-
Notifications
You must be signed in to change notification settings - Fork 116
/
scheduler.go
1682 lines (1451 loc) · 41.6 KB
/
scheduler.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
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Copyright 2019 Sorint.lab
//
// 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 runservice
import (
"bytes"
"context"
"encoding/json"
"fmt"
"net/http"
"strconv"
"time"
"github.com/rs/zerolog"
"github.com/sorintlab/errors"
"agola.io/agola/internal/objectstorage"
"agola.io/agola/internal/runconfig"
rsapi "agola.io/agola/internal/services/runservice/api"
"agola.io/agola/internal/services/runservice/common"
"agola.io/agola/internal/services/runservice/store"
"agola.io/agola/internal/sqlg/sql"
"agola.io/agola/internal/util"
"agola.io/agola/services/runservice/types"
)
const (
changeGroupCompactorInterval = 1 * time.Minute
cacheCleanerInterval = 1 * 24 * time.Hour
workspaceCleanerInterval = 1 * 24 * time.Hour
logCleanerInterval = 1 * 24 * time.Hour
defaultExecutorNotAliveInterval = 60 * time.Second
changeGroupMinDuration = 5 * time.Minute
)
func taskMatchesParentDependCondition(rt *types.RunTask, r *types.Run, rc *types.RunConfig) bool {
rct := rc.Tasks[rt.ID]
parents := runconfig.GetParents(rc.Tasks, rct)
matchedNum := 0
for _, p := range parents {
matched := false
rp := r.Tasks[p.ID]
conds := runconfig.GetParentDependConditions(rct, p)
for _, cond := range conds {
switch cond {
case types.RunConfigTaskDependConditionOnSuccess:
if rp.Status == types.RunTaskStatusSuccess {
matched = true
}
case types.RunConfigTaskDependConditionOnFailure:
if rp.Status == types.RunTaskStatusFailed {
matched = true
}
case types.RunConfigTaskDependConditionOnSkipped:
if rp.Status == types.RunTaskStatusSkipped {
matched = true
}
}
}
if matched {
matchedNum++
}
}
return len(parents) == matchedNum
}
func advanceRunTasks(log zerolog.Logger, curRun *types.Run, rc *types.RunConfig, scheduledExecutorTasks []*types.ExecutorTask) (*types.Run, error) {
log.Debug().Msgf("run: %s", util.Dump(curRun))
log.Debug().Msgf("rc: %s", util.Dump(rc))
// take a deepcopy of r so we do logic only on fixed status and not affected by current changes (due to random map iteration)
newRun := curRun.DeepCopy()
if newRun.Stop {
// if the run is set to stop, skip all not running tasks
for _, rt := range newRun.Tasks {
isScheduled := false
for _, et := range scheduledExecutorTasks {
if rt.ID == et.RunTaskID {
isScheduled = true
}
}
if isScheduled {
continue
}
if rt.Status == types.RunTaskStatusNotStarted {
rt.Status = types.RunTaskStatusSkipped
}
}
}
// handle root tasks
for _, rt := range newRun.Tasks {
if rt.Skip {
continue
}
if rt.Status != types.RunTaskStatusNotStarted {
continue
}
rct := rc.Tasks[rt.ID]
parents := runconfig.GetParents(rc.Tasks, rct)
if len(parents) > 0 {
continue
}
// cancel task if the run has a result set and is not yet scheduled
if curRun.Result.IsSet() {
isScheduled := false
for _, et := range scheduledExecutorTasks {
if rt.ID == et.RunTaskID {
isScheduled = true
}
}
if isScheduled {
continue
}
if rt.Status == types.RunTaskStatusNotStarted {
rt.Status = types.RunTaskStatusCancelled
}
}
}
// handle all tasks
// TODO(sgotti) process tasks by their level (from 0) so we'll calculate the
// final state in just one loop. Currently the call to this function won't
// calculate a deterministic final state since we could process the tasks in
// any order
for _, rt := range newRun.Tasks {
if rt.Skip {
continue
}
if rt.Status != types.RunTaskStatusNotStarted {
continue
}
rct := rc.Tasks[rt.ID]
parents := runconfig.GetParents(rc.Tasks, rct)
finishedParents := 0
for _, p := range parents {
// use current run status to not be affected by previous changes to to random map iteration
rp := curRun.Tasks[p.ID]
if rp.Status.IsFinished() && rp.ArchivesFetchFinished() {
finishedParents++
}
}
allParentsFinished := finishedParents == len(parents)
// if all parents are finished check if the task could be executed or be skipped
if allParentsFinished {
matched := taskMatchesParentDependCondition(rt, curRun, rc)
// if all parents are matched then we can start it, otherwise we mark the step to be skipped
skip := !matched
if skip {
rt.Status = types.RunTaskStatusSkipped
continue
}
// now that the task can run set it to waiting approval if needed
if rct.NeedsApproval && !rt.WaitingApproval && !rt.Approved {
rt.WaitingApproval = true
}
}
}
return newRun, nil
}
func getTasksToRun(log zerolog.Logger, r *types.Run, rc *types.RunConfig) ([]*types.RunTask, error) {
log.Debug().Msgf("run: %s", util.Dump(r))
log.Debug().Msgf("rc: %s", util.Dump(rc))
tasksToRun := []*types.RunTask{}
// get tasks that can be executed
for _, rt := range r.Tasks {
if rt.Skip {
continue
}
if rt.Status != types.RunTaskStatusNotStarted {
continue
}
rct := rc.Tasks[rt.ID]
parents := runconfig.GetParents(rc.Tasks, rct)
finishedParents := 0
for _, p := range parents {
rp := r.Tasks[p.ID]
if rp.Status.IsFinished() && rp.ArchivesFetchFinished() {
finishedParents++
}
}
allParentsFinished := finishedParents == len(parents)
if allParentsFinished {
// TODO(sgotti) This could be removed when advanceRunTasks will calculate the
// state in a deterministic a complete way in one loop (see the related TODO)
if !taskMatchesParentDependCondition(rt, r, rc) {
continue
}
// Run only if approved (when needs approval)
if !rct.NeedsApproval || (rct.NeedsApproval && rt.Approved) {
tasksToRun = append(tasksToRun, rt)
}
}
}
return tasksToRun, nil
}
func (s *Runservice) submitRunTasks(ctx context.Context, r *types.Run, rc *types.RunConfig, tasks []*types.RunTask) error {
s.log.Debug().Msgf("tasksToRun: %s", util.Dump(tasks))
for _, rt := range tasks {
rct := rc.Tasks[rt.ID]
// check that the executorTask for this runTask id wasn't already scheduled
var executorTask *types.ExecutorTask
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
executorTask, err = s.d.GetExecutorTaskByRunTask(tx, r.ID, rt.ID)
if err != nil {
return errors.WithStack(err)
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
if executorTask != nil {
continue
}
executor, err := s.chooseExecutor(ctx, rct)
if err != nil {
return errors.WithStack(err)
}
if executor == nil {
s.log.Warn().Msgf("cannot choose an executor")
return nil
}
// check again that the executorTask for this runTask id wasn't already scheduled
// if not existing, save and submit it
var shouldSend bool
err = s.d.Do(ctx, func(tx *sql.Tx) error {
curExecutorTask, err := s.d.GetExecutorTaskByRunTask(tx, r.ID, rt.ID)
if err != nil {
return errors.WithStack(err)
}
if curExecutorTask != nil {
return nil
}
executorTask = common.GenExecutorTask(tx, r, rt, rc, executor)
s.log.Debug().Msgf("et: %s", util.Dump(executorTask))
if err := s.d.InsertExecutorTask(tx, executorTask); err != nil {
return errors.WithStack(err)
}
shouldSend = true
return nil
})
if err != nil {
return errors.WithStack(err)
}
if shouldSend {
if err := s.sendExecutorTask(ctx, executorTask); err != nil {
return errors.WithStack(err)
}
}
}
return nil
}
// chooseExecutor chooses the executor to schedule the task on. Now it's a very simple/dumb selection
// TODO(sgotti) improve this to use executor statistic, labels (arch type) etc...
func (s *Runservice) chooseExecutor(ctx context.Context, rct *types.RunConfigTask) (*types.Executor, error) {
var executors []*types.Executor
executorTasksCount := map[string]int{}
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
executors, err = s.d.GetExecutors(tx)
if err != nil {
return errors.WithStack(err)
}
// TODO(sgotti) implement a db method that just returns the count
for _, executor := range executors {
executorTasks, err := s.d.GetExecutorTasksByExecutor(tx, executor.ExecutorID)
if err != nil {
return errors.WithStack(err)
}
executorTasksCount[executor.ExecutorID] = len(executorTasks)
}
return nil
})
if err != nil {
return nil, errors.WithStack(err)
}
return chooseExecutor(executors, executorTasksCount, rct), nil
}
func chooseExecutor(executors []*types.Executor, executorTasksCount map[string]int, rct *types.RunConfigTask) *types.Executor {
requiresPrivilegedContainers := false
for _, c := range rct.Runtime.Containers {
if c.Privileged {
requiresPrivilegedContainers = true
break
}
}
for _, e := range executors {
if time.Since(e.UpdateTime) > defaultExecutorNotAliveInterval {
continue
}
// skip executor provileged containers are required but not allowed
if requiresPrivilegedContainers && !e.AllowPrivilegedContainers {
continue
}
// if arch is not defined use any executor arch
if rct.Runtime.Arch != "" {
hasArch := false
for _, arch := range e.Archs {
if arch == rct.Runtime.Arch {
hasArch = true
}
}
if !hasArch {
continue
}
}
if e.ActiveTasksLimit != 0 {
// will be 0 when executorTasksCount[e.ExecutorID] doesn't exist
activeTasks := executorTasksCount[e.ExecutorID]
if e.ActiveTasks > activeTasks {
activeTasks = e.ActiveTasks
}
// calculate the active tasks by the max between the current scheduled
// tasks in the store and the executor reported tasks
if activeTasks >= e.ActiveTasksLimit {
continue
}
}
return e
}
return nil
}
// sendExecutorTask sends executor task to executor, if this fails the executor
// will periodically fetch the executortask anyway
func (s *Runservice) sendExecutorTask(ctx context.Context, et *types.ExecutorTask) error {
var executor *types.Executor
var r *types.Run
var rc *types.RunConfig
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
executor, err = s.d.GetExecutorByExecutorID(tx, et.ExecutorID)
if err != nil {
return errors.WithStack(err)
}
r, err = s.d.GetRun(tx, et.RunID)
if err != nil {
return errors.WithStack(err)
}
if r == nil {
return errors.Errorf("run with id %q doesn't exist", et.RunID)
}
rc, err = s.d.GetRunConfig(tx, r.RunConfigID)
if err != nil {
return errors.WithStack(err)
}
if rc == nil {
return errors.Errorf("runconfig with id %q doesn't exist", r.RunConfigID)
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
if executor == nil {
s.log.Warn().Msgf("executor with id %q doesn't exist", et.ExecutorID)
return nil
}
rt, ok := r.Tasks[et.RunTaskID]
if !ok {
return errors.Errorf("no such run task with id %s for run %s", et.RunTaskID, r.ID)
}
// generate ExecutorTaskSpecData
etSpecData := common.GenExecutorTaskSpecData(r, rt, rc)
etRes := rsapi.GenExecutorTaskResponse(et, etSpecData)
etResj, err := json.Marshal(etRes)
if err != nil {
return errors.WithStack(err)
}
req, err := http.NewRequestWithContext(ctx, "POST", executor.ListenURL+"/api/v1alpha/executor", bytes.NewReader(etResj))
if err != nil {
return errors.WithStack(err)
}
if s.c.ExecutorAPIToken != "" {
req.Header.Set("Authorization", "token "+s.c.ExecutorAPIToken)
}
resp, err := http.DefaultClient.Do(req)
if err != nil {
return errors.WithStack(err)
}
if resp.StatusCode != http.StatusOK {
return errors.Errorf("received http status: %d", resp.StatusCode)
}
return nil
}
func (s *Runservice) compactChangeGroupsLoop(ctx context.Context) {
for {
if err := s.compactChangeGroups(ctx); err != nil {
s.log.Err(err).Send()
}
sleepCh := time.NewTimer(changeGroupCompactorInterval).C
select {
case <-ctx.Done():
return
case <-sleepCh:
}
}
}
func (s *Runservice) compactChangeGroups(ctx context.Context) error {
// TODO(sgotti) do this in batches of N changegroups and/or filter by update time directly in the query
err := s.d.Do(ctx, func(tx *sql.Tx) error {
changeGroups, err := s.d.GetChangeGroups(tx)
if err != nil {
return errors.WithStack(err)
}
for _, changeGroup := range changeGroups {
if time.Since(changeGroup.UpdateTime) < changeGroupMinDuration {
continue
}
if err := s.d.DeleteChangeGroup(tx, changeGroup.ID); err != nil {
return errors.WithStack(err)
}
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
return nil
}
func (s *Runservice) scheduleRun(ctx context.Context, runID string) error {
// we use multiple transactions to split the logic in multiple steps and
// rely on optimistic object locking.
// We could probably also use a single transaction without many issues
// (perhaps more serializable transaction errors) since it should be quite
// faster.
var shouldSubmitRunTasks bool
var etsToSend []*types.ExecutorTask
var r *types.Run
var rc *types.RunConfig
var scheduledExecutorTasks []*types.ExecutorTask
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
r, err = s.d.GetRun(tx, runID)
if err != nil {
return errors.WithStack(err)
}
if r == nil {
return errors.Errorf("run with id %q doesn't exist", runID)
}
rc, err = s.d.GetRunConfig(tx, r.RunConfigID)
if err != nil {
return errors.WithStack(err)
}
if rc == nil {
return errors.Errorf("runconfig with id %q doesn't exist", r.RunConfigID)
}
// the real source of active tasks is the number of executor tasks in the db
// we can't rely on RunTask.Status since it's only updated when receiveing
// updated from the executor so it could be in a NotStarted state but have an
// executor tasks scheduled and running
scheduledExecutorTasks, err = s.d.GetExecutorTasksByRun(tx, r.ID)
if err != nil {
return errors.WithStack(err)
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
s.log.Debug().Msgf("r: %s", util.Dump(r))
prevPhase := r.Phase
prevResult := r.Result
if err := advanceRun(s.log, r, rc, scheduledExecutorTasks); err != nil {
return errors.WithStack(err)
}
curRunRevision := r.Revision
curScheduledExecutorTasksRevisions := map[string]uint64{}
for _, et := range scheduledExecutorTasks {
curScheduledExecutorTasksRevisions[et.ID] = et.Revision
}
err = s.d.Do(ctx, func(tx *sql.Tx) error {
// if the run is set to stop, stop all active tasks
if r.Stop {
for _, et := range scheduledExecutorTasks {
et.Stop = true
et.TxID = tx.ID()
et.Revision = curScheduledExecutorTasksRevisions[et.ID]
if err := s.d.UpdateExecutorTask(tx, et); err != nil {
return errors.WithStack(err)
}
etsToSend = append(etsToSend, et)
}
}
// advance tasks
if r.Phase == types.RunPhaseRunning {
r, err = advanceRunTasks(s.log, r, rc, scheduledExecutorTasks)
if err != nil {
return errors.WithStack(err)
}
shouldSubmitRunTasks = true
}
r.TxID = tx.ID()
r.Revision = curRunRevision
if err := s.d.UpdateRun(tx, r); err != nil {
return errors.WithStack(err)
}
// detect changes to phase and result and set related events
if prevPhase != r.Phase || prevResult != r.Result {
runEvent, err := common.NewRunEvent(s.d, tx, r, rc, types.RunPhaseChanged)
if err != nil {
return errors.WithStack(err)
}
if err := s.d.InsertRunEvent(tx, runEvent); err != nil {
return errors.WithStack(err)
}
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
for _, et := range etsToSend {
if err := s.sendExecutorTask(ctx, et); err != nil {
return errors.WithStack(err)
}
}
if shouldSubmitRunTasks {
tasksToRun, err := getTasksToRun(s.log, r, rc)
if err != nil {
return errors.WithStack(err)
}
if err := s.submitRunTasks(ctx, r, rc, tasksToRun); err != nil {
return errors.WithStack(err)
}
}
return nil
}
// advanceRun updates the run result and phase. It must be the unique function that
// should update them.
func advanceRun(log zerolog.Logger, r *types.Run, rc *types.RunConfig, scheduledExecutorTasks []*types.ExecutorTask) error {
log.Debug().Msgf("run: %s", util.Dump(r))
hasScheduledTasks := len(scheduledExecutorTasks) > 0
// fail run if a task is failed
if !r.Result.IsSet() && r.Phase == types.RunPhaseRunning {
for _, rt := range r.Tasks {
rct, ok := rc.Tasks[rt.ID]
log.Debug().Msgf("rct: %s", util.Dump(rct))
if !ok {
return errors.Errorf("no such run config task with id %s for run config %s", rt.ID, rc.ID)
}
if rt.Status == types.RunTaskStatusFailed {
if !rct.IgnoreFailure {
log.Debug().Msgf("marking run %q as failed is task %q is failed", r.ID, rt.ID)
r.Result = types.RunResultFailed
break
}
}
}
}
// see if run could be marked as success
if !r.Result.IsSet() && r.Phase == types.RunPhaseRunning {
finished := true
for _, rt := range r.Tasks {
if !rt.Status.IsFinished() {
finished = false
}
}
if finished {
r.Result = types.RunResultSuccess
return nil
}
}
// if run is set to stop set result as stopped
if !r.Result.IsSet() && r.Phase == types.RunPhaseRunning {
if r.Stop {
r.Result = types.RunResultStopped
}
}
// if the run has a result defined AND all tasks are finished AND there're no executor tasks scheduled we can mark
// the run phase as finished
if r.Result.IsSet() {
finished := true
for _, rt := range r.Tasks {
if !rt.Status.IsFinished() {
finished = false
}
}
if finished && !r.Phase.IsFinished() {
if !hasScheduledTasks {
r.ChangePhase(types.RunPhaseFinished)
}
}
// if the run is finished AND there're no executor tasks scheduled we can mark
// all not started runtasks' fetch phases (setup step, logs and archives) as finished
if r.Phase.IsFinished() {
for _, rt := range r.Tasks {
log.Debug().Msgf("rt: %s", util.Dump(rt))
if rt.Status == types.RunTaskStatusNotStarted {
rt.SetupStep.LogPhase = types.RunTaskFetchPhaseFinished
for _, s := range rt.Steps {
s.LogPhase = types.RunTaskFetchPhaseFinished
}
for i := range rt.WorkspaceArchivesPhase {
rt.WorkspaceArchivesPhase[i] = types.RunTaskFetchPhaseFinished
}
}
}
}
}
return nil
}
func (s *Runservice) handleExecutorTaskUpdate(ctx context.Context, executorTaskID string) error {
var r *types.Run
err := s.d.Do(ctx, func(tx *sql.Tx) error {
et, err := s.d.GetExecutorTask(tx, executorTaskID)
if err != nil {
return errors.WithStack(err)
}
if et == nil {
return errors.Errorf("executor task with id %q doesn't exist", executorTaskID)
}
r, err = s.d.GetRun(tx, et.RunID)
if err != nil {
return errors.WithStack(err)
}
if r == nil {
return errors.Errorf("run with id %q doesn't exist", et.RunID)
}
if err := s.updateRunTaskStatus(et, r); err != nil {
return errors.WithStack(err)
}
if err = s.d.UpdateRun(tx, r); err != nil {
return errors.WithStack(err)
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
return s.scheduleRun(ctx, r.ID)
}
func (s *Runservice) updateRunTaskStatus(et *types.ExecutorTask, r *types.Run) error {
s.log.Debug().Msgf("et: %s", util.Dump(et))
rt, ok := r.Tasks[et.RunTaskID]
if !ok {
return errors.Errorf("no such run task with id %s for run %s", et.RunTaskID, r.ID)
}
rt.StartTime = et.StartTime
rt.EndTime = et.EndTime
wrongstatus := false
switch et.Phase {
case types.ExecutorTaskPhaseNotStarted:
if rt.Status != types.RunTaskStatusNotStarted {
wrongstatus = true
}
case types.ExecutorTaskPhaseCancelled:
if rt.Status != types.RunTaskStatusCancelled &&
rt.Status != types.RunTaskStatusNotStarted {
wrongstatus = true
}
case types.ExecutorTaskPhaseRunning:
if rt.Status != types.RunTaskStatusRunning &&
rt.Status != types.RunTaskStatusNotStarted {
wrongstatus = true
}
case types.ExecutorTaskPhaseStopped:
if rt.Status != types.RunTaskStatusStopped &&
rt.Status != types.RunTaskStatusNotStarted &&
rt.Status != types.RunTaskStatusRunning {
wrongstatus = true
}
case types.ExecutorTaskPhaseSuccess:
if rt.Status != types.RunTaskStatusSuccess &&
rt.Status != types.RunTaskStatusNotStarted &&
rt.Status != types.RunTaskStatusRunning {
wrongstatus = true
}
case types.ExecutorTaskPhaseFailed:
if rt.Status != types.RunTaskStatusFailed &&
rt.Status != types.RunTaskStatusNotStarted &&
rt.Status != types.RunTaskStatusRunning {
wrongstatus = true
}
}
if wrongstatus {
s.log.Warn().Msgf("ignoring wrong executor task %q status: %q, rt status: %q", et.ID, et.Phase, rt.Status)
return nil
}
switch et.Phase {
case types.ExecutorTaskPhaseNotStarted:
rt.Status = types.RunTaskStatusNotStarted
case types.ExecutorTaskPhaseCancelled:
rt.Status = types.RunTaskStatusCancelled
case types.ExecutorTaskPhaseRunning:
rt.Status = types.RunTaskStatusRunning
case types.ExecutorTaskPhaseStopped:
rt.Status = types.RunTaskStatusStopped
case types.ExecutorTaskPhaseSuccess:
rt.Status = types.RunTaskStatusSuccess
case types.ExecutorTaskPhaseFailed:
rt.Status = types.RunTaskStatusFailed
}
rt.Timedout = et.Timedout
rt.SetupStep.Phase = et.SetupStep.Phase
rt.SetupStep.StartTime = et.SetupStep.StartTime
rt.SetupStep.EndTime = et.SetupStep.EndTime
for i, s := range et.Steps {
rt.Steps[i].Phase = s.Phase
rt.Steps[i].ExitStatus = s.ExitStatus
rt.Steps[i].StartTime = s.StartTime
rt.Steps[i].EndTime = s.EndTime
}
return nil
}
func (s *Runservice) executorTaskUpdateHandler(ctx context.Context, c <-chan string) {
for {
select {
case <-ctx.Done():
return
case etID := <-c:
go func() {
if err := s.handleExecutorTaskUpdate(ctx, etID); err != nil {
// TODO(sgotti) improve logging to not return "run modified errors" since
// they are normal
s.log.Warn().Err(err).Send()
}
}()
}
}
}
func (s *Runservice) executorTasksCleanerLoop(ctx context.Context) {
for {
s.log.Debug().Msgf("executorTasksCleaner")
if err := s.executorTasksCleaner(ctx); err != nil {
s.log.Err(err).Send()
}
sleepCh := time.NewTimer(1 * time.Second).C
select {
case <-ctx.Done():
return
case <-sleepCh:
}
}
}
func (s *Runservice) executorTasksCleaner(ctx context.Context) error {
// TODO(sgotti) use paged List or get only the executor tasks id
var executorTasks []*types.ExecutorTask
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
executorTasks, err = s.d.GetExecutorTasks(tx)
if err != nil {
return errors.WithStack(err)
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
for _, et := range executorTasks {
if err := s.executorTaskCleaner(ctx, et.ID); err != nil {
s.log.Err(err).Send()
}
}
return nil
}
func (s *Runservice) executorTaskCleaner(ctx context.Context, executorTaskID string) error {
var shouldSend bool
var et *types.ExecutorTask
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
et, err = s.d.GetExecutorTask(tx, executorTaskID)
if err != nil {
return errors.WithStack(err)
}
s.log.Debug().Msgf("et: %s", util.Dump(et))
if et == nil {
return nil
}
if et.Phase.IsFinished() {
r, err := s.d.GetRun(tx, et.RunID)
if err != nil {
return errors.WithStack(err)
}
if r == nil {
// run doesn't exists, remove executor task
s.log.Warn().Msgf("deleting executor task %q since run %q doesn't exist", et.ID, et.RunID)
if err := s.d.DeleteExecutorTask(tx, et.ID); err != nil {
s.log.Err(err).Send()
return errors.WithStack(err)
}
return nil
}
if r.Phase.IsFinished() {
// if the run is finished mark the executor tasks to stop
if !et.Stop {
et.Stop = true
if err := s.d.UpdateExecutorTask(tx, et); err != nil {
return errors.WithStack(err)
}
shouldSend = true
}
}
}
if !et.Phase.IsFinished() {
// if the executor doesn't exists anymore mark the not finished executor tasks as failed
executor, err := s.d.GetExecutorByExecutorID(tx, et.ExecutorID)
if err != nil {
return errors.WithStack(err)
}
if executor == nil {
s.log.Warn().Msgf("executor with id %q doesn't exist. marking executor task %q as failed", et.ExecutorID, et.ID)
et.FailError = "executor deleted"
et.Phase = types.ExecutorTaskPhaseFailed
et.EndTime = util.TimeP(time.Now())
for _, s := range et.Steps {
if s.Phase == types.ExecutorTaskPhaseRunning {
s.Phase = types.ExecutorTaskPhaseFailed
s.EndTime = util.TimeP(time.Now())
}
}
if err := s.d.UpdateExecutorTask(tx, et); err != nil {
return errors.WithStack(err)
}
}
}
return nil
})
if err != nil {
return errors.WithStack(err)
}
if shouldSend {
if err := s.sendExecutorTask(ctx, et); err != nil {
return errors.WithStack(err)
}
}
return nil
}
func (s *Runservice) runTasksUpdaterLoop(ctx context.Context) {
for {
s.log.Debug().Msgf("runTasksUpdater")
if err := s.runTasksUpdater(ctx); err != nil {
s.log.Err(err).Send()
}
sleepCh := time.NewTimer(10 * time.Second).C
select {
case <-ctx.Done():
return
case <-sleepCh:
}
}
}
func (s *Runservice) runTasksUpdater(ctx context.Context) error {
s.log.Debug().Msgf("runTasksUpdater")
l := s.lf.NewLock(common.TaskUpdaterLockKey)
if err := l.Lock(ctx); err != nil {
return errors.Wrap(err, "failed to acquire task updater lock")
}
defer func() { _ = l.Unlock() }()
// TODO(sgotti) use paged List or get only the executor tasks id
var executorTasks []*types.ExecutorTask
err := s.d.Do(ctx, func(tx *sql.Tx) error {
var err error
executorTasks, err = s.d.GetExecutorTasks(tx)
if err != nil {
return errors.WithStack(err)