bit
<-time.After(p.props.HealDSTNSEWaitTick)
}
}
func (p *healProcessor) waitForNSEUpdateContext(ctx context.Context, endpoint *registry.NSERegistration, cc *model.ClientConnection) context.Context {
waitCtx, waitCancel := context.WithTimeout(ctx, p.props.HealTimeout*3)
defer waitCancel()
if !p.waitNSE(waitCtx, endpoint.NetworkServiceEndpoint.Name, cc.GetNetworkService(), p.nseIsNewAndAvailable) {
// Mark endpoint as ignored.
return common.WithIgnoredEndpoints(ctx, map[registry.EndpointNSMName]*registry.NSERegistration{
endpoint.GetEndpointNSMName(): cc.Endpoint,
})
}
return ctx
}