/
status_manager.go
899 lines (799 loc) · 33.9 KB
/
status_manager.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
/*
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.
*/
//go:generate mockgen -source=status_manager.go -destination=testing/mock_pod_status_provider.go -package=testing PodStatusProvider
package status
import (
"context"
"fmt"
"sort"
"strings"
"sync"
"time"
clientset "k8s.io/client-go/kubernetes"
v1 "k8s.io/api/core/v1"
apiequality "k8s.io/apimachinery/pkg/api/equality"
"k8s.io/apimachinery/pkg/api/errors"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/types"
"k8s.io/apimachinery/pkg/util/diff"
"k8s.io/apimachinery/pkg/util/wait"
"k8s.io/klog/v2"
podutil "k8s.io/kubernetes/pkg/api/v1/pod"
kubecontainer "k8s.io/kubernetes/pkg/kubelet/container"
kubepod "k8s.io/kubernetes/pkg/kubelet/pod"
kubetypes "k8s.io/kubernetes/pkg/kubelet/types"
statusutil "k8s.io/kubernetes/pkg/util/pod"
)
// A wrapper around v1.PodStatus that includes a version to enforce that stale pod statuses are
// not sent to the API server.
type versionedPodStatus struct {
status v1.PodStatus
// Monotonically increasing version number (per pod).
version uint64
// Pod name & namespace, for sending updates to API server.
podName string
podNamespace string
}
type podStatusSyncRequest struct {
podUID types.UID
status versionedPodStatus
}
// Updates pod statuses in apiserver. Writes only when new status has changed.
// All methods are thread-safe.
type manager struct {
kubeClient clientset.Interface
podManager kubepod.Manager
// Map from pod UID to sync status of the corresponding pod.
podStatuses map[types.UID]versionedPodStatus
podStatusesLock sync.RWMutex
podStatusChannel chan podStatusSyncRequest
// Map from (mirror) pod UID to latest status version successfully sent to the API server.
// apiStatusVersions must only be accessed from the sync thread.
apiStatusVersions map[kubetypes.MirrorPodUID]uint64
podDeletionSafety PodDeletionSafetyProvider
}
// PodStatusProvider knows how to provide status for a pod. It's intended to be used by other components
// that need to introspect status.
type PodStatusProvider interface {
// GetPodStatus returns the cached status for the provided pod UID, as well as whether it
// was a cache hit.
GetPodStatus(uid types.UID) (v1.PodStatus, bool)
}
// PodDeletionSafetyProvider provides guarantees that a pod can be safely deleted.
type PodDeletionSafetyProvider interface {
// PodResourcesAreReclaimed returns true if the pod can safely be deleted.
PodResourcesAreReclaimed(pod *v1.Pod, status v1.PodStatus) bool
// PodCouldHaveRunningContainers returns true if the pod could have running containers.
PodCouldHaveRunningContainers(pod *v1.Pod) bool
}
// Manager is the Source of truth for kubelet pod status, and should be kept up-to-date with
// the latest v1.PodStatus. It also syncs updates back to the API server.
type Manager interface {
PodStatusProvider
// Start the API server status sync loop.
Start()
// SetPodStatus caches updates the cached status for the given pod, and triggers a status update.
SetPodStatus(pod *v1.Pod, status v1.PodStatus)
// SetContainerReadiness updates the cached container status with the given readiness, and
// triggers a status update.
SetContainerReadiness(podUID types.UID, containerID kubecontainer.ContainerID, ready bool)
// SetContainerStartup updates the cached container status with the given startup, and
// triggers a status update.
SetContainerStartup(podUID types.UID, containerID kubecontainer.ContainerID, started bool)
// TerminatePod resets the container status for the provided pod to terminated and triggers
// a status update.
TerminatePod(pod *v1.Pod)
// RemoveOrphanedStatuses scans the status cache and removes any entries for pods not included in
// the provided podUIDs.
RemoveOrphanedStatuses(podUIDs map[types.UID]bool)
}
const syncPeriod = 10 * time.Second
// NewManager returns a functional Manager.
func NewManager(kubeClient clientset.Interface, podManager kubepod.Manager, podDeletionSafety PodDeletionSafetyProvider) Manager {
return &manager{
kubeClient: kubeClient,
podManager: podManager,
podStatuses: make(map[types.UID]versionedPodStatus),
podStatusChannel: make(chan podStatusSyncRequest, 1000), // Buffer up to 1000 statuses
apiStatusVersions: make(map[kubetypes.MirrorPodUID]uint64),
podDeletionSafety: podDeletionSafety,
}
}
// isPodStatusByKubeletEqual returns true if the given pod statuses are equal when non-kubelet-owned
// pod conditions are excluded.
// This method normalizes the status before comparing so as to make sure that meaningless
// changes will be ignored.
func isPodStatusByKubeletEqual(oldStatus, status *v1.PodStatus) bool {
oldCopy := oldStatus.DeepCopy()
for _, c := range status.Conditions {
if kubetypes.PodConditionByKubelet(c.Type) {
_, oc := podutil.GetPodCondition(oldCopy, c.Type)
if oc == nil || oc.Status != c.Status || oc.Message != c.Message || oc.Reason != c.Reason {
return false
}
}
}
oldCopy.Conditions = status.Conditions
return apiequality.Semantic.DeepEqual(oldCopy, status)
}
func (m *manager) Start() {
// Don't start the status manager if we don't have a client. This will happen
// on the master, where the kubelet is responsible for bootstrapping the pods
// of the master components.
if m.kubeClient == nil {
klog.InfoS("Kubernetes client is nil, not starting status manager")
return
}
klog.InfoS("Starting to sync pod status with apiserver")
//nolint:staticcheck // SA1015 Ticker can leak since this is only called once and doesn't handle termination.
syncTicker := time.NewTicker(syncPeriod).C
// syncPod and syncBatch share the same go routine to avoid sync races.
go wait.Forever(func() {
for {
select {
case syncRequest := <-m.podStatusChannel:
klog.V(5).InfoS("Status Manager: syncing pod with status from podStatusChannel",
"podUID", syncRequest.podUID,
"statusVersion", syncRequest.status.version,
"status", syncRequest.status.status)
m.syncPod(syncRequest.podUID, syncRequest.status)
case <-syncTicker:
klog.V(5).InfoS("Status Manager: syncing batch")
// remove any entries in the status channel since the batch will handle them
for i := len(m.podStatusChannel); i > 0; i-- {
<-m.podStatusChannel
}
m.syncBatch()
}
}
}, 0)
}
func (m *manager) GetPodStatus(uid types.UID) (v1.PodStatus, bool) {
m.podStatusesLock.RLock()
defer m.podStatusesLock.RUnlock()
status, ok := m.podStatuses[types.UID(m.podManager.TranslatePodUID(uid))]
return status.status, ok
}
func (m *manager) SetPodStatus(pod *v1.Pod, status v1.PodStatus) {
m.podStatusesLock.Lock()
defer m.podStatusesLock.Unlock()
// Make sure we're caching a deep copy.
status = *status.DeepCopy()
// Force a status update if deletion timestamp is set. This is necessary
// because if the pod is in the non-running state, the pod worker still
// needs to be able to trigger an update and/or deletion.
m.updateStatusInternal(pod, status, pod.DeletionTimestamp != nil)
}
func (m *manager) SetContainerReadiness(podUID types.UID, containerID kubecontainer.ContainerID, ready bool) {
m.podStatusesLock.Lock()
defer m.podStatusesLock.Unlock()
pod, ok := m.podManager.GetPodByUID(podUID)
if !ok {
klog.V(4).InfoS("Pod has been deleted, no need to update readiness", "podUID", string(podUID))
return
}
oldStatus, found := m.podStatuses[pod.UID]
if !found {
klog.InfoS("Container readiness changed before pod has synced",
"pod", klog.KObj(pod),
"containerID", containerID.String())
return
}
// Find the container to update.
containerStatus, _, ok := findContainerStatus(&oldStatus.status, containerID.String())
if !ok {
klog.InfoS("Container readiness changed for unknown container",
"pod", klog.KObj(pod),
"containerID", containerID.String())
return
}
if containerStatus.Ready == ready {
klog.V(4).InfoS("Container readiness unchanged",
"ready", ready,
"pod", klog.KObj(pod),
"containerID", containerID.String())
return
}
// Make sure we're not updating the cached version.
status := *oldStatus.status.DeepCopy()
containerStatus, _, _ = findContainerStatus(&status, containerID.String())
containerStatus.Ready = ready
// updateConditionFunc updates the corresponding type of condition
updateConditionFunc := func(conditionType v1.PodConditionType, condition v1.PodCondition) {
conditionIndex := -1
for i, condition := range status.Conditions {
if condition.Type == conditionType {
conditionIndex = i
break
}
}
if conditionIndex != -1 {
status.Conditions[conditionIndex] = condition
} else {
klog.InfoS("PodStatus missing condition type", "conditionType", conditionType, "status", status)
status.Conditions = append(status.Conditions, condition)
}
}
updateConditionFunc(v1.PodReady, GeneratePodReadyCondition(&pod.Spec, status.Conditions, status.ContainerStatuses, status.Phase))
updateConditionFunc(v1.ContainersReady, GenerateContainersReadyCondition(&pod.Spec, status.ContainerStatuses, status.Phase))
m.updateStatusInternal(pod, status, false)
}
func (m *manager) SetContainerStartup(podUID types.UID, containerID kubecontainer.ContainerID, started bool) {
m.podStatusesLock.Lock()
defer m.podStatusesLock.Unlock()
pod, ok := m.podManager.GetPodByUID(podUID)
if !ok {
klog.V(4).InfoS("Pod has been deleted, no need to update startup", "podUID", string(podUID))
return
}
oldStatus, found := m.podStatuses[pod.UID]
if !found {
klog.InfoS("Container startup changed before pod has synced",
"pod", klog.KObj(pod),
"containerID", containerID.String())
return
}
// Find the container to update.
containerStatus, _, ok := findContainerStatus(&oldStatus.status, containerID.String())
if !ok {
klog.InfoS("Container startup changed for unknown container",
"pod", klog.KObj(pod),
"containerID", containerID.String())
return
}
if containerStatus.Started != nil && *containerStatus.Started == started {
klog.V(4).InfoS("Container startup unchanged",
"pod", klog.KObj(pod),
"containerID", containerID.String())
return
}
// Make sure we're not updating the cached version.
status := *oldStatus.status.DeepCopy()
containerStatus, _, _ = findContainerStatus(&status, containerID.String())
containerStatus.Started = &started
m.updateStatusInternal(pod, status, false)
}
func findContainerStatus(status *v1.PodStatus, containerID string) (containerStatus *v1.ContainerStatus, init bool, ok bool) {
// Find the container to update.
for i, c := range status.ContainerStatuses {
if c.ContainerID == containerID {
return &status.ContainerStatuses[i], false, true
}
}
for i, c := range status.InitContainerStatuses {
if c.ContainerID == containerID {
return &status.InitContainerStatuses[i], true, true
}
}
return nil, false, false
}
// TerminatePod ensures that the status of containers is properly defaulted at the end of the pod
// lifecycle. As the Kubelet must reconcile with the container runtime to observe container status
// there is always the possibility we are unable to retrieve one or more container statuses due to
// garbage collection, admin action, or loss of temporary data on a restart. This method ensures
// that any absent container status is treated as a failure so that we do not incorrectly describe
// the pod as successful. If we have not yet initialized the pod in the presence of init containers,
// the init container failure status is sufficient to describe the pod as failing, and we do not need
// to override waiting containers (unless there is evidence the pod previously started those containers).
func (m *manager) TerminatePod(pod *v1.Pod) {
m.podStatusesLock.Lock()
defer m.podStatusesLock.Unlock()
// ensure that all containers have a terminated state - because we do not know whether the container
// was successful, always report an error
oldStatus := &pod.Status
if cachedStatus, ok := m.podStatuses[pod.UID]; ok {
oldStatus = &cachedStatus.status
}
status := *oldStatus.DeepCopy()
// once a pod has initialized, any missing status is treated as a failure
if hasPodInitialized(pod) {
for i := range status.ContainerStatuses {
if status.ContainerStatuses[i].State.Terminated != nil {
continue
}
status.ContainerStatuses[i].State = v1.ContainerState{
Terminated: &v1.ContainerStateTerminated{
Reason: "ContainerStatusUnknown",
Message: "The container could not be located when the pod was terminated",
ExitCode: 137,
},
}
}
}
// all but the final suffix of init containers which have no evidence of a container start are
// marked as failed containers
for i := range initializedContainers(status.InitContainerStatuses) {
if status.InitContainerStatuses[i].State.Terminated != nil {
continue
}
status.InitContainerStatuses[i].State = v1.ContainerState{
Terminated: &v1.ContainerStateTerminated{
Reason: "ContainerStatusUnknown",
Message: "The container could not be located when the pod was terminated",
ExitCode: 137,
},
}
}
klog.V(5).InfoS("TerminatePod calling updateStatusInternal", "pod", klog.KObj(pod), "podUID", pod.UID)
m.updateStatusInternal(pod, status, true)
}
// hasPodInitialized returns true if the pod has no evidence of ever starting a regular container, which
// implies those containers should not be transitioned to terminated status.
func hasPodInitialized(pod *v1.Pod) bool {
// a pod without init containers is always initialized
if len(pod.Spec.InitContainers) == 0 {
return true
}
// if any container has ever moved out of waiting state, the pod has initialized
for _, status := range pod.Status.ContainerStatuses {
if status.LastTerminationState.Terminated != nil || status.State.Waiting == nil {
return true
}
}
// if the last init container has ever completed with a zero exit code, the pod is initialized
if l := len(pod.Status.InitContainerStatuses); l > 0 {
container := pod.Status.InitContainerStatuses[l-1]
if state := container.LastTerminationState; state.Terminated != nil && state.Terminated.ExitCode == 0 {
return true
}
if state := container.State; state.Terminated != nil && state.Terminated.ExitCode == 0 {
return true
}
}
// otherwise the pod has no record of being initialized
return false
}
// initializedContainers returns all status except for suffix of containers that are in Waiting
// state, which is the set of containers that have attempted to start at least once. If all containers
// are Watiing, the first container is always returned.
func initializedContainers(containers []v1.ContainerStatus) []v1.ContainerStatus {
for i := len(containers) - 1; i >= 0; i-- {
if containers[i].State.Waiting == nil || containers[i].LastTerminationState.Terminated != nil {
return containers[0 : i+1]
}
}
// always return at least one container
if len(containers) > 0 {
return containers[0:1]
}
return nil
}
// checkContainerStateTransition ensures that no container is trying to transition
// from a terminated to non-terminated state, which is illegal and indicates a
// logical error in the kubelet.
func checkContainerStateTransition(oldStatuses, newStatuses []v1.ContainerStatus, restartPolicy v1.RestartPolicy) error {
// If we should always restart, containers are allowed to leave the terminated state
if restartPolicy == v1.RestartPolicyAlways {
return nil
}
for _, oldStatus := range oldStatuses {
// Skip any container that wasn't terminated
if oldStatus.State.Terminated == nil {
continue
}
// Skip any container that failed but is allowed to restart
if oldStatus.State.Terminated.ExitCode != 0 && restartPolicy == v1.RestartPolicyOnFailure {
continue
}
for _, newStatus := range newStatuses {
if oldStatus.Name == newStatus.Name && newStatus.State.Terminated == nil {
return fmt.Errorf("terminated container %v attempted illegal transition to non-terminated state", newStatus.Name)
}
}
}
return nil
}
// updateStatusInternal updates the internal status cache, and queues an update to the api server if
// necessary. Returns whether an update was triggered.
// This method IS NOT THREAD SAFE and must be called from a locked function.
func (m *manager) updateStatusInternal(pod *v1.Pod, status v1.PodStatus, forceUpdate bool) bool {
var oldStatus v1.PodStatus
cachedStatus, isCached := m.podStatuses[pod.UID]
if isCached {
oldStatus = cachedStatus.status
} else if mirrorPod, ok := m.podManager.GetMirrorPodByPod(pod); ok {
oldStatus = mirrorPod.Status
} else {
oldStatus = pod.Status
}
// Check for illegal state transition in containers
if err := checkContainerStateTransition(oldStatus.ContainerStatuses, status.ContainerStatuses, pod.Spec.RestartPolicy); err != nil {
klog.ErrorS(err, "Status update on pod aborted", "pod", klog.KObj(pod))
return false
}
if err := checkContainerStateTransition(oldStatus.InitContainerStatuses, status.InitContainerStatuses, pod.Spec.RestartPolicy); err != nil {
klog.ErrorS(err, "Status update on pod aborted", "pod", klog.KObj(pod))
return false
}
// Set ContainersReadyCondition.LastTransitionTime.
updateLastTransitionTime(&status, &oldStatus, v1.ContainersReady)
// Set ReadyCondition.LastTransitionTime.
updateLastTransitionTime(&status, &oldStatus, v1.PodReady)
// Set InitializedCondition.LastTransitionTime.
updateLastTransitionTime(&status, &oldStatus, v1.PodInitialized)
// Set PodScheduledCondition.LastTransitionTime.
updateLastTransitionTime(&status, &oldStatus, v1.PodScheduled)
// ensure that the start time does not change across updates.
if oldStatus.StartTime != nil && !oldStatus.StartTime.IsZero() {
status.StartTime = oldStatus.StartTime
} else if status.StartTime.IsZero() {
// if the status has no start time, we need to set an initial time
now := metav1.Now()
status.StartTime = &now
}
normalizeStatus(pod, &status)
// Perform some more extensive logging of container termination state to assist in
// debugging production races (generally not needed).
if klogV := klog.V(5); klogV.Enabled() {
var containers []string
for _, s := range append(append([]v1.ContainerStatus(nil), status.InitContainerStatuses...), status.ContainerStatuses...) {
var current, previous string
switch {
case s.State.Running != nil:
current = "running"
case s.State.Waiting != nil:
current = "waiting"
case s.State.Terminated != nil:
current = fmt.Sprintf("terminated=%d", s.State.Terminated.ExitCode)
default:
current = "unknown"
}
switch {
case s.LastTerminationState.Running != nil:
previous = "running"
case s.LastTerminationState.Waiting != nil:
previous = "waiting"
case s.LastTerminationState.Terminated != nil:
previous = fmt.Sprintf("terminated=%d", s.LastTerminationState.Terminated.ExitCode)
default:
previous = "<none>"
}
containers = append(containers, fmt.Sprintf("(%s state=%s previous=%s)", s.Name, current, previous))
}
sort.Strings(containers)
klogV.InfoS("updateStatusInternal", "version", cachedStatus.version+1, "pod", klog.KObj(pod), "podUID", pod.UID, "containers", strings.Join(containers, " "))
}
// The intent here is to prevent concurrent updates to a pod's status from
// clobbering each other so the phase of a pod progresses monotonically.
if isCached && isPodStatusByKubeletEqual(&cachedStatus.status, &status) && !forceUpdate {
klog.V(3).InfoS("Ignoring same status for pod", "pod", klog.KObj(pod), "status", status)
return false // No new status.
}
newStatus := versionedPodStatus{
status: status,
version: cachedStatus.version + 1,
podName: pod.Name,
podNamespace: pod.Namespace,
}
m.podStatuses[pod.UID] = newStatus
select {
case m.podStatusChannel <- podStatusSyncRequest{pod.UID, newStatus}:
klog.V(5).InfoS("Status Manager: adding pod with new status to podStatusChannel",
"pod", klog.KObj(pod),
"podUID", pod.UID,
"statusVersion", newStatus.version,
"status", newStatus.status)
return true
default:
// Let the periodic syncBatch handle the update if the channel is full.
// We can't block, since we hold the mutex lock.
klog.V(4).InfoS("Skipping the status update for pod for now because the channel is full",
"pod", klog.KObj(pod),
"status", status)
return false
}
}
// updateLastTransitionTime updates the LastTransitionTime of a pod condition.
func updateLastTransitionTime(status, oldStatus *v1.PodStatus, conditionType v1.PodConditionType) {
_, condition := podutil.GetPodCondition(status, conditionType)
if condition == nil {
return
}
// Need to set LastTransitionTime.
lastTransitionTime := metav1.Now()
_, oldCondition := podutil.GetPodCondition(oldStatus, conditionType)
if oldCondition != nil && condition.Status == oldCondition.Status {
lastTransitionTime = oldCondition.LastTransitionTime
}
condition.LastTransitionTime = lastTransitionTime
}
// deletePodStatus simply removes the given pod from the status cache.
func (m *manager) deletePodStatus(uid types.UID) {
m.podStatusesLock.Lock()
defer m.podStatusesLock.Unlock()
delete(m.podStatuses, uid)
}
// TODO(filipg): It'd be cleaner if we can do this without signal from user.
func (m *manager) RemoveOrphanedStatuses(podUIDs map[types.UID]bool) {
m.podStatusesLock.Lock()
defer m.podStatusesLock.Unlock()
for key := range m.podStatuses {
if _, ok := podUIDs[key]; !ok {
klog.V(5).InfoS("Removing pod from status map.", "podUID", key)
delete(m.podStatuses, key)
}
}
}
// syncBatch syncs pods statuses with the apiserver.
func (m *manager) syncBatch() {
var updatedStatuses []podStatusSyncRequest
podToMirror, mirrorToPod := m.podManager.GetUIDTranslations()
func() { // Critical section
m.podStatusesLock.RLock()
defer m.podStatusesLock.RUnlock()
// Clean up orphaned versions.
for uid := range m.apiStatusVersions {
_, hasPod := m.podStatuses[types.UID(uid)]
_, hasMirror := mirrorToPod[uid]
if !hasPod && !hasMirror {
delete(m.apiStatusVersions, uid)
}
}
for uid, status := range m.podStatuses {
syncedUID := kubetypes.MirrorPodUID(uid)
if mirrorUID, ok := podToMirror[kubetypes.ResolvedPodUID(uid)]; ok {
if mirrorUID == "" {
klog.V(5).InfoS("Static pod does not have a corresponding mirror pod; skipping",
"podUID", uid,
"pod", klog.KRef(status.podNamespace, status.podName))
continue
}
syncedUID = mirrorUID
}
if m.needsUpdate(types.UID(syncedUID), status) {
updatedStatuses = append(updatedStatuses, podStatusSyncRequest{uid, status})
} else if m.needsReconcile(uid, status.status) {
// Delete the apiStatusVersions here to force an update on the pod status
// In most cases the deleted apiStatusVersions here should be filled
// soon after the following syncPod() [If the syncPod() sync an update
// successfully].
delete(m.apiStatusVersions, syncedUID)
updatedStatuses = append(updatedStatuses, podStatusSyncRequest{uid, status})
}
}
}()
for _, update := range updatedStatuses {
klog.V(5).InfoS("Status Manager: syncPod in syncbatch", "podUID", update.podUID)
m.syncPod(update.podUID, update.status)
}
}
// syncPod syncs the given status with the API server. The caller must not hold the lock.
func (m *manager) syncPod(uid types.UID, status versionedPodStatus) {
if !m.needsUpdate(uid, status) {
klog.V(1).InfoS("Status for pod is up-to-date; skipping", "podUID", uid)
return
}
// TODO: make me easier to express from client code
pod, err := m.kubeClient.CoreV1().Pods(status.podNamespace).Get(context.TODO(), status.podName, metav1.GetOptions{})
if errors.IsNotFound(err) {
klog.V(3).InfoS("Pod does not exist on the server",
"podUID", uid,
"pod", klog.KRef(status.podNamespace, status.podName))
// If the Pod is deleted the status will be cleared in
// RemoveOrphanedStatuses, so we just ignore the update here.
return
}
if err != nil {
klog.InfoS("Failed to get status for pod",
"podUID", uid,
"pod", klog.KRef(status.podNamespace, status.podName),
"err", err)
return
}
translatedUID := m.podManager.TranslatePodUID(pod.UID)
// Type convert original uid just for the purpose of comparison.
if len(translatedUID) > 0 && translatedUID != kubetypes.ResolvedPodUID(uid) {
klog.V(2).InfoS("Pod was deleted and then recreated, skipping status update",
"pod", klog.KObj(pod),
"oldPodUID", uid,
"podUID", translatedUID)
m.deletePodStatus(uid)
return
}
mergedStatus := mergePodStatus(pod.Status, status.status, m.podDeletionSafety.PodCouldHaveRunningContainers(pod))
newPod, patchBytes, unchanged, err := statusutil.PatchPodStatus(m.kubeClient, pod.Namespace, pod.Name, pod.UID, pod.Status, mergedStatus)
klog.V(3).InfoS("Patch status for pod", "pod", klog.KObj(pod), "patch", string(patchBytes))
if err != nil {
klog.InfoS("Failed to update status for pod", "pod", klog.KObj(pod), "err", err)
return
}
if unchanged {
klog.V(3).InfoS("Status for pod is up-to-date", "pod", klog.KObj(pod), "statusVersion", status.version)
} else {
klog.V(3).InfoS("Status for pod updated successfully", "pod", klog.KObj(pod), "statusVersion", status.version, "status", mergedStatus)
pod = newPod
}
m.apiStatusVersions[kubetypes.MirrorPodUID(pod.UID)] = status.version
// We don't handle graceful deletion of mirror pods.
if m.canBeDeleted(pod, status.status) {
deleteOptions := metav1.DeleteOptions{
GracePeriodSeconds: new(int64),
// Use the pod UID as the precondition for deletion to prevent deleting a
// newly created pod with the same name and namespace.
Preconditions: metav1.NewUIDPreconditions(string(pod.UID)),
}
err = m.kubeClient.CoreV1().Pods(pod.Namespace).Delete(context.TODO(), pod.Name, deleteOptions)
if err != nil {
klog.InfoS("Failed to delete status for pod", "pod", klog.KObj(pod), "err", err)
return
}
klog.V(3).InfoS("Pod fully terminated and removed from etcd", "pod", klog.KObj(pod))
m.deletePodStatus(uid)
}
}
// needsUpdate returns whether the status is stale for the given pod UID.
// This method is not thread safe, and must only be accessed by the sync thread.
func (m *manager) needsUpdate(uid types.UID, status versionedPodStatus) bool {
latest, ok := m.apiStatusVersions[kubetypes.MirrorPodUID(uid)]
if !ok || latest < status.version {
return true
}
pod, ok := m.podManager.GetPodByUID(uid)
if !ok {
return false
}
return m.canBeDeleted(pod, status.status)
}
func (m *manager) canBeDeleted(pod *v1.Pod, status v1.PodStatus) bool {
if pod.DeletionTimestamp == nil || kubetypes.IsMirrorPod(pod) {
return false
}
return m.podDeletionSafety.PodResourcesAreReclaimed(pod, status)
}
// needsReconcile compares the given status with the status in the pod manager (which
// in fact comes from apiserver), returns whether the status needs to be reconciled with
// the apiserver. Now when pod status is inconsistent between apiserver and kubelet,
// kubelet should forcibly send an update to reconcile the inconsistence, because kubelet
// should be the source of truth of pod status.
// NOTE(random-liu): It's simpler to pass in mirror pod uid and get mirror pod by uid, but
// now the pod manager only supports getting mirror pod by static pod, so we have to pass
// static pod uid here.
// TODO(random-liu): Simplify the logic when mirror pod manager is added.
func (m *manager) needsReconcile(uid types.UID, status v1.PodStatus) bool {
// The pod could be a static pod, so we should translate first.
pod, ok := m.podManager.GetPodByUID(uid)
if !ok {
klog.V(4).InfoS("Pod has been deleted, no need to reconcile", "podUID", string(uid))
return false
}
// If the pod is a static pod, we should check its mirror pod, because only status in mirror pod is meaningful to us.
if kubetypes.IsStaticPod(pod) {
mirrorPod, ok := m.podManager.GetMirrorPodByPod(pod)
if !ok {
klog.V(4).InfoS("Static pod has no corresponding mirror pod, no need to reconcile", "pod", klog.KObj(pod))
return false
}
pod = mirrorPod
}
podStatus := pod.Status.DeepCopy()
normalizeStatus(pod, podStatus)
if isPodStatusByKubeletEqual(podStatus, &status) {
// If the status from the source is the same with the cached status,
// reconcile is not needed. Just return.
return false
}
klog.V(3).InfoS("Pod status is inconsistent with cached status for pod, a reconciliation should be triggered",
"pod", klog.KObj(pod),
"statusDiff", diff.ObjectDiff(podStatus, &status))
return true
}
// normalizeStatus normalizes nanosecond precision timestamps in podStatus
// down to second precision (*RFC339NANO* -> *RFC3339*). This must be done
// before comparing podStatus to the status returned by apiserver because
// apiserver does not support RFC339NANO.
// Related issue #15262/PR #15263 to move apiserver to RFC339NANO is closed.
func normalizeStatus(pod *v1.Pod, status *v1.PodStatus) *v1.PodStatus {
bytesPerStatus := kubecontainer.MaxPodTerminationMessageLogLength
if containers := len(pod.Spec.Containers) + len(pod.Spec.InitContainers); containers > 0 {
bytesPerStatus = bytesPerStatus / containers
}
normalizeTimeStamp := func(t *metav1.Time) {
*t = t.Rfc3339Copy()
}
normalizeContainerState := func(c *v1.ContainerState) {
if c.Running != nil {
normalizeTimeStamp(&c.Running.StartedAt)
}
if c.Terminated != nil {
normalizeTimeStamp(&c.Terminated.StartedAt)
normalizeTimeStamp(&c.Terminated.FinishedAt)
if len(c.Terminated.Message) > bytesPerStatus {
c.Terminated.Message = c.Terminated.Message[:bytesPerStatus]
}
}
}
if status.StartTime != nil {
normalizeTimeStamp(status.StartTime)
}
for i := range status.Conditions {
condition := &status.Conditions[i]
normalizeTimeStamp(&condition.LastProbeTime)
normalizeTimeStamp(&condition.LastTransitionTime)
}
// update container statuses
for i := range status.ContainerStatuses {
cstatus := &status.ContainerStatuses[i]
normalizeContainerState(&cstatus.State)
normalizeContainerState(&cstatus.LastTerminationState)
}
// Sort the container statuses, so that the order won't affect the result of comparison
sort.Sort(kubetypes.SortedContainerStatuses(status.ContainerStatuses))
// update init container statuses
for i := range status.InitContainerStatuses {
cstatus := &status.InitContainerStatuses[i]
normalizeContainerState(&cstatus.State)
normalizeContainerState(&cstatus.LastTerminationState)
}
// Sort the container statuses, so that the order won't affect the result of comparison
kubetypes.SortInitContainerStatuses(pod, status.InitContainerStatuses)
return status
}
// mergePodStatus merges oldPodStatus and newPodStatus to preserve where pod conditions
// not owned by kubelet and to ensure terminal phase transition only happens after all
// running containers have terminated. This method does not modify the old status.
func mergePodStatus(oldPodStatus, newPodStatus v1.PodStatus, couldHaveRunningContainers bool) v1.PodStatus {
podConditions := make([]v1.PodCondition, 0, len(oldPodStatus.Conditions)+len(newPodStatus.Conditions))
for _, c := range oldPodStatus.Conditions {
if !kubetypes.PodConditionByKubelet(c.Type) {
podConditions = append(podConditions, c)
}
}
for _, c := range newPodStatus.Conditions {
if kubetypes.PodConditionByKubelet(c.Type) {
podConditions = append(podConditions, c)
}
}
newPodStatus.Conditions = podConditions
// Delay transitioning a pod to a terminal status unless the pod is actually terminal.
// The Kubelet should never transition a pod to terminal status that could have running
// containers and thus actively be leveraging exclusive resources. Note that resources
// like volumes are reconciled by a subsystem in the Kubelet and will converge if a new
// pod reuses an exclusive resource (unmount -> free -> mount), which means we do not
// need wait for those resources to be detached by the Kubelet. In general, resources
// the Kubelet exclusively owns must be released prior to a pod being reported terminal,
// while resources that have participanting components above the API use the pod's
// transition to a terminal phase (or full deletion) to release those resources.
if !podutil.IsPodPhaseTerminal(oldPodStatus.Phase) && podutil.IsPodPhaseTerminal(newPodStatus.Phase) {
if couldHaveRunningContainers {
newPodStatus.Phase = oldPodStatus.Phase
newPodStatus.Reason = oldPodStatus.Reason
newPodStatus.Message = oldPodStatus.Message
}
}
// If the new phase is terminal, explicitly set the ready condition to false for v1.PodReady and v1.ContainersReady.
// It may take some time for kubelet to reconcile the ready condition, so explicitly set ready conditions to false if the phase is terminal.
// This is done to ensure kubelet does not report a status update with terminal pod phase and ready=true.
// See https://issues.k8s.io/108594 for more details.
if podutil.IsPodPhaseTerminal(newPodStatus.Phase) {
if podutil.IsPodReadyConditionTrue(newPodStatus) || podutil.IsContainersReadyConditionTrue(newPodStatus) {
containersReadyCondition := generateContainersReadyConditionForTerminalPhase(newPodStatus.Phase)
podutil.UpdatePodCondition(&newPodStatus, &containersReadyCondition)
podReadyCondition := generatePodReadyConditionForTerminalPhase(newPodStatus.Phase)
podutil.UpdatePodCondition(&newPodStatus, &podReadyCondition)
}
}
return newPodStatus
}
// NeedToReconcilePodReadiness returns if the pod "Ready" condition need to be reconcile
func NeedToReconcilePodReadiness(pod *v1.Pod) bool {
if len(pod.Spec.ReadinessGates) == 0 {
return false
}
podReadyCondition := GeneratePodReadyCondition(&pod.Spec, pod.Status.Conditions, pod.Status.ContainerStatuses, pod.Status.Phase)
i, curCondition := podutil.GetPodConditionFromList(pod.Status.Conditions, v1.PodReady)
// Only reconcile if "Ready" condition is present and Status or Message is not expected
if i >= 0 && (curCondition.Status != podReadyCondition.Status || curCondition.Message != podReadyCondition.Message) {
return true
}
return false
}