forked from actions/actions-runner-controller
-
Notifications
You must be signed in to change notification settings - Fork 0
/
runner_controller.go
978 lines (831 loc) · 30.4 KB
/
runner_controller.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
/*
Copyright 2020 The actions-runner-controller 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 controllers
import (
"context"
"errors"
"fmt"
"strings"
"time"
gogithub "github.com/google/go-github/v33/github"
"github.com/summerwind/actions-runner-controller/hash"
"k8s.io/apimachinery/pkg/util/wait"
"github.com/go-logr/logr"
kerrors "k8s.io/apimachinery/pkg/api/errors"
"k8s.io/apimachinery/pkg/runtime"
"k8s.io/client-go/tools/record"
ctrl "sigs.k8s.io/controller-runtime"
"sigs.k8s.io/controller-runtime/pkg/client"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"github.com/summerwind/actions-runner-controller/api/v1alpha1"
"github.com/summerwind/actions-runner-controller/github"
)
const (
containerName = "runner"
finalizerName = "runner.actions.summerwind.dev"
LabelKeyPodTemplateHash = "pod-template-hash"
retryDelayOnGitHubAPIRateLimitError = 30 * time.Second
// This is an annotation internal to actions-runner-controller and can change in backward-incompatible ways
annotationKeyRegistrationOnly = "actions-runner-controller/registration-only"
)
// RunnerReconciler reconciles a Runner object
type RunnerReconciler struct {
client.Client
Log logr.Logger
Recorder record.EventRecorder
Scheme *runtime.Scheme
GitHubClient *github.Client
RunnerImage string
DockerImage string
Name string
RegistrationRecheckInterval time.Duration
RegistrationRecheckJitter time.Duration
}
// +kubebuilder:rbac:groups=actions.summerwind.dev,resources=runners,verbs=get;list;watch;create;update;patch;delete
// +kubebuilder:rbac:groups=actions.summerwind.dev,resources=runners/finalizers,verbs=get;list;watch;create;update;patch;delete
// +kubebuilder:rbac:groups=actions.summerwind.dev,resources=runners/status,verbs=get;update;patch
// +kubebuilder:rbac:groups=core,resources=pods,verbs=get;list;watch;create;update;patch;delete
// +kubebuilder:rbac:groups=core,resources=pods/finalizers,verbs=get;list;watch;create;update;patch;delete
// +kubebuilder:rbac:groups=core,resources=events,verbs=create;patch
func (r *RunnerReconciler) Reconcile(req ctrl.Request) (ctrl.Result, error) {
ctx := context.Background()
log := r.Log.WithValues("runner", req.NamespacedName)
var runner v1alpha1.Runner
if err := r.Get(ctx, req.NamespacedName, &runner); err != nil {
return ctrl.Result{}, client.IgnoreNotFound(err)
}
err := runner.Validate()
if err != nil {
log.Info("Failed to validate runner spec", "error", err.Error())
return ctrl.Result{}, nil
}
if runner.ObjectMeta.DeletionTimestamp.IsZero() {
finalizers, added := addFinalizer(runner.ObjectMeta.Finalizers)
if added {
newRunner := runner.DeepCopy()
newRunner.ObjectMeta.Finalizers = finalizers
if err := r.Update(ctx, newRunner); err != nil {
log.Error(err, "Failed to update runner")
return ctrl.Result{}, err
}
return ctrl.Result{}, nil
}
} else {
finalizers, removed := removeFinalizer(runner.ObjectMeta.Finalizers)
if removed {
if len(runner.Status.Registration.Token) > 0 {
ok, err := r.unregisterRunner(ctx, runner.Spec.Enterprise, runner.Spec.Organization, runner.Spec.Repository, runner.Name)
if err != nil {
if errors.Is(err, &gogithub.RateLimitError{}) {
// We log the underlying error when we failed calling GitHub API to list or unregisters,
// or the runner is still busy.
log.Error(
err,
fmt.Sprintf(
"Failed to unregister runner due to GitHub API rate limits. Delaying retry for %s to avoid excessive GitHub API calls",
retryDelayOnGitHubAPIRateLimitError,
),
)
return ctrl.Result{RequeueAfter: retryDelayOnGitHubAPIRateLimitError}, err
}
return ctrl.Result{}, err
}
if !ok {
log.V(1).Info("Runner no longer exists on GitHub")
}
} else {
log.V(1).Info("Runner was never registered on GitHub")
}
newRunner := runner.DeepCopy()
newRunner.ObjectMeta.Finalizers = finalizers
if err := r.Patch(ctx, newRunner, client.MergeFrom(&runner)); err != nil {
log.Error(err, "Failed to update runner for finalizer removal")
return ctrl.Result{}, err
}
log.Info("Removed runner from GitHub", "repository", runner.Spec.Repository, "organization", runner.Spec.Organization)
}
return ctrl.Result{}, nil
}
registrationOnly := metav1.HasAnnotation(runner.ObjectMeta, annotationKeyRegistrationOnly)
if registrationOnly && runner.Status.Phase != "" {
// At this point we are sure that the registration-only runner has successfully configured and
// is of `offline` status, because we set runner.Status.Phase to that of the runner pod only after
// successful registration.
var pod corev1.Pod
if err := r.Get(ctx, req.NamespacedName, &pod); err != nil {
if !kerrors.IsNotFound(err) {
log.Info(fmt.Sprintf("Retrying soon as we failed to get registration-only runner pod: %v", err))
return ctrl.Result{Requeue: true}, nil
}
} else if err := r.Delete(ctx, &pod); err != nil {
if !kerrors.IsNotFound(err) {
log.Info(fmt.Sprintf("Retrying soon as we failed to delete registration-only runner pod: %v", err))
return ctrl.Result{Requeue: true}, nil
}
}
log.Info("Successfully deleted egistration-only runner pod to free node and cluster resource")
// Return here to not recreate the deleted pod, because recreating it is the waste of cluster and node resource,
// and also defeats the original purpose of scale-from/to-zero we're trying to implement by using the registration-only runner.
return ctrl.Result{}, nil
}
var pod corev1.Pod
if err := r.Get(ctx, req.NamespacedName, &pod); err != nil {
if !kerrors.IsNotFound(err) {
return ctrl.Result{}, err
}
if updated, err := r.updateRegistrationToken(ctx, runner); err != nil {
return ctrl.Result{}, err
} else if updated {
return ctrl.Result{Requeue: true}, nil
}
newPod, err := r.newPod(runner)
if err != nil {
log.Error(err, "Could not create pod")
return ctrl.Result{}, err
}
if err := r.Create(ctx, &newPod); err != nil {
if kerrors.IsAlreadyExists(err) {
// Gracefully handle pod-already-exists errors due to informer cache delay.
// Without this we got a few errors like the below on new runner pod:
// 2021-03-16T00:23:10.116Z ERROR controller-runtime.controller Reconciler error {"controller": "runner-controller", "request": "default/example-runnerdeploy-b2g2g-j4mcp", "error": "pods \"example-runnerdeploy-b2g2g-j4mcp\" already exists"}
log.Info(
"Failed to create pod due to AlreadyExists error. Probably this pod has been already created in previous reconcilation but is still not in the informer cache. Will retry on pod created. If it doesn't repeat, there's no problem",
)
return ctrl.Result{}, nil
}
log.Error(err, "Failed to create pod resource")
return ctrl.Result{}, err
}
r.Recorder.Event(&runner, corev1.EventTypeNormal, "PodCreated", fmt.Sprintf("Created pod '%s'", newPod.Name))
log.Info("Created runner pod", "repository", runner.Spec.Repository)
} else {
if !pod.ObjectMeta.DeletionTimestamp.IsZero() {
deletionTimeout := 1 * time.Minute
currentTime := time.Now()
deletionDidTimeout := currentTime.Sub(pod.DeletionTimestamp.Add(deletionTimeout)) > 0
if deletionDidTimeout {
log.Info(
fmt.Sprintf("Failed to delete pod within %s. ", deletionTimeout)+
"This is typically the case when a Kubernetes node became unreachable "+
"and the kube controller started evicting nodes. Forcefully deleting the pod to not get stuck.",
"podDeletionTimestamp", pod.DeletionTimestamp,
"currentTime", currentTime,
"configuredDeletionTimeout", deletionTimeout,
)
var force int64 = 0
// forcefully delete runner as we would otherwise get stuck if the node stays unreachable
if err := r.Delete(ctx, &pod, &client.DeleteOptions{GracePeriodSeconds: &force}); err != nil {
// probably
if !kerrors.IsNotFound(err) {
log.Error(err, "Failed to forcefully delete pod resource ...")
return ctrl.Result{}, err
}
// forceful deletion finally succeeded
return ctrl.Result{Requeue: true}, nil
}
r.Recorder.Event(&runner, corev1.EventTypeNormal, "PodDeleted", fmt.Sprintf("Forcefully deleted pod '%s'", pod.Name))
log.Info("Forcefully deleted runner pod", "repository", runner.Spec.Repository)
// give kube manager a little time to forcefully delete the stuck pod
return ctrl.Result{RequeueAfter: 3 * time.Second}, err
} else {
return ctrl.Result{}, err
}
}
// If pod has ended up succeeded we need to restart it
// Happens e.g. when dind is in runner and run completes
stopped := pod.Status.Phase == corev1.PodSucceeded
if !stopped {
if pod.Status.Phase == corev1.PodRunning {
for _, status := range pod.Status.ContainerStatuses {
if status.Name != containerName {
continue
}
if status.State.Terminated != nil && status.State.Terminated.ExitCode == 0 {
stopped = true
}
}
}
}
restart := stopped
if registrationOnly && stopped {
restart = false
log.Info(
"Observed that registration-only runner for scaling-from-zero has successfully stopped. " +
"Unlike other pods, this one will be recreated only when runner spec changes.",
)
}
if updated, err := r.updateRegistrationToken(ctx, runner); err != nil {
return ctrl.Result{}, err
} else if updated {
return ctrl.Result{Requeue: true}, nil
}
newPod, err := r.newPod(runner)
if err != nil {
log.Error(err, "Could not create pod")
return ctrl.Result{}, err
}
if registrationOnly {
newPod.Spec.Containers[0].Env = append(
newPod.Spec.Containers[0].Env,
corev1.EnvVar{
Name: "RUNNER_REGISTRATION_ONLY",
Value: "true",
},
)
}
var registrationRecheckDelay time.Duration
// all checks done below only decide whether a restart is needed
// if a restart was already decided before, there is no need for the checks
// saving API calls and scary log messages
if !restart {
registrationCheckInterval := time.Minute
if r.RegistrationRecheckInterval > 0 {
registrationCheckInterval = r.RegistrationRecheckInterval
}
// We want to call ListRunners GitHub Actions API only once per runner per minute.
// This if block, in conjunction with:
// return ctrl.Result{RequeueAfter: registrationRecheckDelay}, nil
// achieves that.
if lastCheckTime := runner.Status.LastRegistrationCheckTime; lastCheckTime != nil {
nextCheckTime := lastCheckTime.Add(registrationCheckInterval)
now := time.Now()
// Requeue scheduled by RequeueAfter can happen a bit earlier (like dozens of milliseconds)
// so to avoid excessive, in-effective retry, we heuristically ignore the remaining delay in case it is
// shorter than 1s
requeueAfter := nextCheckTime.Sub(now) - time.Second
if requeueAfter > 0 {
log.Info(
fmt.Sprintf("Skipped registration check because it's deferred until %s. Retrying in %s at latest", nextCheckTime, requeueAfter),
"lastRegistrationCheckTime", lastCheckTime,
"registrationCheckInterval", registrationCheckInterval,
)
// Without RequeueAfter, the controller may not retry on scheduled. Instead, it must wait until the
// next sync period passes, which can be too much later than nextCheckTime.
//
// We need to requeue on this reconcilation even though we have already scheduled the initial
// requeue previously with `return ctrl.Result{RequeueAfter: registrationRecheckDelay}, nil`.
// Apparently, the workqueue used by controller-runtime seems to deduplicate and resets the delay on
// other requeues- so the initial scheduled requeue may have been reset due to requeue on
// spec/status change.
return ctrl.Result{RequeueAfter: requeueAfter}, nil
}
}
notFound := false
offline := false
runnerBusy, err := r.GitHubClient.IsRunnerBusy(ctx, runner.Spec.Enterprise, runner.Spec.Organization, runner.Spec.Repository, runner.Name)
currentTime := time.Now()
if err != nil {
var notFoundException *github.RunnerNotFound
var offlineException *github.RunnerOffline
if errors.As(err, ¬FoundException) {
notFound = true
} else if errors.As(err, &offlineException) {
offline = true
} else {
var e *gogithub.RateLimitError
if errors.As(err, &e) {
// We log the underlying error when we failed calling GitHub API to list or unregisters,
// or the runner is still busy.
log.Error(
err,
fmt.Sprintf(
"Failed to check if runner is busy due to Github API rate limit. Retrying in %s to avoid excessive GitHub API calls",
retryDelayOnGitHubAPIRateLimitError,
),
)
return ctrl.Result{RequeueAfter: retryDelayOnGitHubAPIRateLimitError}, err
}
return ctrl.Result{}, err
}
}
// See the `newPod` function called above for more information
// about when this hash changes.
curHash := pod.Labels[LabelKeyPodTemplateHash]
newHash := newPod.Labels[LabelKeyPodTemplateHash]
if !runnerBusy && curHash != newHash {
restart = true
}
registrationTimeout := 10 * time.Minute
durationAfterRegistrationTimeout := currentTime.Sub(pod.CreationTimestamp.Add(registrationTimeout))
registrationDidTimeout := durationAfterRegistrationTimeout > 0
if notFound {
if registrationDidTimeout {
log.Info(
"Runner failed to register itself to GitHub in timely manner. "+
"Recreating the pod to see if it resolves the issue. "+
"CAUTION: If you see this a lot, you should investigate the root cause. "+
"See https://github.com/summerwind/actions-runner-controller/issues/288",
"podCreationTimestamp", pod.CreationTimestamp,
"currentTime", currentTime,
"configuredRegistrationTimeout", registrationTimeout,
)
restart = true
} else {
log.V(1).Info(
"Runner pod exists but we failed to check if runner is busy. Apparently it still needs more time.",
"runnerName", runner.Name,
)
}
} else if offline {
if registrationOnly {
log.Info(
"Observed that registration-only runner for scaling-from-zero has successfully been registered.",
"podCreationTimestamp", pod.CreationTimestamp,
"currentTime", currentTime,
"configuredRegistrationTimeout", registrationTimeout,
)
} else if registrationDidTimeout {
log.Info(
"Already existing GitHub runner still appears offline . "+
"Recreating the pod to see if it resolves the issue. "+
"CAUTION: If you see this a lot, you should investigate the root cause. ",
"podCreationTimestamp", pod.CreationTimestamp,
"currentTime", currentTime,
"configuredRegistrationTimeout", registrationTimeout,
)
restart = true
} else {
log.V(1).Info(
"Runner pod exists but the GitHub runner appears to be still offline. Waiting for runner to get online ...",
"runnerName", runner.Name,
)
}
}
if (notFound || (offline && !registrationOnly)) && !registrationDidTimeout {
registrationRecheckJitter := 10 * time.Second
if r.RegistrationRecheckJitter > 0 {
registrationRecheckJitter = r.RegistrationRecheckJitter
}
registrationRecheckDelay = registrationCheckInterval + wait.Jitter(registrationRecheckJitter, 0.1)
}
}
// Don't do anything if there's no need to restart the runner
if !restart {
// This guard enables us to update runner.Status.Phase to `Running` only after
// the runner is registered to GitHub.
if registrationRecheckDelay > 0 {
log.V(1).Info(fmt.Sprintf("Rechecking the runner registration in %s", registrationRecheckDelay))
updated := runner.DeepCopy()
updated.Status.LastRegistrationCheckTime = &metav1.Time{Time: time.Now()}
if err := r.Status().Patch(ctx, updated, client.MergeFrom(&runner)); err != nil {
log.Error(err, "Failed to update runner status for LastRegistrationCheckTime")
return ctrl.Result{}, err
}
return ctrl.Result{RequeueAfter: registrationRecheckDelay}, nil
}
if runner.Status.Phase != string(pod.Status.Phase) {
if pod.Status.Phase == corev1.PodRunning {
// Seeing this message, you can expect the runner to become `Running` soon.
log.Info(
"Runner appears to have registered and running.",
"podCreationTimestamp", pod.CreationTimestamp,
)
}
updated := runner.DeepCopy()
updated.Status.Phase = string(pod.Status.Phase)
updated.Status.Reason = pod.Status.Reason
updated.Status.Message = pod.Status.Message
if err := r.Status().Patch(ctx, updated, client.MergeFrom(&runner)); err != nil {
log.Error(err, "Failed to update runner status for Phase/Reason/Message")
return ctrl.Result{}, err
}
}
return ctrl.Result{}, nil
}
// Delete current pod if recreation is needed
if err := r.Delete(ctx, &pod); err != nil {
log.Error(err, "Failed to delete pod resource")
return ctrl.Result{}, err
}
r.Recorder.Event(&runner, corev1.EventTypeNormal, "PodDeleted", fmt.Sprintf("Deleted pod '%s'", newPod.Name))
log.Info("Deleted runner pod", "repository", runner.Spec.Repository)
}
return ctrl.Result{}, nil
}
func (r *RunnerReconciler) unregisterRunner(ctx context.Context, enterprise, org, repo, name string) (bool, error) {
runners, err := r.GitHubClient.ListRunners(ctx, enterprise, org, repo)
if err != nil {
return false, err
}
id := int64(0)
for _, runner := range runners {
if runner.GetName() == name {
if runner.GetBusy() {
return false, fmt.Errorf("runner is busy")
}
id = runner.GetID()
break
}
}
if id == int64(0) {
return false, nil
}
if err := r.GitHubClient.RemoveRunner(ctx, enterprise, org, repo, id); err != nil {
return false, err
}
return true, nil
}
func (r *RunnerReconciler) updateRegistrationToken(ctx context.Context, runner v1alpha1.Runner) (bool, error) {
if runner.IsRegisterable() {
return false, nil
}
log := r.Log.WithValues("runner", runner.Name)
rt, err := r.GitHubClient.GetRegistrationToken(ctx, runner.Spec.Enterprise, runner.Spec.Organization, runner.Spec.Repository, runner.Name)
if err != nil {
r.Recorder.Event(&runner, corev1.EventTypeWarning, "FailedUpdateRegistrationToken", "Updating registration token failed")
log.Error(err, "Failed to get new registration token")
return false, err
}
updated := runner.DeepCopy()
updated.Status.Registration = v1alpha1.RunnerStatusRegistration{
Organization: runner.Spec.Organization,
Repository: runner.Spec.Repository,
Labels: runner.Spec.Labels,
Token: rt.GetToken(),
ExpiresAt: metav1.NewTime(rt.GetExpiresAt().Time),
}
if err := r.Status().Patch(ctx, updated, client.MergeFrom(&runner)); err != nil {
log.Error(err, "Failed to update runner status for Registration")
return false, err
}
r.Recorder.Event(&runner, corev1.EventTypeNormal, "RegistrationTokenUpdated", "Successfully update registration token")
log.Info("Updated registration token", "repository", runner.Spec.Repository)
return true, nil
}
func (r *RunnerReconciler) newPod(runner v1alpha1.Runner) (corev1.Pod, error) {
var (
privileged bool = true
dockerdInRunner bool = runner.Spec.DockerdWithinRunnerContainer != nil && *runner.Spec.DockerdWithinRunnerContainer
dockerEnabled bool = runner.Spec.DockerEnabled == nil || *runner.Spec.DockerEnabled
ephemeral bool = runner.Spec.Ephemeral == nil || *runner.Spec.Ephemeral
dockerdInRunnerPrivileged bool = dockerdInRunner
)
runnerImage := runner.Spec.Image
if runnerImage == "" {
runnerImage = r.RunnerImage
}
workDir := runner.Spec.WorkDir
if workDir == "" {
workDir = "/runner/_work"
}
runnerImagePullPolicy := runner.Spec.ImagePullPolicy
if runnerImagePullPolicy == "" {
runnerImagePullPolicy = corev1.PullAlways
}
env := []corev1.EnvVar{
{
Name: "RUNNER_NAME",
Value: runner.Name,
},
{
Name: "RUNNER_ORG",
Value: runner.Spec.Organization,
},
{
Name: "RUNNER_REPO",
Value: runner.Spec.Repository,
},
{
Name: "RUNNER_ENTERPRISE",
Value: runner.Spec.Enterprise,
},
{
Name: "RUNNER_LABELS",
Value: strings.Join(runner.Spec.Labels, ","),
},
{
Name: "RUNNER_GROUP",
Value: runner.Spec.Group,
},
{
Name: "RUNNER_TOKEN",
Value: runner.Status.Registration.Token,
},
{
Name: "DOCKERD_IN_RUNNER",
Value: fmt.Sprintf("%v", dockerdInRunner),
},
{
Name: "GITHUB_URL",
Value: r.GitHubClient.GithubBaseURL,
},
{
Name: "RUNNER_WORKDIR",
Value: workDir,
},
{
Name: "RUNNER_EPHEMERAL",
Value: fmt.Sprintf("%v", ephemeral),
},
}
if metav1.HasAnnotation(runner.ObjectMeta, annotationKeyRegistrationOnly) {
env = append(env, corev1.EnvVar{
Name: "RUNNER_REGISTRATION_ONLY",
Value: "true",
},
)
}
env = append(env, runner.Spec.Env...)
labels := map[string]string{}
for k, v := range runner.Labels {
labels[k] = v
}
// This implies that...
//
// (1) We recreate the runner pod whenever the runner has changes in:
// - metadata.labels (excluding "runner-template-hash" added by the parent RunnerReplicaSet
// - metadata.annotations
// - metadata.spec (including image, env, organization, repository, group, and so on)
// - GithubBaseURL setting of the controller (can be configured via GITHUB_ENTERPRISE_URL)
//
// (2) We don't recreate the runner pod when there are changes in:
// - runner.status.registration.token
// - This token expires and changes hourly, but you don't need to recreate the pod due to that.
// It's the opposite.
// An unexpired token is required only when the runner agent is registering itself on launch.
//
// In other words, the registered runner doesn't get invalidated on registration token expiration.
// A registered runner's session and the a registration token seem to have two different and independent
// lifecycles.
//
// See https://github.com/summerwind/actions-runner-controller/issues/143 for more context.
labels[LabelKeyPodTemplateHash] = hash.FNVHashStringObjects(
filterLabels(runner.Labels, LabelKeyRunnerTemplateHash),
runner.Annotations,
runner.Spec,
r.GitHubClient.GithubBaseURL,
)
var seLinuxOptions *corev1.SELinuxOptions
if runner.Spec.SecurityContext != nil {
seLinuxOptions = runner.Spec.SecurityContext.SELinuxOptions
if seLinuxOptions != nil {
privileged = false
dockerdInRunnerPrivileged = false
}
}
pod := corev1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: runner.Name,
Namespace: runner.Namespace,
Labels: labels,
Annotations: runner.Annotations,
},
Spec: corev1.PodSpec{
RestartPolicy: "OnFailure",
Containers: []corev1.Container{
{
Name: containerName,
Image: runnerImage,
ImagePullPolicy: runnerImagePullPolicy,
Env: env,
EnvFrom: runner.Spec.EnvFrom,
SecurityContext: &corev1.SecurityContext{
// Runner need to run privileged if it contains DinD
Privileged: &dockerdInRunnerPrivileged,
},
Resources: runner.Spec.Resources,
},
},
},
}
if mtu := runner.Spec.DockerMTU; mtu != nil && dockerdInRunner {
pod.Spec.Containers[0].Env = append(pod.Spec.Containers[0].Env, []corev1.EnvVar{
{
Name: "MTU",
Value: fmt.Sprintf("%d", *runner.Spec.DockerMTU),
},
}...)
}
if mirror := runner.Spec.DockerRegistryMirror; mirror != nil && dockerdInRunner {
pod.Spec.Containers[0].Env = append(pod.Spec.Containers[0].Env, []corev1.EnvVar{
{
Name: "DOCKER_REGISTRY_MIRROR",
Value: *runner.Spec.DockerRegistryMirror,
},
}...)
}
//
// /runner must be generated on runtime from /runnertmp embedded in the container image.
//
// When you're NOT using dindWithinRunner=true,
// it must also be shared with the dind container as it seems like required to run docker steps.
//
runnerVolumeName := "runner"
runnerVolumeMountPath := "/runner"
runnerVolumeEmptyDir := &corev1.EmptyDirVolumeSource{}
if runner.Spec.VolumeSizeLimit != nil {
runnerVolumeEmptyDir.SizeLimit = runner.Spec.VolumeSizeLimit
}
pod.Spec.Volumes = append(pod.Spec.Volumes,
corev1.Volume{
Name: runnerVolumeName,
VolumeSource: corev1.VolumeSource{
EmptyDir: runnerVolumeEmptyDir,
},
},
)
pod.Spec.Containers[0].VolumeMounts = append(pod.Spec.Containers[0].VolumeMounts,
corev1.VolumeMount{
Name: runnerVolumeName,
MountPath: runnerVolumeMountPath,
},
)
if !dockerdInRunner && dockerEnabled {
pod.Spec.Volumes = append(pod.Spec.Volumes,
corev1.Volume{
Name: "work",
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{},
},
},
corev1.Volume{
Name: "certs-client",
VolumeSource: corev1.VolumeSource{
EmptyDir: &corev1.EmptyDirVolumeSource{},
},
},
)
pod.Spec.Containers[0].VolumeMounts = append(pod.Spec.Containers[0].VolumeMounts,
corev1.VolumeMount{
Name: "work",
MountPath: workDir,
},
corev1.VolumeMount{
Name: "certs-client",
MountPath: "/certs/client",
ReadOnly: true,
},
)
pod.Spec.Containers[0].Env = append(pod.Spec.Containers[0].Env, []corev1.EnvVar{
{
Name: "DOCKER_HOST",
Value: "tcp://localhost:2376",
},
{
Name: "DOCKER_TLS_VERIFY",
Value: "1",
},
{
Name: "DOCKER_CERT_PATH",
Value: "/certs/client",
},
}...)
// Determine the volume mounts assigned to the docker sidecar. In case extra mounts are included in the RunnerSpec, append them to the standard
// set of mounts. See https://github.com/summerwind/actions-runner-controller/issues/435 for context.
dockerVolumeMounts := []corev1.VolumeMount{
{
Name: "work",
MountPath: workDir,
},
{
Name: runnerVolumeName,
MountPath: runnerVolumeMountPath,
},
{
Name: "certs-client",
MountPath: "/certs/client",
},
}
if extraDockerVolumeMounts := runner.Spec.DockerVolumeMounts; extraDockerVolumeMounts != nil {
dockerVolumeMounts = append(dockerVolumeMounts, extraDockerVolumeMounts...)
}
pod.Spec.Containers = append(pod.Spec.Containers, corev1.Container{
Name: "docker",
Image: r.DockerImage,
VolumeMounts: dockerVolumeMounts,
Env: []corev1.EnvVar{
{
Name: "DOCKER_TLS_CERTDIR",
Value: "/certs",
},
},
SecurityContext: &corev1.SecurityContext{
Privileged: &privileged,
SELinuxOptions: seLinuxOptions,
},
Resources: runner.Spec.DockerdContainerResources,
})
if mtu := runner.Spec.DockerMTU; mtu != nil {
pod.Spec.Containers[1].Env = append(pod.Spec.Containers[1].Env, []corev1.EnvVar{
// See https://docs.docker.com/engine/security/rootless/
{
Name: "DOCKERD_ROOTLESS_ROOTLESSKIT_MTU",
Value: fmt.Sprintf("%d", *runner.Spec.DockerMTU),
},
}...)
pod.Spec.Containers[1].Args = append(pod.Spec.Containers[1].Args,
"--mtu",
fmt.Sprintf("%d", *runner.Spec.DockerMTU),
)
}
if mirror := runner.Spec.DockerRegistryMirror; mirror != nil {
pod.Spec.Containers[1].Args = append(pod.Spec.Containers[1].Args,
fmt.Sprintf("--registry-mirror=%s", *runner.Spec.DockerRegistryMirror),
)
}
}
if len(runner.Spec.Containers) != 0 {
pod.Spec.Containers = runner.Spec.Containers
for i := 0; i < len(pod.Spec.Containers); i++ {
if pod.Spec.Containers[i].Name == containerName {
pod.Spec.Containers[i].Env = append(pod.Spec.Containers[i].Env, env...)
}
}
}
if len(runner.Spec.VolumeMounts) != 0 {
pod.Spec.Containers[0].VolumeMounts = append(pod.Spec.Containers[0].VolumeMounts, runner.Spec.VolumeMounts...)
}
if len(runner.Spec.Volumes) != 0 {
pod.Spec.Volumes = append(pod.Spec.Volumes, runner.Spec.Volumes...)
}
if len(runner.Spec.InitContainers) != 0 {
pod.Spec.InitContainers = append(pod.Spec.InitContainers, runner.Spec.InitContainers...)
}
if runner.Spec.NodeSelector != nil {
pod.Spec.NodeSelector = runner.Spec.NodeSelector
}
if runner.Spec.ServiceAccountName != "" {
pod.Spec.ServiceAccountName = runner.Spec.ServiceAccountName
}
if runner.Spec.AutomountServiceAccountToken != nil {
pod.Spec.AutomountServiceAccountToken = runner.Spec.AutomountServiceAccountToken
}
if len(runner.Spec.SidecarContainers) != 0 {
pod.Spec.Containers = append(pod.Spec.Containers, runner.Spec.SidecarContainers...)
}
if runner.Spec.SecurityContext != nil {
pod.Spec.SecurityContext = runner.Spec.SecurityContext
}
if len(runner.Spec.ImagePullSecrets) != 0 {
pod.Spec.ImagePullSecrets = runner.Spec.ImagePullSecrets
}
if runner.Spec.Affinity != nil {
pod.Spec.Affinity = runner.Spec.Affinity
}
if len(runner.Spec.Tolerations) != 0 {
pod.Spec.Tolerations = runner.Spec.Tolerations
}
if len(runner.Spec.EphemeralContainers) != 0 {
pod.Spec.EphemeralContainers = runner.Spec.EphemeralContainers
}
if runner.Spec.TerminationGracePeriodSeconds != nil {
pod.Spec.TerminationGracePeriodSeconds = runner.Spec.TerminationGracePeriodSeconds
}
if len(runner.Spec.HostAliases) != 0 {
pod.Spec.HostAliases = runner.Spec.HostAliases
}
if runner.Spec.RuntimeClassName != nil {
pod.Spec.RuntimeClassName = runner.Spec.RuntimeClassName
}
if err := ctrl.SetControllerReference(&runner, &pod, r.Scheme); err != nil {
return pod, err
}
return pod, nil
}
func (r *RunnerReconciler) SetupWithManager(mgr ctrl.Manager) error {
name := "runner-controller"
if r.Name != "" {
name = r.Name
}
r.Recorder = mgr.GetEventRecorderFor(name)
return ctrl.NewControllerManagedBy(mgr).
For(&v1alpha1.Runner{}).
Owns(&corev1.Pod{}).
Named(name).
Complete(r)
}
func addFinalizer(finalizers []string) ([]string, bool) {
exists := false
for _, name := range finalizers {
if name == finalizerName {
exists = true
}
}
if exists {
return finalizers, false
}
return append(finalizers, finalizerName), true
}
func removeFinalizer(finalizers []string) ([]string, bool) {
removed := false
result := []string{}
for _, name := range finalizers {
if name == finalizerName {
removed = true
continue
}
result = append(result, name)
}
return result, removed
}