-
Notifications
You must be signed in to change notification settings - Fork 815
/
Copy pathdistributor.go
1657 lines (1422 loc) · 65.8 KB
/
distributor.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
package distributor
import (
"context"
"flag"
"fmt"
io "io"
"net/http"
"sort"
"strings"
"sync"
"time"
"github.com/go-kit/log"
"github.com/go-kit/log/level"
"github.com/opentracing/opentracing-go"
"github.com/pkg/errors"
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promauto"
"github.com/prometheus/common/model"
"github.com/prometheus/prometheus/model/labels"
"github.com/prometheus/prometheus/model/relabel"
"github.com/prometheus/prometheus/scrape"
"github.com/prometheus/prometheus/storage"
"github.com/weaveworks/common/httpgrpc"
"github.com/weaveworks/common/instrument"
"github.com/weaveworks/common/user"
"go.uber.org/atomic"
"github.com/cortexproject/cortex/pkg/cortexpb"
"github.com/cortexproject/cortex/pkg/ha"
"github.com/cortexproject/cortex/pkg/ingester"
ingester_client "github.com/cortexproject/cortex/pkg/ingester/client"
"github.com/cortexproject/cortex/pkg/ring"
ring_client "github.com/cortexproject/cortex/pkg/ring/client"
"github.com/cortexproject/cortex/pkg/tenant"
"github.com/cortexproject/cortex/pkg/util"
"github.com/cortexproject/cortex/pkg/util/extract"
"github.com/cortexproject/cortex/pkg/util/labelset"
"github.com/cortexproject/cortex/pkg/util/limiter"
util_log "github.com/cortexproject/cortex/pkg/util/log"
util_math "github.com/cortexproject/cortex/pkg/util/math"
"github.com/cortexproject/cortex/pkg/util/services"
"github.com/cortexproject/cortex/pkg/util/validation"
)
var (
emptyPreallocSeries = cortexpb.PreallocTimeseries{}
supportedShardingStrategies = []string{util.ShardingStrategyDefault, util.ShardingStrategyShuffle}
// Validation errors.
errInvalidShardingStrategy = errors.New("invalid sharding strategy")
errInvalidTenantShardSize = errors.New("invalid tenant shard size. The value must be greater than or equal to 0")
)
const (
// ringKey is the key under which we store the distributors ring in the KVStore.
ringKey = "distributor"
typeSamples = "samples"
typeMetadata = "metadata"
instanceIngestionRateTickInterval = time.Second
clearStaleIngesterMetricsInterval = time.Minute
labelSetMetricsTickInterval = 30 * time.Second
// mergeSlicesParallelism is a constant of how much go routines we should use to merge slices, and
// it was based on empirical observation: See BenchmarkMergeSlicesParallel
mergeSlicesParallelism = 8
sampleMetricTypeFloat = "float"
sampleMetricTypeHistogram = "histogram"
)
// Distributor is a storage.SampleAppender and a client.Querier which
// forwards appends and queries to individual ingesters.
type Distributor struct {
services.Service
cfg Config
log log.Logger
ingestersRing ring.ReadRing
ingesterPool *ring_client.Pool
limits *validation.Overrides
// The global rate limiter requires a distributors ring to count
// the number of healthy instances
distributorsLifeCycler *ring.Lifecycler
distributorsRing *ring.Ring
// For handling HA replicas.
HATracker *ha.HATracker
// Per-user rate limiter.
ingestionRateLimiter *limiter.RateLimiter
// Manager for subservices (HA Tracker, distributor ring and client pool)
subservices *services.Manager
subservicesWatcher *services.FailureWatcher
activeUsers *util.ActiveUsersCleanupService
ingestionRate *util_math.EwmaRate
inflightPushRequests atomic.Int64
inflightClientRequests atomic.Int64
// Metrics
queryDuration *instrument.HistogramCollector
receivedSamples *prometheus.CounterVec
receivedSamplesPerLabelSet *prometheus.CounterVec
receivedExemplars *prometheus.CounterVec
receivedMetadata *prometheus.CounterVec
incomingSamples *prometheus.CounterVec
incomingExemplars *prometheus.CounterVec
incomingMetadata *prometheus.CounterVec
nonHASamples *prometheus.CounterVec
dedupedSamples *prometheus.CounterVec
labelsHistogram prometheus.Histogram
ingesterAppends *prometheus.CounterVec
ingesterAppendFailures *prometheus.CounterVec
ingesterQueries *prometheus.CounterVec
ingesterQueryFailures *prometheus.CounterVec
ingesterPartialDataQueries prometheus.Counter
replicationFactor prometheus.Gauge
latestSeenSampleTimestampPerUser *prometheus.GaugeVec
validateMetrics *validation.ValidateMetrics
asyncExecutor util.AsyncExecutor
// Map to track label sets from user.
labelSetTracker *labelset.LabelSetTracker
}
// Config contains the configuration required to
// create a Distributor
type Config struct {
PoolConfig PoolConfig `yaml:"pool"`
HATrackerConfig HATrackerConfig `yaml:"ha_tracker"`
MaxRecvMsgSize int `yaml:"max_recv_msg_size"`
OTLPMaxRecvMsgSize int `yaml:"otlp_max_recv_msg_size"`
RemoteTimeout time.Duration `yaml:"remote_timeout"`
ExtraQueryDelay time.Duration `yaml:"extra_queue_delay"`
ShardingStrategy string `yaml:"sharding_strategy"`
ShardByAllLabels bool `yaml:"shard_by_all_labels"`
ExtendWrites bool `yaml:"extend_writes"`
SignWriteRequestsEnabled bool `yaml:"sign_write_requests"`
// Distributors ring
DistributorRing RingConfig `yaml:"ring"`
// for testing and for extending the ingester by adding calls to the client
IngesterClientFactory ring_client.PoolFactory `yaml:"-"`
// when true the distributor does not validate the label name, Cortex doesn't directly use
// this (and should never use it) but this feature is used by other projects built on top of it
SkipLabelNameValidation bool `yaml:"-"`
// This config is dynamically injected because defined in the querier config.
ShuffleShardingLookbackPeriod time.Duration `yaml:"-"`
// ZoneResultsQuorumMetadata enables zone results quorum when querying ingester replication set
// with metadata APIs (labels names and values for now). When zone awareness is enabled, only results
// from quorum number of zones will be included to reduce data merged and improve performance.
ZoneResultsQuorumMetadata bool `yaml:"zone_results_quorum_metadata" doc:"hidden"`
// Number of go routines to handle push calls from distributors to ingesters.
// If set to 0 (default), workers are disabled, and a new goroutine will be created for each push request.
// When no workers are available, a new goroutine will be spawned automatically.
NumPushWorkers int `yaml:"num_push_workers"`
// Limits for distributor
InstanceLimits InstanceLimits `yaml:"instance_limits"`
// OTLPConfig
OTLPConfig OTLPConfig `yaml:"otlp"`
}
type InstanceLimits struct {
MaxIngestionRate float64 `yaml:"max_ingestion_rate"`
MaxInflightPushRequests int `yaml:"max_inflight_push_requests"`
MaxInflightClientRequests int `yaml:"max_inflight_client_requests"`
}
type OTLPConfig struct {
ConvertAllAttributes bool `yaml:"convert_all_attributes"`
DisableTargetInfo bool `yaml:"disable_target_info"`
}
// RegisterFlags adds the flags required to config this to the given FlagSet
func (cfg *Config) RegisterFlags(f *flag.FlagSet) {
cfg.PoolConfig.RegisterFlags(f)
cfg.HATrackerConfig.RegisterFlags(f)
cfg.DistributorRing.RegisterFlags(f)
f.IntVar(&cfg.MaxRecvMsgSize, "distributor.max-recv-msg-size", 100<<20, "remote_write API max receive message size (bytes).")
f.IntVar(&cfg.OTLPMaxRecvMsgSize, "distributor.otlp-max-recv-msg-size", 100<<20, "Maximum OTLP request size in bytes that the Distributor can accept.")
f.DurationVar(&cfg.RemoteTimeout, "distributor.remote-timeout", 2*time.Second, "Timeout for downstream ingesters.")
f.DurationVar(&cfg.ExtraQueryDelay, "distributor.extra-query-delay", 0, "Time to wait before sending more than the minimum successful query requests.")
f.BoolVar(&cfg.ShardByAllLabels, "distributor.shard-by-all-labels", false, "Distribute samples based on all labels, as opposed to solely by user and metric name.")
f.BoolVar(&cfg.SignWriteRequestsEnabled, "distributor.sign-write-requests", false, "EXPERIMENTAL: If enabled, sign the write request between distributors and ingesters.")
f.StringVar(&cfg.ShardingStrategy, "distributor.sharding-strategy", util.ShardingStrategyDefault, fmt.Sprintf("The sharding strategy to use. Supported values are: %s.", strings.Join(supportedShardingStrategies, ", ")))
f.BoolVar(&cfg.ExtendWrites, "distributor.extend-writes", true, "Try writing to an additional ingester in the presence of an ingester not in the ACTIVE state. It is useful to disable this along with -ingester.unregister-on-shutdown=false in order to not spread samples to extra ingesters during rolling restarts with consistent naming.")
f.BoolVar(&cfg.ZoneResultsQuorumMetadata, "distributor.zone-results-quorum-metadata", false, "Experimental, this flag may change in the future. If zone awareness and this both enabled, when querying metadata APIs (labels names and values for now), only results from quorum number of zones will be included.")
f.IntVar(&cfg.NumPushWorkers, "distributor.num-push-workers", 0, "EXPERIMENTAL: Number of go routines to handle push calls from distributors to ingesters. When no workers are available, a new goroutine will be spawned automatically. If set to 0 (default), workers are disabled, and a new goroutine will be created for each push request.")
f.Float64Var(&cfg.InstanceLimits.MaxIngestionRate, "distributor.instance-limits.max-ingestion-rate", 0, "Max ingestion rate (samples/sec) that this distributor will accept. This limit is per-distributor, not per-tenant. Additional push requests will be rejected. Current ingestion rate is computed as exponentially weighted moving average, updated every second. 0 = unlimited.")
f.IntVar(&cfg.InstanceLimits.MaxInflightPushRequests, "distributor.instance-limits.max-inflight-push-requests", 0, "Max inflight push requests that this distributor can handle. This limit is per-distributor, not per-tenant. Additional requests will be rejected. 0 = unlimited.")
f.IntVar(&cfg.InstanceLimits.MaxInflightClientRequests, "distributor.instance-limits.max-inflight-client-requests", 0, "Max inflight ingester client requests that this distributor can handle. This limit is per-distributor, not per-tenant. Additional requests will be rejected. 0 = unlimited.")
f.BoolVar(&cfg.OTLPConfig.ConvertAllAttributes, "distributor.otlp.convert-all-attributes", false, "If true, all resource attributes are converted to labels.")
f.BoolVar(&cfg.OTLPConfig.DisableTargetInfo, "distributor.otlp.disable-target-info", false, "If true, a target_info metric is not ingested. (refer to: https://github.com/prometheus/OpenMetrics/blob/main/specification/OpenMetrics.md#supporting-target-metadata-in-both-push-based-and-pull-based-systems)")
}
// Validate config and returns error on failure
func (cfg *Config) Validate(limits validation.Limits) error {
if !util.StringsContain(supportedShardingStrategies, cfg.ShardingStrategy) {
return errInvalidShardingStrategy
}
if cfg.ShardingStrategy == util.ShardingStrategyShuffle && limits.IngestionTenantShardSize < 0 {
return errInvalidTenantShardSize
}
haHATrackerConfig := cfg.HATrackerConfig.ToHATrackerConfig()
return haHATrackerConfig.Validate()
}
const (
instanceLimitsMetric = "cortex_distributor_instance_limits"
instanceLimitsMetricHelp = "Instance limits used by this distributor." // Must be same for all registrations.
limitLabel = "limit"
)
// New constructs a new Distributor
func New(cfg Config, clientConfig ingester_client.Config, limits *validation.Overrides, ingestersRing ring.ReadRing, canJoinDistributorsRing bool, reg prometheus.Registerer, log log.Logger) (*Distributor, error) {
if cfg.IngesterClientFactory == nil {
cfg.IngesterClientFactory = func(addr string) (ring_client.PoolClient, error) {
return ingester_client.MakeIngesterClient(addr, clientConfig)
}
}
cfg.PoolConfig.RemoteTimeout = cfg.RemoteTimeout
haTrackerStatusConfig := ha.HATrackerStatusConfig{
Title: "Cortex HA Tracker Status",
ReplicaGroupLabel: "Cluster",
}
haTracker, err := ha.NewHATracker(cfg.HATrackerConfig.ToHATrackerConfig(), limits, haTrackerStatusConfig, prometheus.WrapRegistererWithPrefix("cortex_", reg), "distributor-hatracker", log)
if err != nil {
return nil, err
}
subservices := []services.Service(nil)
subservices = append(subservices, haTracker)
// Create the configured ingestion rate limit strategy (local or global). In case
// it's an internal dependency and can't join the distributors ring, we skip rate
// limiting.
var ingestionRateStrategy limiter.RateLimiterStrategy
var distributorsLifeCycler *ring.Lifecycler
var distributorsRing *ring.Ring
if !canJoinDistributorsRing {
ingestionRateStrategy = newInfiniteIngestionRateStrategy()
} else if limits.IngestionRateStrategy() == validation.GlobalIngestionRateStrategy {
distributorsLifeCycler, err = ring.NewLifecycler(cfg.DistributorRing.ToLifecyclerConfig(), nil, "distributor", ringKey, true, true, log, prometheus.WrapRegistererWithPrefix("cortex_", reg))
if err != nil {
return nil, err
}
distributorsRing, err = ring.New(cfg.DistributorRing.ToRingConfig(), "distributor", ringKey, log, prometheus.WrapRegistererWithPrefix("cortex_", reg))
if err != nil {
return nil, errors.Wrap(err, "failed to initialize distributors' ring client")
}
subservices = append(subservices, distributorsLifeCycler, distributorsRing)
ingestionRateStrategy = newGlobalIngestionRateStrategy(limits, distributorsLifeCycler)
} else {
ingestionRateStrategy = newLocalIngestionRateStrategy(limits)
}
d := &Distributor{
cfg: cfg,
log: log,
ingestersRing: ingestersRing,
ingesterPool: NewPool(cfg.PoolConfig, ingestersRing, cfg.IngesterClientFactory, log),
distributorsLifeCycler: distributorsLifeCycler,
distributorsRing: distributorsRing,
limits: limits,
ingestionRateLimiter: limiter.NewRateLimiter(ingestionRateStrategy, 10*time.Second),
HATracker: haTracker,
ingestionRate: util_math.NewEWMARate(0.2, instanceIngestionRateTickInterval),
queryDuration: instrument.NewHistogramCollector(promauto.With(reg).NewHistogramVec(prometheus.HistogramOpts{
Namespace: "cortex",
Name: "distributor_query_duration_seconds",
Help: "Time spent executing expression and exemplar queries.",
Buckets: []float64{.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10, 20, 30},
}, []string{"method", "status_code"})),
receivedSamples: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_received_samples_total",
Help: "The total number of received samples, excluding rejected and deduped samples.",
}, []string{"user", "type"}),
receivedSamplesPerLabelSet: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_received_samples_per_labelset_total",
Help: "The total number of received samples per label set, excluding rejected and deduped samples.",
}, []string{"user", "type", "labelset"}),
receivedExemplars: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_received_exemplars_total",
Help: "The total number of received exemplars, excluding rejected and deduped exemplars.",
}, []string{"user"}),
receivedMetadata: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_received_metadata_total",
Help: "The total number of received metadata, excluding rejected.",
}, []string{"user"}),
incomingSamples: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_samples_in_total",
Help: "The total number of samples that have come in to the distributor, including rejected or deduped samples.",
}, []string{"user", "type"}),
incomingExemplars: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_exemplars_in_total",
Help: "The total number of exemplars that have come in to the distributor, including rejected or deduped exemplars.",
}, []string{"user"}),
incomingMetadata: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_metadata_in_total",
Help: "The total number of metadata that have come in to the distributor, including rejected.",
}, []string{"user"}),
nonHASamples: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_non_ha_samples_received_total",
Help: "The total number of received samples for a user that has HA tracking turned on, but the sample didn't contain both HA labels.",
}, []string{"user"}),
dedupedSamples: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_deduped_samples_total",
Help: "The total number of deduplicated samples.",
}, []string{"user", "cluster"}),
labelsHistogram: promauto.With(reg).NewHistogram(prometheus.HistogramOpts{
Namespace: "cortex",
Name: "labels_per_sample",
Help: "Number of labels per sample.",
Buckets: []float64{5, 10, 15, 20, 25},
}),
ingesterAppends: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_ingester_appends_total",
Help: "The total number of batch appends sent to ingesters.",
}, []string{"ingester", "type"}),
ingesterAppendFailures: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_ingester_append_failures_total",
Help: "The total number of failed batch appends sent to ingesters.",
}, []string{"ingester", "type", "status"}),
ingesterQueries: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_ingester_queries_total",
Help: "The total number of queries sent to ingesters.",
}, []string{"ingester"}),
ingesterQueryFailures: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_ingester_query_failures_total",
Help: "The total number of failed queries sent to ingesters.",
}, []string{"ingester"}),
ingesterPartialDataQueries: promauto.With(reg).NewCounter(prometheus.CounterOpts{
Namespace: "cortex",
Name: "distributor_ingester_partial_data_queries_total",
Help: "The total number of queries sent to ingesters that may have returned partial data.",
}),
replicationFactor: promauto.With(reg).NewGauge(prometheus.GaugeOpts{
Namespace: "cortex",
Name: "distributor_replication_factor",
Help: "The configured replication factor.",
}),
latestSeenSampleTimestampPerUser: promauto.With(reg).NewGaugeVec(prometheus.GaugeOpts{
Name: "cortex_distributor_latest_seen_sample_timestamp_seconds",
Help: "Unix timestamp of latest received sample per user.",
}, []string{"user"}),
validateMetrics: validation.NewValidateMetrics(reg),
asyncExecutor: util.NewNoOpExecutor(),
}
d.labelSetTracker = labelset.NewLabelSetTracker()
if cfg.NumPushWorkers > 0 {
util_log.WarnExperimentalUse("Distributor: using goroutine worker pool")
d.asyncExecutor = util.NewWorkerPool("distributor", cfg.NumPushWorkers, reg)
}
promauto.With(reg).NewGauge(prometheus.GaugeOpts{
Name: instanceLimitsMetric,
Help: instanceLimitsMetricHelp,
ConstLabels: map[string]string{limitLabel: "max_inflight_push_requests"},
}).Set(float64(cfg.InstanceLimits.MaxInflightPushRequests))
promauto.With(reg).NewGauge(prometheus.GaugeOpts{
Name: instanceLimitsMetric,
Help: instanceLimitsMetricHelp,
ConstLabels: map[string]string{limitLabel: "max_inflight_client_requests"},
}).Set(float64(cfg.InstanceLimits.MaxInflightClientRequests))
promauto.With(reg).NewGauge(prometheus.GaugeOpts{
Name: instanceLimitsMetric,
Help: instanceLimitsMetricHelp,
ConstLabels: map[string]string{limitLabel: "max_ingestion_rate"},
}).Set(cfg.InstanceLimits.MaxIngestionRate)
promauto.With(reg).NewGaugeFunc(prometheus.GaugeOpts{
Name: "cortex_distributor_inflight_push_requests",
Help: "Current number of inflight push requests in distributor.",
}, func() float64 {
return float64(d.inflightPushRequests.Load())
})
promauto.With(reg).NewGaugeFunc(prometheus.GaugeOpts{
Name: "cortex_distributor_inflight_client_requests",
Help: "Current number of inflight client requests in distributor.",
}, func() float64 {
return float64(d.inflightClientRequests.Load())
})
promauto.With(reg).NewGaugeFunc(prometheus.GaugeOpts{
Name: "cortex_distributor_ingestion_rate_samples_per_second",
Help: "Current ingestion rate in samples/sec that distributor is using to limit access.",
}, func() float64 {
return d.ingestionRate.Rate()
})
d.replicationFactor.Set(float64(ingestersRing.ReplicationFactor()))
d.activeUsers = util.NewActiveUsersCleanupWithDefaultValues(d.cleanupInactiveUser)
subservices = append(subservices, d.ingesterPool, d.activeUsers)
d.subservices, err = services.NewManager(subservices...)
if err != nil {
return nil, err
}
d.subservicesWatcher = services.NewFailureWatcher()
d.subservicesWatcher.WatchManager(d.subservices)
d.Service = services.NewBasicService(d.starting, d.running, d.stopping)
return d, nil
}
func (d *Distributor) starting(ctx context.Context) error {
if d.cfg.InstanceLimits != (InstanceLimits{}) {
util_log.WarnExperimentalUse("distributor instance limits")
}
// Only report success if all sub-services start properly
return services.StartManagerAndAwaitHealthy(ctx, d.subservices)
}
func (d *Distributor) running(ctx context.Context) error {
ingestionRateTicker := time.NewTicker(instanceIngestionRateTickInterval)
defer ingestionRateTicker.Stop()
staleIngesterMetricTicker := time.NewTicker(clearStaleIngesterMetricsInterval)
defer staleIngesterMetricTicker.Stop()
labelSetMetricsTicker := time.NewTicker(labelSetMetricsTickInterval)
defer labelSetMetricsTicker.Stop()
for {
select {
case <-ctx.Done():
return nil
case <-ingestionRateTicker.C:
d.ingestionRate.Tick()
case <-staleIngesterMetricTicker.C:
d.cleanStaleIngesterMetrics()
case <-labelSetMetricsTicker.C:
d.updateLabelSetMetrics()
case err := <-d.subservicesWatcher.Chan():
return errors.Wrap(err, "distributor subservice failed")
}
}
}
func (d *Distributor) cleanupInactiveUser(userID string) {
d.ingestersRing.CleanupShuffleShardCache(userID)
d.HATracker.CleanupHATrackerMetricsForUser(userID)
d.receivedSamples.DeleteLabelValues(userID, sampleMetricTypeFloat)
d.receivedSamples.DeleteLabelValues(userID, sampleMetricTypeHistogram)
d.receivedExemplars.DeleteLabelValues(userID)
d.receivedMetadata.DeleteLabelValues(userID)
d.incomingSamples.DeleteLabelValues(userID, sampleMetricTypeFloat)
d.incomingSamples.DeleteLabelValues(userID, sampleMetricTypeHistogram)
d.incomingExemplars.DeleteLabelValues(userID)
d.incomingMetadata.DeleteLabelValues(userID)
d.nonHASamples.DeleteLabelValues(userID)
d.latestSeenSampleTimestampPerUser.DeleteLabelValues(userID)
if err := util.DeleteMatchingLabels(d.dedupedSamples, map[string]string{"user": userID}); err != nil {
level.Warn(d.log).Log("msg", "failed to remove cortex_distributor_deduped_samples_total metric for user", "user", userID, "err", err)
}
if err := util.DeleteMatchingLabels(d.receivedSamplesPerLabelSet, map[string]string{"user": userID}); err != nil {
level.Warn(d.log).Log("msg", "failed to remove cortex_distributor_received_samples_per_labelset_total metric for user", "user", userID, "err", err)
}
validation.DeletePerUserValidationMetrics(d.validateMetrics, userID, d.log)
}
// Called after distributor is asked to stop via StopAsync.
func (d *Distributor) stopping(_ error) error {
return services.StopManagerAndAwaitStopped(context.Background(), d.subservices)
}
func (d *Distributor) tokenForLabels(userID string, labels []cortexpb.LabelAdapter) (uint32, error) {
if d.cfg.ShardByAllLabels {
return shardByAllLabels(userID, labels), nil
}
unsafeMetricName, err := extract.UnsafeMetricNameFromLabelAdapters(labels)
if err != nil {
return 0, err
}
return shardByMetricName(userID, unsafeMetricName), nil
}
func (d *Distributor) tokenForMetadata(userID string, metricName string) uint32 {
if d.cfg.ShardByAllLabels {
return shardByMetricName(userID, metricName)
}
return shardByUser(userID)
}
// shardByMetricName returns the token for the given metric. The provided metricName
// is guaranteed to not be retained.
func shardByMetricName(userID string, metricName string) uint32 {
h := shardByUser(userID)
h = ingester_client.HashAdd32(h, metricName)
return h
}
func shardByUser(userID string) uint32 {
h := ingester_client.HashNew32()
h = ingester_client.HashAdd32(h, userID)
return h
}
// This function generates different values for different order of same labels.
func shardByAllLabels(userID string, labels []cortexpb.LabelAdapter) uint32 {
h := shardByUser(userID)
for _, label := range labels {
if len(label.Value) > 0 {
h = ingester_client.HashAdd32(h, label.Name)
h = ingester_client.HashAdd32(h, label.Value)
}
}
return h
}
// Remove the label labelname from a slice of LabelPairs if it exists.
func removeLabel(labelName string, labels *[]cortexpb.LabelAdapter) {
for i := 0; i < len(*labels); i++ {
pair := (*labels)[i]
if pair.Name == labelName {
*labels = append((*labels)[:i], (*labels)[i+1:]...)
return
}
}
}
// Returns a boolean that indicates whether or not we want to remove the replica label going forward,
// and an error that indicates whether we want to accept samples based on the cluster/replica found in ts.
// nil for the error means accept the sample.
func (d *Distributor) checkSample(ctx context.Context, userID, cluster, replica string, limits *validation.Limits) (removeReplicaLabel bool, _ error) {
// If the sample doesn't have either HA label, accept it.
// At the moment we want to accept these samples by default.
if cluster == "" || replica == "" {
return false, nil
}
// If replica label is too long, don't use it. We accept the sample here, but it will fail validation later anyway.
if len(replica) > limits.MaxLabelValueLength {
return false, nil
}
// At this point we know we have both HA labels, we should lookup
// the cluster/instance here to see if we want to accept this sample.
err := d.HATracker.CheckReplica(ctx, userID, cluster, replica, time.Now())
// checkReplica should only have returned an error if there was a real error talking to Consul, or if the replica labels don't match.
if err != nil { // Don't accept the sample.
return false, err
}
return true, nil
}
// Validates a single series from a write request. Will remove labels if
// any are configured to be dropped for the user ID.
// Returns the validated series with it's labels/samples, and any error.
// The returned error may retain the series labels.
func (d *Distributor) validateSeries(ts cortexpb.PreallocTimeseries, userID string, skipLabelNameValidation bool, limits *validation.Limits) (cortexpb.PreallocTimeseries, validation.ValidationError) {
d.labelsHistogram.Observe(float64(len(ts.Labels)))
if err := validation.ValidateLabels(d.validateMetrics, limits, userID, ts.Labels, skipLabelNameValidation); err != nil {
return emptyPreallocSeries, err
}
var samples []cortexpb.Sample
if len(ts.Samples) > 0 {
// Only alloc when data present
samples = make([]cortexpb.Sample, 0, len(ts.Samples))
for _, s := range ts.Samples {
if err := validation.ValidateSampleTimestamp(d.validateMetrics, limits, userID, ts.Labels, s.TimestampMs); err != nil {
return emptyPreallocSeries, err
}
samples = append(samples, s)
}
}
var exemplars []cortexpb.Exemplar
if len(ts.Exemplars) > 0 {
// Only alloc when data present
exemplars = make([]cortexpb.Exemplar, 0, len(ts.Exemplars))
for _, e := range ts.Exemplars {
if err := validation.ValidateExemplar(d.validateMetrics, userID, ts.Labels, e); err != nil {
// An exemplar validation error prevents ingesting samples
// in the same series object. However, because the current Prometheus
// remote write implementation only populates one or the other,
// there never will be any.
return emptyPreallocSeries, err
}
exemplars = append(exemplars, e)
}
}
var histograms []cortexpb.Histogram
if len(ts.Histograms) > 0 {
// Only alloc when data present
histograms = make([]cortexpb.Histogram, 0, len(ts.Histograms))
for i, h := range ts.Histograms {
if err := validation.ValidateSampleTimestamp(d.validateMetrics, limits, userID, ts.Labels, h.TimestampMs); err != nil {
return emptyPreallocSeries, err
}
// TODO(yeya24): add max schema validation for native histogram if needed.
convertedHistogram, err := validation.ValidateNativeHistogram(d.validateMetrics, limits, userID, ts.Labels, h)
if err != nil {
return emptyPreallocSeries, err
}
ts.Histograms[i] = convertedHistogram
}
histograms = append(histograms, ts.Histograms...)
}
return cortexpb.PreallocTimeseries{
TimeSeries: &cortexpb.TimeSeries{
Labels: ts.Labels,
Samples: samples,
Exemplars: exemplars,
Histograms: histograms,
},
},
nil
}
// Push implements client.IngesterServer
func (d *Distributor) Push(ctx context.Context, req *cortexpb.WriteRequest) (*cortexpb.WriteResponse, error) {
userID, err := tenant.TenantID(ctx)
if err != nil {
return nil, err
}
span, ctx := opentracing.StartSpanFromContext(ctx, "Distributor.Push")
defer span.Finish()
// We will report *this* request in the error too.
inflight := d.inflightPushRequests.Inc()
defer d.inflightPushRequests.Dec()
now := time.Now()
d.activeUsers.UpdateUserTimestamp(userID, now)
numFloatSamples := 0
numHistogramSamples := 0
numExemplars := 0
for _, ts := range req.Timeseries {
numFloatSamples += len(ts.Samples)
numHistogramSamples += len(ts.Histograms)
numExemplars += len(ts.Exemplars)
}
// Count the total samples, exemplars in, prior to validation or deduplication, for comparison with other metrics.
d.incomingSamples.WithLabelValues(userID, sampleMetricTypeFloat).Add(float64(numFloatSamples))
d.incomingSamples.WithLabelValues(userID, sampleMetricTypeHistogram).Add(float64(numHistogramSamples))
d.incomingExemplars.WithLabelValues(userID).Add(float64(numExemplars))
// Count the total number of metadata in.
d.incomingMetadata.WithLabelValues(userID).Add(float64(len(req.Metadata)))
if d.cfg.InstanceLimits.MaxInflightPushRequests > 0 && inflight > int64(d.cfg.InstanceLimits.MaxInflightPushRequests) {
return nil, httpgrpc.Errorf(http.StatusServiceUnavailable, "too many inflight push requests in distributor")
}
if d.cfg.InstanceLimits.MaxIngestionRate > 0 {
if rate := d.ingestionRate.Rate(); rate >= d.cfg.InstanceLimits.MaxIngestionRate {
return nil, httpgrpc.Errorf(http.StatusServiceUnavailable, "distributor's samples push rate limit reached")
}
}
// only reject requests at this stage to allow distributor to finish sending the current batch request to all ingesters
// even if we've exceeded the MaxInflightClientRequests in the `doBatch`
if d.cfg.InstanceLimits.MaxInflightClientRequests > 0 && d.inflightClientRequests.Load() > int64(d.cfg.InstanceLimits.MaxInflightClientRequests) {
return nil, httpgrpc.Errorf(http.StatusServiceUnavailable, "too many inflight ingester client requests in distributor")
}
removeReplica := false
// Cache user limit with overrides so we spend less CPU doing locking. See issue #4904
limits := d.limits.GetOverridesForUser(userID)
if limits.AcceptHASamples && len(req.Timeseries) > 0 && !limits.AcceptMixedHASamples {
cluster, replica := findHALabels(limits.HAReplicaLabel, limits.HAClusterLabel, req.Timeseries[0].Labels)
removeReplica, err = d.checkSample(ctx, userID, cluster, replica, limits)
if err != nil {
// Ensure the request slice is reused if the series get deduped.
cortexpb.ReuseSlice(req.Timeseries)
if errors.Is(err, ha.ReplicasNotMatchError{}) {
// These samples have been deduped.
d.dedupedSamples.WithLabelValues(userID, cluster).Add(float64(numFloatSamples + numHistogramSamples))
return nil, httpgrpc.Errorf(http.StatusAccepted, "%s", err.Error())
}
if errors.Is(err, ha.TooManyReplicaGroupsError{}) {
d.validateMetrics.DiscardedSamples.WithLabelValues(validation.TooManyHAClusters, userID).Add(float64(numFloatSamples + numHistogramSamples))
return nil, httpgrpc.Errorf(http.StatusBadRequest, "%s", err.Error())
}
return nil, err
}
// If there wasn't an error but removeReplica is false that means we didn't find both HA labels.
if !removeReplica { // False, Nil
d.nonHASamples.WithLabelValues(userID).Add(float64(numFloatSamples + numHistogramSamples))
}
}
// A WriteRequest can only contain series or metadata but not both. This might change in the future.
seriesKeys, validatedTimeseries, validatedFloatSamples, validatedHistogramSamples, validatedExemplars, firstPartialErr, err := d.prepareSeriesKeys(ctx, req, userID, limits, removeReplica)
if err != nil {
return nil, err
}
metadataKeys, validatedMetadata, firstPartialErr := d.prepareMetadataKeys(req, limits, userID, firstPartialErr)
d.receivedSamples.WithLabelValues(userID, sampleMetricTypeFloat).Add(float64(validatedFloatSamples))
d.receivedSamples.WithLabelValues(userID, sampleMetricTypeHistogram).Add(float64(validatedHistogramSamples))
d.receivedExemplars.WithLabelValues(userID).Add(float64(validatedExemplars))
d.receivedMetadata.WithLabelValues(userID).Add(float64(len(validatedMetadata)))
if len(seriesKeys) == 0 && len(metadataKeys) == 0 {
// Ensure the request slice is reused if there's no series or metadata passing the validation.
cortexpb.ReuseSlice(req.Timeseries)
return &cortexpb.WriteResponse{}, firstPartialErr
}
totalSamples := validatedFloatSamples + validatedHistogramSamples
totalN := totalSamples + validatedExemplars + len(validatedMetadata)
if !d.ingestionRateLimiter.AllowN(now, userID, totalN) {
// Ensure the request slice is reused if the request is rate limited.
cortexpb.ReuseSlice(req.Timeseries)
d.validateMetrics.DiscardedSamples.WithLabelValues(validation.RateLimited, userID).Add(float64(totalSamples))
d.validateMetrics.DiscardedExemplars.WithLabelValues(validation.RateLimited, userID).Add(float64(validatedExemplars))
d.validateMetrics.DiscardedMetadata.WithLabelValues(validation.RateLimited, userID).Add(float64(len(validatedMetadata)))
// Return a 429 here to tell the client it is going too fast.
// Client may discard the data or slow down and re-send.
// Prometheus v2.26 added a remote-write option 'retry_on_http_429'.
return nil, httpgrpc.Errorf(http.StatusTooManyRequests, "ingestion rate limit (%v) exceeded while adding %d samples and %d metadata", d.ingestionRateLimiter.Limit(now, userID), totalSamples, len(validatedMetadata))
}
// totalN included samples and metadata. Ingester follows this pattern when computing its ingestion rate.
d.ingestionRate.Add(int64(totalN))
subRing := d.ingestersRing
// Obtain a subring if required.
if d.cfg.ShardingStrategy == util.ShardingStrategyShuffle {
subRing = d.ingestersRing.ShuffleShard(userID, limits.IngestionTenantShardSize)
}
keys := append(seriesKeys, metadataKeys...)
initialMetadataIndex := len(seriesKeys)
err = d.doBatch(ctx, req, subRing, keys, initialMetadataIndex, validatedMetadata, validatedTimeseries, userID)
if err != nil {
return nil, err
}
return &cortexpb.WriteResponse{}, firstPartialErr
}
func (d *Distributor) updateLabelSetMetrics() {
activeUserSet := make(map[string]map[uint64]struct{})
for _, user := range d.activeUsers.ActiveUsers() {
limits := d.limits.LimitsPerLabelSet(user)
activeUserSet[user] = make(map[uint64]struct{}, len(limits))
for _, l := range limits {
activeUserSet[user][l.Hash] = struct{}{}
}
}
d.labelSetTracker.UpdateMetrics(activeUserSet, func(user, labelSetStr string, removeUser bool) {
if removeUser {
if err := util.DeleteMatchingLabels(d.receivedSamplesPerLabelSet, map[string]string{"user": user}); err != nil {
level.Warn(d.log).Log("msg", "failed to remove cortex_distributor_received_samples_per_labelset_total metric for user", "user", user, "err", err)
}
return
}
d.receivedSamplesPerLabelSet.DeleteLabelValues(user, sampleMetricTypeFloat, labelSetStr)
d.receivedSamplesPerLabelSet.DeleteLabelValues(user, sampleMetricTypeHistogram, labelSetStr)
})
}
func (d *Distributor) cleanStaleIngesterMetrics() {
healthy, unhealthy, err := d.ingestersRing.GetAllInstanceDescs(ring.WriteNoExtend)
if err != nil {
level.Warn(d.log).Log("msg", "error cleaning metrics: GetAllInstanceDescs", "err", err)
return
}
idsMap := map[string]struct{}{}
for _, ing := range append(healthy, unhealthy...) {
if id, err := d.ingestersRing.GetInstanceIdByAddr(ing.Addr); err == nil {
idsMap[id] = struct{}{}
}
}
ingesterMetrics := []*prometheus.CounterVec{d.ingesterAppends, d.ingesterAppendFailures, d.ingesterQueries, d.ingesterQueryFailures}
for _, m := range ingesterMetrics {
metrics, err := util.GetLabels(m, make(map[string]string))
if err != nil {
level.Warn(d.log).Log("msg", "error cleaning metrics: GetLabels", "err", err)
return
}
for _, lbls := range metrics {
if _, ok := idsMap[lbls.Get("ingester")]; !ok {
err := util.DeleteMatchingLabels(m, map[string]string{"ingester": lbls.Get("ingester")})
if err != nil {
level.Warn(d.log).Log("msg", "error cleaning metrics: DeleteMatchingLabels", "err", err)
return
}
}
}
}
}
func (d *Distributor) doBatch(ctx context.Context, req *cortexpb.WriteRequest, subRing ring.ReadRing, keys []uint32, initialMetadataIndex int, validatedMetadata []*cortexpb.MetricMetadata, validatedTimeseries []cortexpb.PreallocTimeseries, userID string) error {
span, _ := opentracing.StartSpanFromContext(ctx, "doBatch")
defer span.Finish()
// Use a background context to make sure all ingesters get samples even if we return early
localCtx, cancel := context.WithTimeout(context.Background(), d.cfg.RemoteTimeout)
localCtx = user.InjectOrgID(localCtx, userID)
if sp := opentracing.SpanFromContext(ctx); sp != nil {
localCtx = opentracing.ContextWithSpan(localCtx, sp)
}
// Get any HTTP headers that are supposed to be added to logs and add to localCtx for later use
if headerMap := util_log.HeaderMapFromContext(ctx); headerMap != nil {
localCtx = util_log.ContextWithHeaderMap(localCtx, headerMap)
}
// Get clientIP(s) from Context and add it to localCtx
source := util.GetSourceIPsFromOutgoingCtx(ctx)
localCtx = util.AddSourceIPsToOutgoingContext(localCtx, source)
op := ring.WriteNoExtend
if d.cfg.ExtendWrites {
op = ring.Write
}
return ring.DoBatch(ctx, op, subRing, d.asyncExecutor, keys, func(ingester ring.InstanceDesc, indexes []int) error {
timeseries := make([]cortexpb.PreallocTimeseries, 0, len(indexes))
var metadata []*cortexpb.MetricMetadata
for _, i := range indexes {
if i >= initialMetadataIndex {
metadata = append(metadata, validatedMetadata[i-initialMetadataIndex])
} else {
timeseries = append(timeseries, validatedTimeseries[i])
}
}
return d.send(localCtx, ingester, timeseries, metadata, req.Source)
}, func() {
cortexpb.ReuseSlice(req.Timeseries)
cancel()
})
}
func (d *Distributor) prepareMetadataKeys(req *cortexpb.WriteRequest, limits *validation.Limits, userID string, firstPartialErr error) ([]uint32, []*cortexpb.MetricMetadata, error) {
validatedMetadata := make([]*cortexpb.MetricMetadata, 0, len(req.Metadata))
metadataKeys := make([]uint32, 0, len(req.Metadata))
for _, m := range req.Metadata {
err := validation.ValidateMetadata(d.validateMetrics, limits, userID, m)
if err != nil {
if firstPartialErr == nil {
firstPartialErr = err
}
continue
}
metadataKeys = append(metadataKeys, d.tokenForMetadata(userID, m.MetricFamilyName))
validatedMetadata = append(validatedMetadata, m)
}
return metadataKeys, validatedMetadata, firstPartialErr
}
type samplesLabelSetEntry struct {
floatSamples int64
histogramSamples int64
labels labels.Labels
}
func (d *Distributor) prepareSeriesKeys(ctx context.Context, req *cortexpb.WriteRequest, userID string, limits *validation.Limits, removeReplica bool) ([]uint32, []cortexpb.PreallocTimeseries, int, int, int, error, error) {
pSpan, _ := opentracing.StartSpanFromContext(ctx, "prepareSeriesKeys")
defer pSpan.Finish()
// For each timeseries or samples, we compute a hash to distribute across ingesters;
// check each sample/metadata and discard if outside limits.
validatedTimeseries := make([]cortexpb.PreallocTimeseries, 0, len(req.Timeseries))
seriesKeys := make([]uint32, 0, len(req.Timeseries))
validatedFloatSamples := 0
validatedHistogramSamples := 0
validatedExemplars := 0
limitsPerLabelSet := d.limits.LimitsPerLabelSet(userID)
var (
labelSetCounters map[uint64]*samplesLabelSetEntry
firstPartialErr error
)
latestSampleTimestampMs := int64(0)
defer func() {
// Update this metric even in case of errors.
if latestSampleTimestampMs > 0 {
d.latestSeenSampleTimestampPerUser.WithLabelValues(userID).Set(float64(latestSampleTimestampMs) / 1000)
}
}()
// For each timeseries, compute a hash to distribute across ingesters;
// check each sample and discard if outside limits.
skipLabelNameValidation := d.cfg.SkipLabelNameValidation || req.GetSkipLabelNameValidation()
for _, ts := range req.Timeseries {
if limits.AcceptHASamples && limits.AcceptMixedHASamples {
cluster, replica := findHALabels(limits.HAReplicaLabel, limits.HAClusterLabel, ts.Labels)
if cluster != "" && replica != "" {
_, err := d.checkSample(ctx, userID, cluster, replica, limits)
if err != nil {
// discard sample
if errors.Is(err, ha.ReplicasNotMatchError{}) {
// These samples have been deduped.
d.dedupedSamples.WithLabelValues(userID, cluster).Add(float64(len(ts.Samples) + len(ts.Histograms)))
}
if errors.Is(err, ha.TooManyReplicaGroupsError{}) {
d.validateMetrics.DiscardedSamples.WithLabelValues(validation.TooManyHAClusters, userID).Add(float64(len(ts.Samples) + len(ts.Histograms)))
}
continue
}
removeReplica = true // valid HA sample
} else {
removeReplica = false // non HA sample
d.nonHASamples.WithLabelValues(userID).Add(float64(len(ts.Samples) + len(ts.Histograms)))
}
}
// Use timestamp of latest sample in the series. If samples for series are not ordered, metric for user may be wrong.
if len(ts.Samples) > 0 {
latestSampleTimestampMs = max(latestSampleTimestampMs, ts.Samples[len(ts.Samples)-1].TimestampMs)
}
if len(ts.Histograms) > 0 {
latestSampleTimestampMs = max(latestSampleTimestampMs, ts.Histograms[len(ts.Histograms)-1].TimestampMs)
}
if mrc := limits.MetricRelabelConfigs; len(mrc) > 0 {
l, _ := relabel.Process(cortexpb.FromLabelAdaptersToLabels(ts.Labels), mrc...)