Repository navigation
Expand file tree
/
Copy pathvm.go
More file actions
2526 lines (2184 loc) · 88.1 KB
/
Copy pathvm.go
File metadata and controls
2526 lines (2184 loc) · 88.1 KB
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
/*
* This file is part of the KubeVirt project
*
* 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.
*
* Copyright The KubeVirt Authors.
*
*/
package virthandler
import (
"bytes"
"context"
"encoding/json"
goerror "errors"
"fmt"
"os"
"path/filepath"
"regexp"
"runtime"
"sort"
"strings"
"sync"
"time"
"github.com/opencontainers/cgroups"
"libvirt.org/go/libvirtxml"
k8sv1 "k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/api/equality"
"k8s.io/apimachinery/pkg/api/resource"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/util/errors"
"k8s.io/apimachinery/pkg/util/wait"
"k8s.io/client-go/tools/cache"
"k8s.io/client-go/tools/record"
"k8s.io/client-go/util/workqueue"
backupv1 "kubevirt.io/api/backup/v1alpha1"
v1 "kubevirt.io/api/core/v1"
"kubevirt.io/client-go/kubecli"
"kubevirt.io/client-go/log"
"kubevirt.io/kubevirt/pkg/config"
"kubevirt.io/kubevirt/pkg/controller"
"kubevirt.io/kubevirt/pkg/executor"
hostdisk "kubevirt.io/kubevirt/pkg/host-disk"
hotplugdisk "kubevirt.io/kubevirt/pkg/hotplug-disk"
"kubevirt.io/kubevirt/pkg/hypervisor"
metrics "kubevirt.io/kubevirt/pkg/monitoring/metrics/common/vmisync"
vhmetrics "kubevirt.io/kubevirt/pkg/monitoring/metrics/virt-handler"
"kubevirt.io/kubevirt/pkg/network/domainspec"
neterrors "kubevirt.io/kubevirt/pkg/network/errors"
netsetup "kubevirt.io/kubevirt/pkg/network/setup"
netvmispec "kubevirt.io/kubevirt/pkg/network/vmispec"
pluginv1alpha1 "kubevirt.io/api/plugin/v1alpha1"
"kubevirt.io/kubevirt/pkg/safepath"
"kubevirt.io/kubevirt/pkg/storage/reservation"
storagetypes "kubevirt.io/kubevirt/pkg/storage/types"
"kubevirt.io/kubevirt/pkg/util"
"kubevirt.io/kubevirt/pkg/util/migrations"
virtconfig "kubevirt.io/kubevirt/pkg/virt-config"
"kubevirt.io/kubevirt/pkg/virt-config/featuregate"
"kubevirt.io/kubevirt/pkg/virt-controller/watch/topology"
virtcache "kubevirt.io/kubevirt/pkg/virt-handler/cache"
"kubevirt.io/kubevirt/pkg/virt-handler/cgroup"
cmdclient "kubevirt.io/kubevirt/pkg/virt-handler/cmd-client"
containerdisk "kubevirt.io/kubevirt/pkg/virt-handler/container-disk"
deviceManager "kubevirt.io/kubevirt/pkg/virt-handler/device-manager"
"kubevirt.io/kubevirt/pkg/virt-handler/heartbeat"
hotplugvolume "kubevirt.io/kubevirt/pkg/virt-handler/hotplug-disk"
"kubevirt.io/kubevirt/pkg/virt-handler/isolation"
launcherclients "kubevirt.io/kubevirt/pkg/virt-handler/launcher-clients"
migrationproxy "kubevirt.io/kubevirt/pkg/virt-handler/migration-proxy"
multipathmonitor "kubevirt.io/kubevirt/pkg/virt-handler/multipath-monitor"
"kubevirt.io/kubevirt/pkg/virt-handler/plugins"
"kubevirt.io/kubevirt/pkg/virt-handler/selinux"
"kubevirt.io/kubevirt/pkg/virt-launcher/virtwrap/api"
)
type netstat interface {
UpdateStatus(vmi *v1.VirtualMachineInstance, domain *api.Domain) error
Teardown(vmi *v1.VirtualMachineInstance)
}
type downwardMetricsManager interface {
Run(stopCh chan struct{})
StartServer(vmi *v1.VirtualMachineInstance, pid int) error
StopServer(vmi *v1.VirtualMachineInstance)
}
type VirtualMachineController struct {
*BaseController
capabilities *libvirtxml.Caps
clientset kubecli.KubevirtClient
containerDiskMounter containerdisk.Mounter
downwardMetricsManager downwardMetricsManager
hotplugVolumeMounter hotplugvolume.VolumeMounter
hostCpuModel string
ioErrorRetryManager *FailRetryManager
pluginExecutor plugins.NodeHookExecutor
deviceManagerController *deviceManager.DeviceController
heartBeat *heartbeat.HeartBeat
heartBeatInterval time.Duration
lastSeenPanicCount sync.Map
netConf netconf
sriovHotplugExecutorPool *executor.RateLimitedExecutorPool
vmiExpectations *controller.UIDTrackingControllerExpectations
vmiGlobalStore cache.Store
multipathSocketMonitor *multipathmonitor.MultipathSocketMonitor
cbtHandler *CBTHandler
}
var getCgroupManager = func(vmi *v1.VirtualMachineInstance, host string, hypervisorNodeInfo hypervisor.HypervisorNodeInformation, allowEmulation bool) (cgroup.Manager, error) {
return cgroup.NewManagerFromVM(vmi, host, hypervisorNodeInfo.GetHypervisorDevice(), allowEmulation)
}
func NewVirtualMachineController(
recorder record.EventRecorder,
clientset kubecli.KubevirtClient,
nodeStore cache.Store,
host string,
virtPrivateDir string,
kubeletPodsDir string,
launcherClients launcherclients.LauncherClientsManager,
vmiInformer cache.SharedIndexInformer,
vmiGlobalStore cache.Store,
domainInformer cache.SharedInformer,
maxDevices int,
clusterConfig *virtconfig.ClusterConfig,
podIsolationDetector isolation.PodIsolationDetector,
migrationProxy migrationproxy.ProxyManager,
downwardMetricsManager downwardMetricsManager,
capabilities *libvirtxml.Caps,
hostCpuModel string,
netConf netconf,
netStat netstat,
cbtHandler *CBTHandler,
pluginStore cache.Store,
pluginExecutor plugins.NodeHookExecutor,
) (*VirtualMachineController, error) {
queue := workqueue.NewTypedRateLimitingQueueWithConfig[string](
workqueue.DefaultTypedControllerRateLimiter[string](),
workqueue.TypedRateLimitingQueueConfig[string]{Name: "virt-handler-vm"},
)
logger := log.Log.With("controller", "vm")
hypervisorName := clusterConfig.GetHypervisor().Name
baseCtrl, err := NewBaseController(
logger,
host,
recorder,
clientset,
queue,
vmiInformer,
domainInformer,
clusterConfig,
podIsolationDetector,
launcherClients,
migrationProxy,
"/proc/%d/root/var/run",
netStat,
hypervisor.NewHypervisorNodeInformation(hypervisorName),
hypervisor.GetVirtRuntime(podIsolationDetector, hypervisorName),
pluginStore,
)
if err != nil {
return nil, err
}
containerDiskState := filepath.Join(virtPrivateDir, "container-disk-mount-state")
if err := os.MkdirAll(containerDiskState, 0700); err != nil {
return nil, err
}
hotplugState := filepath.Join(virtPrivateDir, "hotplug-volume-mount-state")
if err := os.MkdirAll(hotplugState, 0700); err != nil {
return nil, err
}
c := &VirtualMachineController{
BaseController: baseCtrl,
capabilities: capabilities,
clientset: clientset,
containerDiskMounter: containerdisk.NewMounter(podIsolationDetector, containerDiskState, clusterConfig),
downwardMetricsManager: downwardMetricsManager,
hotplugVolumeMounter: hotplugvolume.NewVolumeMounter(hotplugState, kubeletPodsDir, host),
hostCpuModel: hostCpuModel,
ioErrorRetryManager: NewFailRetryManager("io-error-retry", 10*time.Second, 3*time.Minute, 30*time.Second),
heartBeatInterval: 1 * time.Minute,
netConf: netConf,
sriovHotplugExecutorPool: executor.NewRateLimitedExecutorPool(executor.NewExponentialLimitedBackoffCreator()),
vmiExpectations: controller.NewUIDTrackingControllerExpectations(controller.NewControllerExpectations()),
vmiGlobalStore: vmiGlobalStore,
multipathSocketMonitor: multipathmonitor.NewMultipathSocketMonitor(),
cbtHandler: cbtHandler,
pluginExecutor: pluginExecutor,
}
_, err = vmiInformer.AddEventHandler(cache.ResourceEventHandlerFuncs{
AddFunc: c.addDeleteFunc,
DeleteFunc: c.addDeleteFunc,
UpdateFunc: c.updateFunc,
})
if err != nil {
return nil, err
}
_, err = domainInformer.AddEventHandler(cache.ResourceEventHandlerFuncs{
AddFunc: c.addDomainFunc,
DeleteFunc: c.deleteDomainFunc,
UpdateFunc: c.updateDomainFunc,
})
if err != nil {
return nil, err
}
permissions := "rw"
if cgroups.IsCgroup2UnifiedMode() {
// Need 'rwm' permissions otherwise ebpf filtering program attached by runc
// will deny probing the device file with 'access' syscall. That in turn
// will lead to virtqemud failure on VM startup.
// This has been fixed upstream:
// https://github.com/opencontainers/runc/pull/2796
// but the workaround is still needed to support previous versions without
// the patch.
permissions = "rwm"
}
c.deviceManagerController = deviceManager.NewDeviceController(
c.host,
maxDevices,
permissions,
deviceManager.PermanentHostDevicePlugins(c.hypervisorNodeInfo.GetHypervisorDevice(), maxDevices, permissions),
clusterConfig,
nodeStore)
c.heartBeat = heartbeat.NewHeartBeat(clientset.CoreV1(), c.deviceManagerController, clusterConfig, host)
return c, nil
}
func (c *VirtualMachineController) Run(threadiness int, stopCh chan struct{}) {
defer c.queue.ShutDown()
c.logger.Info("Starting virt-handler vms controller.")
go c.deviceManagerController.Run(stopCh)
go c.downwardMetricsManager.Run(stopCh)
cache.WaitForCacheSync(stopCh, c.hasSynced)
// queue keys for previous Domains on the host that no longer exist
// in the cache. This ensures we perform local cleanup of deleted VMs.
for _, domain := range c.domainStore.List() {
d := domain.(*api.Domain)
vmiRef := v1.NewVMIReferenceWithUUID(
d.ObjectMeta.Namespace,
d.ObjectMeta.Name,
d.Spec.Metadata.KubeVirt.UID)
key := controller.VirtualMachineInstanceKey(vmiRef)
_, exists, _ := c.vmiStore.GetByKey(key)
if !exists {
c.queue.Add(key)
}
}
c.multipathSocketMonitor.Run()
heartBeatDone := c.heartBeat.Run(c.heartBeatInterval, stopCh)
go c.ioErrorRetryManager.Run(stopCh)
// Start the actual work
for i := 0; i < threadiness; i++ {
go wait.Until(c.runWorker, time.Second, stopCh)
}
<-heartBeatDone
<-stopCh
c.multipathSocketMonitor.Close()
c.logger.Info("Stopping virt-handler vms controller.")
}
func (c *VirtualMachineController) runWorker() {
for c.Execute() {
}
}
func (c *VirtualMachineController) Execute() bool {
key, quit := c.queue.Get()
if quit {
return false
}
defer c.queue.Done(key)
if err := c.execute(key); err != nil {
c.logger.Reason(err).Infof("re-enqueuing VirtualMachineInstance %v", key)
c.queue.AddRateLimited(key)
} else {
c.logger.V(4).Infof("processed VirtualMachineInstance %v", key)
c.queue.Forget(key)
}
return true
}
func (c *VirtualMachineController) execute(key string) error {
vmi, vmiExists, err := c.getVMIFromCache(key)
if err != nil {
return err
}
if !vmiExists {
// the vmiInformer probably has to catch up to the domainInformer
// which already sees the vmi, so let's fetch it from the global
// vmi informer to make sure the vmi has actually been deleted
c.logger.V(4).Infof("fetching vmi for key %v from the global informer", key)
obj, exists, err := c.vmiGlobalStore.GetByKey(key)
if err != nil {
return err
}
if exists {
vmi = obj.(*v1.VirtualMachineInstance)
}
vmiExists = exists
}
if !vmiExists {
c.vmiExpectations.DeleteExpectations(key)
} else if !c.vmiExpectations.SatisfiedExpectations(key) {
return nil
}
domain, domainExists, domainCachedUID, err := c.getDomainFromCache(key)
if err != nil {
return err
}
c.logger.Object(vmi).V(4).Infof("domain exists %v", domainExists)
if !vmiExists && string(domainCachedUID) != "" {
// it's possible to discover the UID from cache even if the domain
// doesn't technically exist anymore
vmi.UID = domainCachedUID
c.logger.Object(vmi).Infof("Using cached UID for vmi found in domain cache")
}
// As a last effort, if the UID still can't be determined attempt
// to retrieve it from the ghost record
if string(vmi.UID) == "" {
uid := virtcache.GhostRecordGlobalStore.LastKnownUID(key)
if uid != "" {
c.logger.Object(vmi).V(3).Infof("ghost record cache provided %s as UID", uid)
vmi.UID = uid
}
}
if vmiExists && domainExists && domain.Spec.Metadata.KubeVirt.UID != vmi.UID {
oldVMI := v1.NewVMIReferenceFromNameWithNS(vmi.Namespace, vmi.Name)
oldVMI.UID = domain.Spec.Metadata.KubeVirt.UID
expired, initialized, err := c.launcherClients.IsLauncherClientUnresponsive(oldVMI)
if err != nil {
return err
}
// If we found an outdated domain which is also not alive anymore, clean up
if !initialized {
c.queue.AddAfter(controller.VirtualMachineInstanceKey(vmi), time.Second*1)
return nil
} else if expired {
c.logger.Object(oldVMI).Infof("Detected stale vmi %s that still needs cleanup before new vmi %s with identical name/namespace can be processed", oldVMI.UID, vmi.UID)
err = c.processVmCleanup(oldVMI)
if err != nil {
return err
}
// Make sure we re-enqueue the key to ensure this new VMI is processed
// after the stale domain is removed
c.queue.AddAfter(controller.VirtualMachineInstanceKey(vmi), time.Second*5)
}
return nil
}
if domainExists &&
(domainMigrated(domain) || domain.DeletionTimestamp != nil) {
c.logger.Object(vmi).V(4).Info("detected orphan vmi")
return c.deleteVM(vmi)
}
if migrations.IsMigrating(vmi) && (vmi.Status.Phase == v1.Failed) {
c.logger.V(1).Infof("cleaning up VMI key %v as migration is in progress and the vmi is failed", key)
err = c.processVmCleanup(vmi)
if err != nil {
return err
}
}
if vmi.DeletionTimestamp == nil && isMigrationInProgress(vmi, domain) {
c.logger.V(4).Infof("ignoring key %v as migration is in progress", key)
return nil
}
if vmiExists && !c.isVMIOwnedByNode(vmi) {
c.logger.Object(vmi).V(4).Info("ignoring vmi as it is not owned by this node")
return nil
}
if vmiExists && vmi.IsMigrationSource() {
c.logger.Object(vmi).V(4).Info("ignoring vmi as it is a migration source")
return nil
}
err = c.sync(key,
vmi.DeepCopy(),
vmiExists,
domain,
domainExists)
_, localExists, _ := c.getVMIFromCache(key)
if !localExists {
metrics.ResetVMISync(key)
}
return err
}
type vmiIrrecoverableError struct {
msg string
}
func (e *vmiIrrecoverableError) Error() string { return e.msg }
func formatIrrecoverableErrorMessage(domain *api.Domain) string {
msg := "unknown reason"
if domainPausedFailedPostCopy(domain) {
msg = "VMI is irrecoverable due to failed post-copy migration"
}
return msg
}
func domainPausedFailedPostCopy(domain *api.Domain) bool {
return domain != nil && domain.Status.Status == api.Paused && domain.Status.Reason == api.ReasonPausedPostcopyFailed
}
// teardownNetwork performs network cache cleanup for a specific VMI.
func (c *VirtualMachineController) teardownNetwork(vmi *v1.VirtualMachineInstance) {
if string(vmi.UID) == "" {
return
}
if err := c.netConf.Teardown(vmi); err != nil {
c.logger.Reason(err).Errorf("failed to delete VMI Network cache files: %s", err.Error())
}
c.netStat.Teardown(vmi)
}
func canUpdateToMounted(currentPhase v1.VolumePhase) bool {
return currentPhase == v1.VolumeBound || currentPhase == v1.VolumePending || currentPhase == v1.HotplugVolumeAttachedToNode
}
func canUpdateToUnmounted(currentPhase v1.VolumePhase) bool {
return currentPhase == v1.VolumeReady || currentPhase == v1.HotplugVolumeMounted || currentPhase == v1.HotplugVolumeAttachedToNode
}
func (c *VirtualMachineController) generateEventsForVolumeStatusChange(vmi *v1.VirtualMachineInstance, newStatusMap map[string]v1.VolumeStatus) {
newStatusMapCopy := make(map[string]v1.VolumeStatus)
for k, v := range newStatusMap {
newStatusMapCopy[k] = v
}
for _, oldStatus := range vmi.Status.VolumeStatus {
newStatus, ok := newStatusMap[oldStatus.Name]
if !ok {
// status got removed
c.recorder.Event(vmi, k8sv1.EventTypeNormal, VolumeUnplugged, fmt.Sprintf("Volume %s has been unplugged", oldStatus.Name))
continue
}
if newStatus.Phase != oldStatus.Phase {
c.recorder.Event(vmi, k8sv1.EventTypeNormal, newStatus.Reason, newStatus.Message)
}
delete(newStatusMapCopy, newStatus.Name)
}
// Send events for any new statuses.
for _, v := range newStatusMapCopy {
c.recorder.Event(vmi, k8sv1.EventTypeNormal, v.Reason, v.Message)
}
}
func (c *VirtualMachineController) updateHotplugVolumeStatus(vmi *v1.VirtualMachineInstance, volumeStatus v1.VolumeStatus, specVolumeMap map[string]struct{}) (v1.VolumeStatus, bool) {
needsRefresh := false
if volumeStatus.Target == "" {
needsRefresh = true
mounted, err := c.hotplugVolumeMounter.IsMounted(vmi, volumeStatus.Name, volumeStatus.HotplugVolume.AttachPodUID)
if err != nil {
c.logger.Object(vmi).Errorf("error occurred while checking if volume is mounted: %v", err)
}
if mounted {
if _, ok := specVolumeMap[volumeStatus.Name]; ok && canUpdateToMounted(volumeStatus.Phase) {
log.DefaultLogger().Infof("Marking volume %s as mounted in pod, it can now be attached", volumeStatus.Name)
// mounted, and still in spec, and in phase we can change, update status to mounted.
volumeStatus.Phase = v1.HotplugVolumeMounted
volumeStatus.Message = fmt.Sprintf("Volume %s has been mounted in virt-launcher pod", volumeStatus.Name)
volumeStatus.Reason = VolumeMountedToPodReason
}
} else {
// Not mounted, check if the volume is in the spec, if not update status
if _, ok := specVolumeMap[volumeStatus.Name]; !ok && canUpdateToUnmounted(volumeStatus.Phase) {
log.DefaultLogger().Infof("Marking volume %s as unmounted from pod, it can now be detached", volumeStatus.Name)
// Not mounted.
volumeStatus.Phase = v1.HotplugVolumeUnMounted
volumeStatus.Message = fmt.Sprintf("Volume %s has been unmounted from virt-launcher pod", volumeStatus.Name)
volumeStatus.Reason = VolumeUnMountedFromPodReason
}
}
} else {
// Successfully attached to VM.
volumeStatus.Phase = v1.VolumeReady
volumeStatus.Message = fmt.Sprintf("Successfully attach hotplugged volume %s to VM", volumeStatus.Name)
volumeStatus.Reason = VolumeReadyReason
}
return volumeStatus, needsRefresh
}
func needToComputeChecksums(vmi *v1.VirtualMachineInstance) bool {
containerDisks := map[string]*v1.Volume{}
for _, volume := range vmi.Spec.Volumes {
if volume.VolumeSource.ContainerDisk != nil {
containerDisks[volume.Name] = &volume
}
}
for i := range vmi.Status.VolumeStatus {
_, isContainerDisk := containerDisks[vmi.Status.VolumeStatus[i].Name]
if !isContainerDisk {
continue
}
if vmi.Status.VolumeStatus[i].ContainerDiskVolume == nil ||
vmi.Status.VolumeStatus[i].ContainerDiskVolume.Checksum == 0 {
return true
}
}
if util.HasKernelBootContainerImage(vmi) {
if vmi.Status.KernelBootStatus == nil {
return true
}
kernelBootContainer := vmi.Spec.Domain.Firmware.KernelBoot.Container
if kernelBootContainer.KernelPath != "" &&
(vmi.Status.KernelBootStatus.KernelInfo == nil ||
vmi.Status.KernelBootStatus.KernelInfo.Checksum == 0) {
return true
}
if kernelBootContainer.InitrdPath != "" &&
(vmi.Status.KernelBootStatus.InitrdInfo == nil ||
vmi.Status.KernelBootStatus.InitrdInfo.Checksum == 0) {
return true
}
}
return false
}
// updateChecksumInfo is kept for compatibility with older virt-handlers
// that validate checksum calculations in vmi.status. This validation was
// removed in PR #14021, but we had to keep the checksum calculations for upgrades.
// Once we're sure old handlers won't interrupt upgrades, this can be removed.
func (c *VirtualMachineController) updateChecksumInfo(vmi *v1.VirtualMachineInstance, syncError error) error {
// If the imageVolume feature gate is enabled, upgrade support isn't required,
// and we can skip the checksum calculation. By the time the feature gate is GA,
// the checksum calculation should be removed.
if syncError != nil || vmi.DeletionTimestamp != nil || !needToComputeChecksums(vmi) || c.clusterConfig.ImageVolumeEnabled() {
return nil
}
diskChecksums, err := c.containerDiskMounter.ComputeChecksums(vmi)
if goerror.Is(err, containerdisk.ErrDiskContainerGone) {
c.logger.Errorf("cannot compute checksums as containerdisk/kernelboot containers seem to have been terminated")
return nil
}
if err != nil {
return err
}
// containerdisks
for i := range vmi.Status.VolumeStatus {
checksum, exists := diskChecksums.ContainerDiskChecksums[vmi.Status.VolumeStatus[i].Name]
if !exists {
// not a containerdisk
continue
}
vmi.Status.VolumeStatus[i].ContainerDiskVolume = &v1.ContainerDiskInfo{
Checksum: checksum,
}
}
// kernelboot
if util.HasKernelBootContainerImage(vmi) {
vmi.Status.KernelBootStatus = &v1.KernelBootStatus{}
if diskChecksums.KernelBootChecksum.Kernel != nil {
vmi.Status.KernelBootStatus.KernelInfo = &v1.KernelInfo{
Checksum: *diskChecksums.KernelBootChecksum.Kernel,
}
}
if diskChecksums.KernelBootChecksum.Initrd != nil {
vmi.Status.KernelBootStatus.InitrdInfo = &v1.InitrdInfo{
Checksum: *diskChecksums.KernelBootChecksum.Initrd,
}
}
}
return nil
}
func (c *VirtualMachineController) updateVolumeStatusesFromDomain(vmi *v1.VirtualMachineInstance, domain *api.Domain) bool {
// The return value is only used by unit tests
hasHotplug := false
if len(vmi.Status.VolumeStatus) == 0 {
return false
}
diskDeviceMap := make(map[string]string)
if domain != nil {
for _, disk := range domain.Spec.Devices.Disks {
// don't care about empty cdroms
if disk.Source.File != "" || disk.Source.Dev != "" {
diskDeviceMap[disk.Alias.GetName()] = disk.Target.Device
}
}
}
specVolumeMap := make(map[string]struct{})
for _, volume := range vmi.Spec.Volumes {
specVolumeMap[volume.Name] = struct{}{}
}
for _, utilityVolume := range vmi.Spec.UtilityVolumes {
specVolumeMap[utilityVolume.Name] = struct{}{}
}
newStatusMap := make(map[string]v1.VolumeStatus)
var newStatuses []v1.VolumeStatus
needsRefresh := false
for _, volumeStatus := range vmi.Status.VolumeStatus {
tmpNeedsRefresh := false
// relying on the fact that target will be "" if not in the map
// see updateHotplugVolumeStatus
volumeStatus.Target = diskDeviceMap[volumeStatus.Name]
if volumeStatus.HotplugVolume != nil {
hasHotplug = true
volumeStatus, tmpNeedsRefresh = c.updateHotplugVolumeStatus(vmi, volumeStatus, specVolumeMap)
needsRefresh = needsRefresh || tmpNeedsRefresh
}
if volumeStatus.MemoryDumpVolume != nil {
volumeStatus, tmpNeedsRefresh = c.updateMemoryDumpInfo(vmi, volumeStatus, domain)
needsRefresh = needsRefresh || tmpNeedsRefresh
}
newStatuses = append(newStatuses, volumeStatus)
newStatusMap[volumeStatus.Name] = volumeStatus
}
sort.SliceStable(newStatuses, func(i, j int) bool {
return strings.Compare(newStatuses[i].Name, newStatuses[j].Name) == -1
})
if needsRefresh {
c.queue.AddAfter(controller.VirtualMachineInstanceKey(vmi), time.Second)
}
c.generateEventsForVolumeStatusChange(vmi, newStatusMap)
vmi.Status.VolumeStatus = newStatuses
return hasHotplug
}
func (c *VirtualMachineController) updateGuestInfoFromDomain(vmi *v1.VirtualMachineInstance, domain *api.Domain) {
if domain == nil || domain.Status.OSInfo.Name == "" || vmi.Status.GuestOSInfo.Name == domain.Status.OSInfo.Name {
return
}
vmi.Status.GuestOSInfo.Name = domain.Status.OSInfo.Name
vmi.Status.GuestOSInfo.Version = domain.Status.OSInfo.Version
vmi.Status.GuestOSInfo.KernelRelease = domain.Status.OSInfo.KernelRelease
vmi.Status.GuestOSInfo.PrettyName = domain.Status.OSInfo.PrettyName
vmi.Status.GuestOSInfo.VersionID = domain.Status.OSInfo.VersionId
vmi.Status.GuestOSInfo.KernelVersion = domain.Status.OSInfo.KernelVersion
vmi.Status.GuestOSInfo.Machine = domain.Status.OSInfo.Machine
vmi.Status.GuestOSInfo.ID = domain.Status.OSInfo.Id
}
func (c *VirtualMachineController) updateAccessCredentialConditions(vmi *v1.VirtualMachineInstance, domain *api.Domain, condManager *controller.VirtualMachineInstanceConditionManager) {
if domain == nil || domain.Spec.Metadata.KubeVirt.AccessCredential == nil {
return
}
message := domain.Spec.Metadata.KubeVirt.AccessCredential.Message
status := k8sv1.ConditionFalse
if domain.Spec.Metadata.KubeVirt.AccessCredential.Succeeded {
status = k8sv1.ConditionTrue
}
add := false
condition := condManager.GetCondition(vmi, v1.VirtualMachineInstanceAccessCredentialsSynchronized)
if condition == nil {
add = true
} else if condition.Status != status || condition.Message != message {
// if not as expected, remove, then add.
condManager.RemoveCondition(vmi, v1.VirtualMachineInstanceAccessCredentialsSynchronized)
add = true
}
if add {
newCondition := v1.VirtualMachineInstanceCondition{
Type: v1.VirtualMachineInstanceAccessCredentialsSynchronized,
LastTransitionTime: metav1.Now(),
Status: status,
Message: message,
}
vmi.Status.Conditions = append(vmi.Status.Conditions, newCondition)
if status == k8sv1.ConditionTrue {
eventMessage := "Access credentials sync successful."
if message != "" {
eventMessage = fmt.Sprintf("Access credentials sync successful: %s", message)
}
c.recorder.Event(vmi, k8sv1.EventTypeNormal, v1.AccessCredentialsSyncSuccess.String(), eventMessage)
} else {
c.recorder.Event(vmi, k8sv1.EventTypeWarning, v1.AccessCredentialsSyncFailed.String(),
fmt.Sprintf("Access credentials sync failed: %s", message),
)
}
}
}
func (c *VirtualMachineController) updateLiveMigrationConditions(vmi *v1.VirtualMachineInstance, condManager *controller.VirtualMachineInstanceConditionManager) {
// Calculate whether the VM is migratable
liveMigrationCondition, isBlockMigration := c.calculateLiveMigrationCondition(vmi)
if !condManager.HasCondition(vmi, v1.VirtualMachineInstanceIsMigratable) {
vmi.Status.Conditions = append(vmi.Status.Conditions, *liveMigrationCondition)
} else {
cond := condManager.GetCondition(vmi, v1.VirtualMachineInstanceIsMigratable)
if !equality.Semantic.DeepEqual(cond, liveMigrationCondition) {
condManager.RemoveCondition(vmi, v1.VirtualMachineInstanceIsMigratable)
vmi.Status.Conditions = append(vmi.Status.Conditions, *liveMigrationCondition)
}
}
// Set VMI Migration Method
if isBlockMigration {
vmi.Status.MigrationMethod = v1.BlockMigration
} else {
vmi.Status.MigrationMethod = v1.LiveMigration
}
storageLiveMigCond := c.calculateLiveStorageMigrationCondition(vmi)
condManager.UpdateCondition(vmi, storageLiveMigCond)
evictable := migrations.VMIMigratableOnEviction(c.clusterConfig, vmi)
if evictable && liveMigrationCondition.Status == k8sv1.ConditionFalse {
c.recorder.Eventf(vmi, k8sv1.EventTypeWarning, v1.Migrated.String(), "EvictionStrategy is set but vmi is not migratable; %s", liveMigrationCondition.Message)
}
}
func guestAgentConnected(domain *api.Domain) bool {
if domain != nil {
for _, channel := range domain.Spec.Devices.Channels {
if channel.Target != nil && channel.Target.Name == "org.qemu.guest_agent.0" &&
channel.Target.State == "connected" {
return true
}
}
}
return false
}
func (c *VirtualMachineController) updateGuestAgentConditions(vmi *v1.VirtualMachineInstance, channelConnected bool, condManager *controller.VirtualMachineInstanceConditionManager) error {
// Update the condition when GA is connected
switch {
case channelConnected && !condManager.HasCondition(vmi, v1.VirtualMachineInstanceAgentConnected):
agentCondition := v1.VirtualMachineInstanceCondition{
Type: v1.VirtualMachineInstanceAgentConnected,
LastProbeTime: metav1.Now(),
Status: k8sv1.ConditionTrue,
}
vmi.Status.Conditions = append(vmi.Status.Conditions, agentCondition)
case !channelConnected:
condManager.RemoveCondition(vmi, v1.VirtualMachineInstanceAgentConnected)
}
if condManager.HasCondition(vmi, v1.VirtualMachineInstanceAgentConnected) {
client, err := c.launcherClients.GetLauncherClient(vmi)
if err != nil {
return err
}
guestInfo, err := client.GetGuestInfo()
if err != nil {
return err
}
var supported = false
var reason = ""
// For current versions, virt-launcher's supported commands will always contain data.
// For backwards compatibility: during upgrade from a previous version of KubeVirt,
// virt-launcher might not provide any supported commands. If the list of supported
// commands is empty, fall back to previous behavior.
if len(guestInfo.SupportedCommands) > 0 {
supported, reason = isGuestAgentSupported(vmi, guestInfo.SupportedCommands)
c.logger.V(3).Object(vmi).Info(reason)
} else {
for _, version := range c.clusterConfig.GetSupportedAgentVersions() {
supported = supported || regexp.MustCompile(version).MatchString(guestInfo.GAVersion)
}
if !supported {
reason = fmt.Sprintf("Guest agent version '%s' is not supported", guestInfo.GAVersion)
}
}
if !supported {
if !condManager.HasCondition(vmi, v1.VirtualMachineInstanceUnsupportedAgent) {
agentCondition := v1.VirtualMachineInstanceCondition{
Type: v1.VirtualMachineInstanceUnsupportedAgent,
LastProbeTime: metav1.Now(),
Status: k8sv1.ConditionTrue,
Reason: reason,
}
vmi.Status.Conditions = append(vmi.Status.Conditions, agentCondition)
}
} else {
condManager.RemoveCondition(vmi, v1.VirtualMachineInstanceUnsupportedAgent)
}
}
return nil
}
func (c *VirtualMachineController) updatePausedConditions(vmi *v1.VirtualMachineInstance, domain *api.Domain, condManager *controller.VirtualMachineInstanceConditionManager) {
// Update paused condition in case VMI was paused / unpaused
if domain != nil && domain.Status.Status == api.Paused {
if !condManager.HasCondition(vmi, v1.VirtualMachineInstancePaused) {
c.calculatePausedCondition(vmi, domain.Status.Reason)
}
} else if condManager.HasCondition(vmi, v1.VirtualMachineInstancePaused) {
c.logger.Object(vmi).V(3).Info("Removing paused condition")
condManager.RemoveCondition(vmi, v1.VirtualMachineInstancePaused)
}
}
func (c *VirtualMachineController) updateSoftwareEmulationCondition(vmi *v1.VirtualMachineInstance, domain *api.Domain, condManager *controller.VirtualMachineInstanceConditionManager) {
if !c.clusterConfig.CrossArchitectureVirtualizationEnabled() {
return
}
isCrossArchEmulation := domain != nil && domain.Spec.Type == "qemu" &&
vmi.Spec.Architecture != "" && vmi.Spec.Architecture != runtime.GOARCH
if isCrossArchEmulation {
if !condManager.HasCondition(vmi, v1.VirtualMachineInstanceSoftwareEmulation) {
now := metav1.Now()
vmi.Status.Conditions = append(vmi.Status.Conditions, v1.VirtualMachineInstanceCondition{
Type: v1.VirtualMachineInstanceSoftwareEmulation,
Status: k8sv1.ConditionTrue,
LastProbeTime: now,
LastTransitionTime: now,
Reason: "CrossArchitectureEmulation",
Message: fmt.Sprintf("VM running with software emulation (host=%s, guest=%s)", runtime.GOARCH, vmi.Spec.Architecture),
})
}
} else if condManager.HasCondition(vmi, v1.VirtualMachineInstanceSoftwareEmulation) {
condManager.RemoveCondition(vmi, v1.VirtualMachineInstanceSoftwareEmulation)
}
}
func dumpTargetFile(vmiName, volName string) string {
targetFileName := fmt.Sprintf("%s-%s-%s.memory.dump", vmiName, volName, time.Now().Format("20060102-150405"))
return targetFileName
}
func (c *VirtualMachineController) updateMemoryDumpInfo(vmi *v1.VirtualMachineInstance, volumeStatus v1.VolumeStatus, domain *api.Domain) (v1.VolumeStatus, bool) {
needsRefresh := false
switch volumeStatus.Phase {
case v1.HotplugVolumeMounted:
needsRefresh = true
c.logger.Object(vmi).V(3).Infof("Memory dump volume %s attached, marking it in progress", volumeStatus.Name)
volumeStatus.Phase = v1.MemoryDumpVolumeInProgress
volumeStatus.Message = fmt.Sprintf("Memory dump Volume %s is attached, getting memory dump", volumeStatus.Name)
volumeStatus.Reason = VolumeMountedToPodReason
volumeStatus.MemoryDumpVolume.TargetFileName = dumpTargetFile(vmi.Name, volumeStatus.Name)
case v1.MemoryDumpVolumeInProgress:
var memoryDumpMetadata *api.MemoryDumpMetadata
if domain != nil {
memoryDumpMetadata = domain.Spec.Metadata.KubeVirt.MemoryDump
}
if memoryDumpMetadata == nil || memoryDumpMetadata.FileName != volumeStatus.MemoryDumpVolume.TargetFileName {
// memory dump wasnt triggered yet
return volumeStatus, needsRefresh
}
needsRefresh = true
if memoryDumpMetadata.StartTimestamp != nil {
volumeStatus.MemoryDumpVolume.StartTimestamp = memoryDumpMetadata.StartTimestamp
}
if memoryDumpMetadata.EndTimestamp != nil && memoryDumpMetadata.Failed {
c.logger.Object(vmi).Errorf("Memory dump to pvc %s failed: %v", volumeStatus.Name, memoryDumpMetadata.FailureReason)
volumeStatus.Message = fmt.Sprintf("Memory dump to pvc %s failed: %v", volumeStatus.Name, memoryDumpMetadata.FailureReason)
volumeStatus.Phase = v1.MemoryDumpVolumeFailed
volumeStatus.MemoryDumpVolume.EndTimestamp = memoryDumpMetadata.EndTimestamp
} else if memoryDumpMetadata.Completed {
c.logger.Object(vmi).V(3).Infof("Marking memory dump to volume %s has completed", volumeStatus.Name)
volumeStatus.Phase = v1.MemoryDumpVolumeCompleted
volumeStatus.Message = fmt.Sprintf("Memory dump to Volume %s has completed successfully", volumeStatus.Name)
volumeStatus.Reason = VolumeReadyReason
volumeStatus.MemoryDumpVolume.EndTimestamp = memoryDumpMetadata.EndTimestamp
}
}
return volumeStatus, needsRefresh
}
func (c *VirtualMachineController) updateFSFreezeStatus(vmi *v1.VirtualMachineInstance, domain *api.Domain) {
if domain == nil || domain.Status.FSFreezeStatus.Status == "" {
return
}
if domain.Status.FSFreezeStatus.Status == api.FSThawed {
vmi.Status.FSFreezeStatus = ""
} else {
vmi.Status.FSFreezeStatus = domain.Status.FSFreezeStatus.Status
}
}
func IsoGuestVolumePath(namespace, name string, volume *v1.Volume) string {
const basepath = "/var/run"
switch {
case volume.CloudInitNoCloud != nil:
return filepath.Join(basepath, "kubevirt-ephemeral-disks", "cloud-init-data", namespace, name, "noCloud.iso")
case volume.CloudInitConfigDrive != nil:
return filepath.Join(basepath, "kubevirt-ephemeral-disks", "cloud-init-data", namespace, name, "configdrive.iso")
case volume.ConfigMap != nil:
return config.GetConfigMapDiskPath(volume.Name)
case volume.DownwardAPI != nil:
return config.GetDownwardAPIDiskPath(volume.Name)
case volume.Secret != nil:
return config.GetSecretDiskPath(volume.Name)
case volume.ServiceAccount != nil:
return config.GetServiceAccountDiskPath()
case volume.Sysprep != nil:
return config.GetSysprepDiskPath(volume.Name)
default:
return ""
}
}
func (c *VirtualMachineController) updateIsoSizeStatus(vmi *v1.VirtualMachineInstance) {
var podUID string
if vmi.Status.Phase != v1.Running {
return
}
for k, v := range vmi.Status.ActivePods {
if v == vmi.Status.NodeName {
podUID = string(k)
break
}
}
if podUID == "" {
log.DefaultLogger().Warningf("failed to find pod UID for VMI %s", vmi.Name)
return
}
volumes := make(map[string]v1.Volume)
for _, volume := range vmi.Spec.Volumes {
volumes[volume.Name] = volume
}
for _, disk := range vmi.Spec.Domain.Devices.Disks {
volume, ok := volumes[disk.Name]
if !ok {
log.DefaultLogger().Warningf("No matching volume with name %s found", disk.Name)
continue
}
volPath := IsoGuestVolumePath(vmi.Namespace, vmi.Name, &volume)
if volPath == "" {
continue
}