forked from hashicorp/nomad
/
eval_endpoint.go
965 lines (831 loc) · 27 KB
/
eval_endpoint.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
// Copyright (c) HashiCorp, Inc.
// SPDX-License-Identifier: MPL-2.0
package nomad
import (
"errors"
"fmt"
"net/http"
"time"
"github.com/armon/go-metrics"
"github.com/hashicorp/go-bexpr"
"github.com/hashicorp/go-hclog"
"github.com/hashicorp/go-memdb"
"github.com/hashicorp/go-multierror"
"github.com/hashicorp/go-version"
"github.com/hernad/nomad/acl"
"github.com/hernad/nomad/nomad/state"
"github.com/hernad/nomad/nomad/state/paginator"
"github.com/hernad/nomad/nomad/structs"
"github.com/hernad/nomad/scheduler"
)
const (
// DefaultDequeueTimeout is used if no dequeue timeout is provided
DefaultDequeueTimeout = time.Second
)
var minVersionEvalDeleteByFilter = version.Must(version.NewVersion("1.4.3"))
// Eval endpoint is used for eval interactions
type Eval struct {
srv *Server
ctx *RPCContext
logger hclog.Logger
}
func NewEvalEndpoint(srv *Server, ctx *RPCContext) *Eval {
return &Eval{srv: srv, ctx: ctx, logger: srv.logger.Named("eval")}
}
// GetEval is used to request information about a specific evaluation
func (e *Eval) GetEval(args *structs.EvalSpecificRequest,
reply *structs.SingleEvalResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
if done, err := e.srv.forward("Eval.GetEval", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricRead, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "get_eval"}, time.Now())
// Check for read-job permissions before performing blocking query.
allowNsOp := acl.NamespaceValidator(acl.NamespaceCapabilityReadJob)
aclObj, err := e.srv.ResolveACL(args)
if err != nil {
return err
} else if !allowNsOp(aclObj, args.RequestNamespace()) {
return structs.ErrPermissionDenied
}
// Setup the blocking query
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, state *state.StateStore) error {
var related []*structs.EvaluationStub
// Look for the eval
eval, err := state.EvalByID(ws, args.EvalID)
if err != nil {
return fmt.Errorf("failed to lookup eval: %v", err)
}
if eval != nil {
// Re-check namespace in case it differs from request.
if !allowNsOp(aclObj, eval.Namespace) {
return structs.ErrPermissionDenied
}
// Lookup related evals if requested.
if args.IncludeRelated {
related, err = state.EvalsRelatedToID(ws, eval.ID)
if err != nil {
return fmt.Errorf("failed to lookup related evals: %v", err)
}
// Use a copy to avoid modifying the original eval.
eval = eval.Copy()
eval.RelatedEvals = related
}
}
// Setup the output.
reply.Eval = eval
if eval != nil {
reply.Index = eval.ModifyIndex
} else {
// Use the last index that affected the evals table
index, err := state.Index("evals")
if err != nil {
return err
}
reply.Index = index
}
// Set the query response
e.srv.setQueryMeta(&reply.QueryMeta)
return nil
}}
return e.srv.blockingRPC(&opts)
}
// Dequeue is used to dequeue a pending evaluation
func (e *Eval) Dequeue(args *structs.EvalDequeueRequest,
reply *structs.EvalDequeueResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Dequeue", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "dequeue"}, time.Now())
// Ensure there is at least one scheduler
if len(args.Schedulers) == 0 {
return fmt.Errorf("dequeue requires at least one scheduler type")
}
// Check that there isn't a scheduler version mismatch
if args.SchedulerVersion != scheduler.SchedulerVersion {
return fmt.Errorf("dequeue disallowed: calling scheduler version is %d; leader version is %d",
args.SchedulerVersion, scheduler.SchedulerVersion)
}
// Ensure there is a default timeout
if args.Timeout <= 0 {
args.Timeout = DefaultDequeueTimeout
}
// If the eval broker is paused, attempt to block and wait for a state
// change before returning. This avoids a tight loop and mimics the
// behaviour where there are no evals to process.
//
// The call can return because either the timeout is reached or the broker
// SetEnabled function was called to modify its state. It is possible this
// is because of leadership transition, therefore the RPC should exit to
// allow all safety checks and RPC forwarding to occur again.
//
// The log line is trace, because the default worker timeout is 500ms which
// produces a large amount of logging.
if !e.srv.evalBroker.Enabled() {
message := e.srv.evalBroker.enabledNotifier.WaitForChange(args.Timeout)
e.logger.Trace("eval broker wait for un-pause", "message", message)
return nil
}
// Attempt the dequeue
eval, token, err := e.srv.evalBroker.Dequeue(args.Schedulers, args.Timeout)
if err != nil {
return err
}
// Provide the output if any
if eval != nil {
// Get the index that the worker should wait until before scheduling.
waitIndex, err := e.getWaitIndex(eval.Namespace, eval.JobID, eval.ModifyIndex)
if err != nil {
var mErr multierror.Error
_ = multierror.Append(&mErr, err)
// We have dequeued the evaluation but won't be returning it to the
// worker so Nack the eval.
if err := e.srv.evalBroker.Nack(eval.ID, token); err != nil {
_ = multierror.Append(&mErr, err)
}
return &mErr
}
reply.Eval = eval
reply.Token = token
reply.WaitIndex = waitIndex
}
// Set the query response
e.srv.setQueryMeta(&reply.QueryMeta)
return nil
}
// getWaitIndex returns the wait index that should be used by the worker before
// invoking the scheduler. The index should be the highest modify index of any
// evaluation for the job. This prevents scheduling races for the same job when
// there are blocked evaluations.
func (e *Eval) getWaitIndex(namespace, job string, evalModifyIndex uint64) (uint64, error) {
snap, err := e.srv.State().Snapshot()
if err != nil {
return 0, err
}
evals, err := snap.EvalsByJob(nil, namespace, job)
if err != nil {
return 0, err
}
// Since dequeueing evals is concurrent with applying Raft messages to
// the state store, initialize to the currently dequeued eval's index
// in case it isn't in the snapshot used by EvalsByJob yet.
max := evalModifyIndex
for _, eval := range evals {
if max < eval.ModifyIndex {
max = eval.ModifyIndex
}
}
return max, nil
}
// Ack is used to acknowledge completion of a dequeued evaluation
func (e *Eval) Ack(args *structs.EvalAckRequest,
reply *structs.GenericResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Ack", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "ack"}, time.Now())
// Ack the EvalID
if err := e.srv.evalBroker.Ack(args.EvalID, args.Token); err != nil {
return err
}
// Wake up the eval cancelation reaper. This never blocks; if the buffer is
// full we know it's going to get picked up by the reaper so we don't need
// another send on that channel.
select {
case e.srv.reapCancelableEvalsCh <- struct{}{}:
default:
}
return nil
}
// Nack is used to negative acknowledge completion of a dequeued evaluation.
func (e *Eval) Nack(args *structs.EvalAckRequest,
reply *structs.GenericResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Nack", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "nack"}, time.Now())
// Nack the EvalID
if err := e.srv.evalBroker.Nack(args.EvalID, args.Token); err != nil {
return err
}
return nil
}
// Update is used to perform an update of an Eval if it is outstanding.
func (e *Eval) Update(args *structs.EvalUpdateRequest,
reply *structs.GenericResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Update", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "update"}, time.Now())
// Ensure there is only a single update with token
if len(args.Evals) != 1 {
return fmt.Errorf("only a single eval can be updated")
}
eval := args.Evals[0]
// Verify the evaluation is outstanding, and that the tokens match.
if err := e.srv.evalBroker.OutstandingReset(eval.ID, args.EvalToken); err != nil {
return err
}
// Update via Raft
_, index, err := e.srv.raftApply(structs.EvalUpdateRequestType, args)
if err != nil {
return err
}
// Update the index
reply.Index = index
return nil
}
// Create is used to make a new evaluation
func (e *Eval) Create(args *structs.EvalUpdateRequest,
reply *structs.GenericResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Create", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "create"}, time.Now())
// Ensure there is only a single update with token
if len(args.Evals) != 1 {
return fmt.Errorf("only a single eval can be created")
}
eval := args.Evals[0]
// Verify the parent evaluation is outstanding, and that the tokens match.
if err := e.srv.evalBroker.OutstandingReset(eval.PreviousEval, args.EvalToken); err != nil {
return err
}
// Look for the eval
snap, err := e.srv.fsm.State().Snapshot()
if err != nil {
return err
}
ws := memdb.NewWatchSet()
out, err := snap.EvalByID(ws, eval.ID)
if err != nil {
return err
}
if out != nil {
return fmt.Errorf("evaluation already exists")
}
// Update via Raft
_, index, err := e.srv.raftApply(structs.EvalUpdateRequestType, args)
if err != nil {
return err
}
// Update the index
reply.Index = index
return nil
}
// Reblock is used to reinsert an existing blocked evaluation into the blocked
// evaluation tracker.
func (e *Eval) Reblock(args *structs.EvalUpdateRequest, reply *structs.GenericResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Reblock", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "reblock"}, time.Now())
// Ensure there is only a single update with token
if len(args.Evals) != 1 {
return fmt.Errorf("only a single eval can be reblocked")
}
eval := args.Evals[0]
// Verify the evaluation is outstanding, and that the tokens match.
if err := e.srv.evalBroker.OutstandingReset(eval.ID, args.EvalToken); err != nil {
return err
}
// Look for the eval
snap, err := e.srv.fsm.State().Snapshot()
if err != nil {
return err
}
ws := memdb.NewWatchSet()
out, err := snap.EvalByID(ws, eval.ID)
if err != nil {
return err
}
if out == nil {
return fmt.Errorf("evaluation does not exist")
}
if out.Status != structs.EvalStatusBlocked {
return fmt.Errorf("evaluation not blocked")
}
// Reblock the eval
e.srv.blockedEvals.Reblock(eval, args.EvalToken)
return nil
}
// Reap is used to cleanup dead evaluations and allocations
func (e *Eval) Reap(args *structs.EvalReapRequest,
reply *structs.GenericResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
// Ensure the connection was initiated by another server if TLS is used.
err := validateTLSCertificateLevel(e.srv, e.ctx, tlsCertificateLevelServer)
if err != nil {
return err
}
if done, err := e.srv.forward("Eval.Reap", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "reap"}, time.Now())
// Update via Raft
_, index, err := e.srv.raftApply(structs.EvalDeleteRequestType, args)
if err != nil {
return err
}
// Update the index
reply.Index = index
return nil
}
// Delete is used by operators to delete evaluations during severe outages. It
// differs from Reap while duplicating some behavior to ensure we have the
// correct controls for user initiated deletions.
func (e *Eval) Delete(
args *structs.EvalDeleteRequest,
reply *structs.EvalDeleteResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
if done, err := e.srv.forward(structs.EvalDeleteRPCMethod, args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricWrite, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "delete"}, time.Now())
// This RPC endpoint is very destructive and alters Nomad's core state,
// meaning only those with management tokens can call it.
if aclObj, err := e.srv.ResolveACL(args); err != nil {
return err
} else if aclObj != nil && !aclObj.IsManagement() {
return structs.ErrPermissionDenied
}
if args.Filter != "" && !ServersMeetMinimumVersion(
e.srv.Members(), e.srv.Region(), minVersionEvalDeleteByFilter, true) {
return fmt.Errorf(
"all servers must be running version %v or later to delete evals by filter",
minVersionEvalDeleteByFilter)
}
if args.Filter != "" && len(args.EvalIDs) > 0 {
return fmt.Errorf("evals cannot be deleted by both ID and filter")
}
if args.Filter == "" && len(args.EvalIDs) == 0 {
return fmt.Errorf("evals must be deleted by either ID or filter")
}
// The eval broker must be disabled otherwise Nomad's state will likely get
// wild in a very un-fun way.
if e.srv.evalBroker.Enabled() {
return errors.New("eval broker is enabled; eval broker must be paused to delete evals")
}
if args.Filter != "" {
count, index, err := e.deleteEvalsByFilter(args)
if err != nil {
return err
}
// Update the index and return.
reply.Index = index
reply.Count = count
return nil
}
// Grab the state snapshot, so we can look up relevant eval information.
serverStateSnapshot, err := e.srv.State().Snapshot()
if err != nil {
return fmt.Errorf("failed to lookup state snapshot: %v", err)
}
ws := memdb.NewWatchSet()
count := 0
// Iterate the evaluations and ensure they are safe to delete. It is
// possible passed evals are not safe to delete and would make Nomads state
// a little wonky. The nature of the RPC return error, means a single
// unsafe eval ID fails the whole call.
for _, evalID := range args.EvalIDs {
evalInfo, err := serverStateSnapshot.EvalByID(ws, evalID)
if err != nil {
return fmt.Errorf("failed to lookup eval: %v", err)
}
if evalInfo == nil {
return errors.New("eval not found")
}
ok, err := serverStateSnapshot.EvalIsUserDeleteSafe(ws, evalInfo)
if err != nil {
return err
}
if !ok {
return fmt.Errorf("eval %s is not safe to delete", evalInfo.ID)
}
count++
}
// Generate the Raft request object using the reap request object. This
// avoids adding new Raft messages types and follows the existing reap
// flow.
raftReq := structs.EvalReapRequest{
Evals: args.EvalIDs,
UserInitiated: true,
WriteRequest: args.WriteRequest,
}
// Update via Raft.
_, index, err := e.srv.raftApply(structs.EvalDeleteRequestType, &raftReq)
if err != nil {
return err
}
// Update the index and return.
reply.Index = index
reply.Count = count
return nil
}
// deleteEvalsByFilter deletes evaluations in batches based on the filter. It
// returns a count, the index, and any error
func (e *Eval) deleteEvalsByFilter(args *structs.EvalDeleteRequest) (int, uint64, error) {
count := 0
index := uint64(0)
filter, err := bexpr.CreateEvaluator(args.Filter)
if err != nil {
return count, index, err
}
// Note that deleting evals by filter is imprecise: For sets of evals larger
// than a single batch eval inserts may occur behind the cursor and therefore
// be missed. This imprecision is not considered to hurt this endpoint's
// purpose of reducing pressure on servers during periods of heavy scheduling
// activity.
snap, err := e.srv.State().Snapshot()
if err != nil {
return count, index, fmt.Errorf("failed to lookup state snapshot: %v", err)
}
iter, err := snap.Evals(nil, state.SortDefault)
if err != nil {
return count, index, err
}
// We *can* send larger raft logs but rough benchmarks for deleting 1M evals
// show that a smaller page size strikes a balance between throughput and
// time we block the FSM apply for other operations
perPage := structs.MaxUUIDsPerWriteRequest / 10
raftReq := structs.EvalReapRequest{
Filter: args.Filter,
PerPage: int32(perPage),
UserInitiated: true,
WriteRequest: args.WriteRequest,
}
// Note: Paginator is designed around fetching a single page for a single
// RPC call and finalizes its state after that page. So we're doing our own
// pagination here.
pageCount := 0
lastToken := ""
for {
raw := iter.Next()
if raw == nil {
break
}
eval := raw.(*structs.Evaluation)
deleteOk, err := snap.EvalIsUserDeleteSafe(nil, eval)
if !deleteOk || err != nil {
continue
}
match, err := filter.Evaluate(eval)
if !match || err != nil {
continue
}
pageCount++
lastToken = eval.ID
if pageCount >= perPage {
raftReq.PerPage = int32(pageCount)
_, index, err = e.srv.raftApply(structs.EvalDeleteRequestType, &raftReq)
if err != nil {
return count, index, err
}
count += pageCount
pageCount = 0
raftReq.NextToken = lastToken
}
}
// send last batch if it's partial
if pageCount > 0 {
raftReq.PerPage = int32(pageCount)
_, index, err = e.srv.raftApply(structs.EvalDeleteRequestType, &raftReq)
if err != nil {
return count, index, err
}
count += pageCount
}
return count, index, nil
}
// List is used to get a list of the evaluations in the system
func (e *Eval) List(args *structs.EvalListRequest, reply *structs.EvalListResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
if done, err := e.srv.forward("Eval.List", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricList, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "list"}, time.Now())
namespace := args.RequestNamespace()
// Check for read-job permissions
aclObj, err := e.srv.ResolveACL(args)
if err != nil {
return err
}
if !aclObj.AllowNsOp(namespace, acl.NamespaceCapabilityReadJob) {
return structs.ErrPermissionDenied
}
allow := aclObj.AllowNsOpFunc(acl.NamespaceCapabilityReadJob)
if args.Filter != "" {
// Check for incompatible filtering.
hasLegacyFilter := args.FilterJobID != "" || args.FilterEvalStatus != ""
if hasLegacyFilter {
return structs.ErrIncompatibleFiltering
}
}
// Setup the blocking query
sort := state.SortOption(args.Reverse)
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, store *state.StateStore) error {
// Scan all the evaluations
var err error
var iter memdb.ResultIterator
var opts paginator.StructsTokenizerOptions
// Get the namespaces the user is allowed to access.
allowableNamespaces, err := allowedNSes(aclObj, store, allow)
if err == structs.ErrPermissionDenied {
// return empty evals if token isn't authorized for any
// namespace, matching other endpoints
reply.Evaluations = make([]*structs.Evaluation, 0)
} else if err != nil {
return err
} else {
if prefix := args.QueryOptions.Prefix; prefix != "" {
iter, err = store.EvalsByIDPrefix(ws, namespace, prefix, sort)
opts = paginator.StructsTokenizerOptions{
WithID: true,
}
} else if namespace != structs.AllNamespacesSentinel {
iter, err = store.EvalsByNamespaceOrdered(ws, namespace, sort)
opts = paginator.StructsTokenizerOptions{
WithCreateIndex: true,
WithID: true,
}
} else {
iter, err = store.Evals(ws, sort)
opts = paginator.StructsTokenizerOptions{
WithCreateIndex: true,
WithID: true,
}
}
if err != nil {
return err
}
iter = memdb.NewFilterIterator(iter, func(raw interface{}) bool {
if eval := raw.(*structs.Evaluation); eval != nil {
return args.ShouldBeFiltered(eval)
}
return false
})
tokenizer := paginator.NewStructsTokenizer(iter, opts)
filters := []paginator.Filter{
paginator.NamespaceFilter{
AllowableNamespaces: allowableNamespaces,
},
}
var evals []*structs.Evaluation
paginator, err := paginator.NewPaginator(iter, tokenizer, filters, args.QueryOptions,
func(raw interface{}) error {
eval := raw.(*structs.Evaluation)
evals = append(evals, eval)
return nil
})
if err != nil {
return structs.NewErrRPCCodedf(
http.StatusBadRequest, "failed to create result paginator: %v", err)
}
nextToken, err := paginator.Page()
if err != nil {
return structs.NewErrRPCCodedf(
http.StatusBadRequest, "failed to read result page: %v", err)
}
reply.QueryMeta.NextToken = nextToken
reply.Evaluations = evals
}
// Use the last index that affected the jobs table
index, err := store.Index("evals")
if err != nil {
return err
}
reply.Index = index
// Set the query response
e.srv.setQueryMeta(&reply.QueryMeta)
return nil
}}
return e.srv.blockingRPC(&opts)
}
// Count is used to get a list of the evaluations in the system
func (e *Eval) Count(args *structs.EvalCountRequest, reply *structs.EvalCountResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
if done, err := e.srv.forward("Eval.Count", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricList, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "count"}, time.Now())
namespace := args.RequestNamespace()
// Check for read-job permissions
aclObj, err := e.srv.ResolveACL(args)
if err != nil {
return err
}
if !aclObj.AllowNsOp(namespace, acl.NamespaceCapabilityReadJob) {
return structs.ErrPermissionDenied
}
allow := aclObj.AllowNsOpFunc(acl.NamespaceCapabilityReadJob)
var filter *bexpr.Evaluator
if args.Filter != "" {
filter, err = bexpr.CreateEvaluator(args.Filter)
if err != nil {
return err
}
}
// Setup the blocking query. This is only superficially like Eval.List,
// because we don't any concerns about pagination, sorting, and legacy
// filter fields.
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, store *state.StateStore) error {
// Scan all the evaluations
var err error
var iter memdb.ResultIterator
// Get the namespaces the user is allowed to access.
allowableNamespaces, err := allowedNSes(aclObj, store, allow)
if err != nil {
return err
}
if prefix := args.QueryOptions.Prefix; prefix != "" {
iter, err = store.EvalsByIDPrefix(ws, namespace, prefix, state.SortDefault)
} else if namespace != structs.AllNamespacesSentinel {
iter, err = store.EvalsByNamespace(ws, namespace)
} else {
iter, err = store.Evals(ws, state.SortDefault)
}
if err != nil {
return err
}
count := 0
iter = memdb.NewFilterIterator(iter, func(raw interface{}) bool {
if raw == nil {
return true
}
eval := raw.(*structs.Evaluation)
if allowableNamespaces != nil && !allowableNamespaces[eval.Namespace] {
return true
}
if filter != nil {
ok, err := filter.Evaluate(eval)
if err != nil {
return true
}
return !ok
}
return false
})
for {
raw := iter.Next()
if raw == nil {
break
}
count++
}
// Use the last index that affected the jobs table
index, err := store.Index("evals")
if err != nil {
return err
}
reply.Index = index
reply.Count = count
// Set the query response
e.srv.setQueryMeta(&reply.QueryMeta)
return nil
}}
return e.srv.blockingRPC(&opts)
}
// Allocations is used to list the allocations for an evaluation
func (e *Eval) Allocations(args *structs.EvalSpecificRequest,
reply *structs.EvalAllocationsResponse) error {
authErr := e.srv.Authenticate(e.ctx, args)
if done, err := e.srv.forward("Eval.Allocations", args, args, reply); done {
return err
}
e.srv.MeasureRPCRate("eval", structs.RateMetricList, args)
if authErr != nil {
return structs.ErrPermissionDenied
}
defer metrics.MeasureSince([]string{"nomad", "eval", "allocations"}, time.Now())
// Check for read-job permissions
allowNsOp := acl.NamespaceValidator(acl.NamespaceCapabilityReadJob)
aclObj, err := e.srv.ResolveACL(args)
if err != nil {
return err
} else if !allowNsOp(aclObj, args.RequestNamespace()) {
return structs.ErrPermissionDenied
}
// Setup the blocking query
opts := blockingOptions{
queryOpts: &args.QueryOptions,
queryMeta: &reply.QueryMeta,
run: func(ws memdb.WatchSet, state *state.StateStore) error {
// Capture the allocations
allocs, err := state.AllocsByEval(ws, args.EvalID)
if err != nil {
return err
}
// Convert to a stub
if len(allocs) > 0 {
// Evaluations do not span namespaces so just check the
// first allocs namespace.
ns := allocs[0].Namespace
if ns != args.RequestNamespace() && !allowNsOp(aclObj, ns) {
return structs.ErrPermissionDenied
}
reply.Allocations = make([]*structs.AllocListStub, 0, len(allocs))
for _, alloc := range allocs {
reply.Allocations = append(reply.Allocations, alloc.Stub(nil))
}
}
// Use the last index that affected the allocs table
index, err := state.Index("allocs")
if err != nil {
return err
}
reply.Index = index
// Set the query response
e.srv.setQueryMeta(&reply.QueryMeta)
return nil
}}
return e.srv.blockingRPC(&opts)
}