forked from influxdata/influxdb
/
index.go
956 lines (824 loc) · 25.7 KB
/
index.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
package tsi1
import (
"errors"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"regexp"
"runtime"
"strconv"
"sync"
"sync/atomic"
"unsafe"
"github.com/cespare/xxhash"
"github.com/influxdata/influxdb/models"
"github.com/influxdata/influxdb/pkg/estimator"
"github.com/influxdata/influxdb/pkg/estimator/hll"
"github.com/influxdata/influxdb/pkg/slices"
"github.com/influxdata/influxdb/tsdb"
"github.com/influxdata/influxql"
"go.uber.org/zap"
)
// IndexName is the name of the index.
const IndexName = "tsi1"
// ErrCompactionInterrupted is returned if compactions are disabled or
// an index is closed while a compaction is occurring.
var ErrCompactionInterrupted = errors.New("tsi1: compaction interrupted")
func init() {
// FIXME(edd): Remove this.
if os.Getenv("TSI_PARTITIONS") != "" {
i, err := strconv.Atoi(os.Getenv("TSI_PARTITIONS"))
if err != nil {
panic(err)
}
DefaultPartitionN = uint64(i)
}
tsdb.RegisterIndex(IndexName, func(_ uint64, db, path string, _ *tsdb.SeriesIDSet, sfile *tsdb.SeriesFile, opt tsdb.EngineOptions) tsdb.Index {
idx := NewIndex(sfile, db, WithPath(path), WithMaximumLogFileSize(int64(opt.Config.MaxIndexLogFileSize)))
return idx
})
}
// DefaultPartitionN determines how many shards the index will be partitioned into.
//
// NOTE: Currently, this must not be change once a database is created. Further,
// it must also be a power of 2.
//
var DefaultPartitionN uint64 = 8
// An IndexOption is a functional option for changing the configuration of
// an Index.
type IndexOption func(i *Index)
// WithPath sets the root path of the Index
var WithPath = func(path string) IndexOption {
return func(i *Index) {
i.path = path
}
}
// DisableCompactions disables compactions on the Index.
var DisableCompactions = func() IndexOption {
return func(i *Index) {
i.disableCompactions = true
}
}
// WithLogger sets the logger for the Index.
var WithLogger = func(l zap.Logger) IndexOption {
return func(i *Index) {
i.logger = l.With(zap.String("index", "tsi"))
}
}
// WithMaximumLogFileSize sets the maximum size of LogFiles before they're
// compacted into IndexFiles.
var WithMaximumLogFileSize = func(size int64) IndexOption {
return func(i *Index) {
i.maxLogFileSize = size
}
}
// DisableFsync disables flushing and syncing of underlying files. Primarily this
// impacts the LogFiles. This option can be set when working with the index in
// an offline manner, for cases where a hard failure can be overcome by re-running the tooling.
var DisableFsync = func() IndexOption {
return func(i *Index) {
i.disableFsync = true
}
}
// WithLogFileBufferSize sets the size of the buffer used within LogFiles.
// Typically appending an entry to a LogFile involves writing 11 or 12 bytes, so
// depending on how many new series are being created within a batch, it may
// be appropriate to set this.
var WithLogFileBufferSize = func(sz int) IndexOption {
return func(i *Index) {
if sz > 1<<17 { // 128K
sz = 1 << 17
} else if sz < 1<<12 {
sz = 1 << 12 // 4K (runtime default)
}
i.logfileBufferSize = sz
}
}
// Index represents a collection of layered index files and WAL.
type Index struct {
mu sync.RWMutex
partitions []*Partition
opened bool
// The following may be set when initializing an Index.
path string // Root directory of the index partitions.
disableCompactions bool // Initially disables compactions on the index.
maxLogFileSize int64 // Maximum size of a LogFile before it's compacted.
logfileBufferSize int // The size of the buffer used by the LogFile.
disableFsync bool // Disables flushing buffers and fsyning files. Used when working with indexes offline.
logger *zap.Logger // Index's logger.
// The following must be set when initializing an Index.
sfile *tsdb.SeriesFile // series lookup file
database string // Name of database.
// Index's version.
version int
// Number of partitions used by the index.
PartitionN uint64
}
func (i *Index) UniqueReferenceID() uintptr {
return uintptr(unsafe.Pointer(i))
}
// NewIndex returns a new instance of Index.
func NewIndex(sfile *tsdb.SeriesFile, database string, options ...IndexOption) *Index {
idx := &Index{
maxLogFileSize: tsdb.DefaultMaxIndexLogFileSize,
logger: zap.NewNop(),
version: Version,
sfile: sfile,
database: database,
PartitionN: DefaultPartitionN,
}
for _, option := range options {
option(idx)
}
return idx
}
// Bytes estimates the memory footprint of this Index, in bytes.
func (i *Index) Bytes() int {
var b int
i.mu.RLock()
b += 24 // mu RWMutex is 24 bytes
b += int(unsafe.Sizeof(i.partitions))
for _, p := range i.partitions {
b += int(unsafe.Sizeof(p)) + p.bytes()
}
b += int(unsafe.Sizeof(i.opened))
b += int(unsafe.Sizeof(i.path)) + len(i.path)
b += int(unsafe.Sizeof(i.disableCompactions))
b += int(unsafe.Sizeof(i.maxLogFileSize))
b += int(unsafe.Sizeof(i.logger))
b += int(unsafe.Sizeof(i.sfile))
// Do not count SeriesFile because it belongs to the code that constructed this Index.
b += int(unsafe.Sizeof(i.database)) + len(i.database)
b += int(unsafe.Sizeof(i.version))
b += int(unsafe.Sizeof(i.PartitionN))
i.mu.RUnlock()
return b
}
// Database returns the name of the database the index was initialized with.
func (i *Index) Database() string {
return i.database
}
// WithLogger sets the logger on the index after it's been created.
//
// It's not safe to call WithLogger after the index has been opened, or before
// it has been closed.
func (i *Index) WithLogger(l *zap.Logger) {
i.mu.Lock()
defer i.mu.Unlock()
i.logger = l.With(zap.String("index", "tsi"))
}
// Type returns the type of Index this is.
func (i *Index) Type() string { return IndexName }
// SeriesFile returns the series file attached to the index.
func (i *Index) SeriesFile() *tsdb.SeriesFile { return i.sfile }
// SeriesIDSet returns the set of series ids associated with series in this
// index. Any series IDs for series no longer present in the index are filtered out.
func (i *Index) SeriesIDSet() *tsdb.SeriesIDSet {
seriesIDSet := tsdb.NewSeriesIDSet()
others := make([]*tsdb.SeriesIDSet, 0, i.PartitionN)
for _, p := range i.partitions {
others = append(others, p.seriesIDSet)
}
seriesIDSet.Merge(others...)
return seriesIDSet
}
// Open opens the index.
func (i *Index) Open() error {
i.mu.Lock()
defer i.mu.Unlock()
if i.opened {
return errors.New("index already open")
}
// Ensure root exists.
if err := os.MkdirAll(i.path, 0777); err != nil {
return err
}
// Initialize index partitions.
i.partitions = make([]*Partition, i.PartitionN)
for j := 0; j < len(i.partitions); j++ {
p := NewPartition(i.sfile, filepath.Join(i.path, fmt.Sprint(j)))
p.MaxLogFileSize = i.maxLogFileSize
p.nosync = i.disableFsync
p.logbufferSize = i.logfileBufferSize
p.logger = i.logger.With(zap.String("tsi1_partition", fmt.Sprint(j+1)))
i.partitions[j] = p
}
// Open all the Partitions in parallel.
partitionN := len(i.partitions)
n := i.availableThreads()
// Store results.
errC := make(chan error, partitionN)
// Run fn on each partition using a fixed number of goroutines.
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func(k int) {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to work on.
if idx >= partitionN {
return // No more work.
}
err := i.partitions[idx].Open()
errC <- err
}
}(k)
}
// Check for error
for i := 0; i < partitionN; i++ {
if err := <-errC; err != nil {
return err
}
}
// Mark opened.
i.opened = true
i.logger.Info(fmt.Sprintf("index opened with %d partitions", partitionN))
return nil
}
// Compact requests a compaction of partitions.
func (i *Index) Compact() {
i.mu.Lock()
defer i.mu.Unlock()
for _, p := range i.partitions {
p.Compact()
}
}
func (i *Index) EnableCompactions() {
for _, p := range i.partitions {
p.EnableCompactions()
}
}
func (i *Index) DisableCompactions() {
for _, p := range i.partitions {
p.DisableCompactions()
}
}
// Wait blocks until all outstanding compactions have completed.
func (i *Index) Wait() {
for _, p := range i.partitions {
p.Wait()
}
}
// Close closes the index.
func (i *Index) Close() error {
// Lock index and close partitions.
i.mu.Lock()
defer i.mu.Unlock()
for _, p := range i.partitions {
if err := p.Close(); err != nil {
return err
}
}
// Mark index as closed.
i.opened = false
return nil
}
// Path returns the path the index was opened with.
func (i *Index) Path() string { return i.path }
// PartitionAt returns the partition by index.
func (i *Index) PartitionAt(index int) *Partition {
return i.partitions[index]
}
// partition returns the appropriate Partition for a provided series key.
func (i *Index) partition(key []byte) *Partition {
return i.partitions[int(xxhash.Sum64(key)&(i.PartitionN-1))]
}
// partitionIdx returns the index of the partition that key belongs in.
func (i *Index) partitionIdx(key []byte) int {
return int(xxhash.Sum64(key) & (i.PartitionN - 1))
}
// availableThreads returns the minimum of GOMAXPROCS and the number of
// partitions in the Index.
func (i *Index) availableThreads() int {
n := runtime.GOMAXPROCS(0)
if len(i.partitions) < n {
return len(i.partitions)
}
return n
}
// SetFieldSet sets a shared field set from the engine.
func (i *Index) SetFieldSet(fs *tsdb.MeasurementFieldSet) {
for _, p := range i.partitions {
p.SetFieldSet(fs)
}
}
// FieldSet returns the assigned fieldset.
func (i *Index) FieldSet() *tsdb.MeasurementFieldSet {
if len(i.partitions) == 0 {
return nil
}
return i.partitions[0].FieldSet()
}
// ForEachMeasurementName iterates over all measurement names in the index,
// applying fn. It returns the first error encountered, if any.
//
// ForEachMeasurementName does not call fn on each partition concurrently so the
// call may provide a non-goroutine safe fn.
func (i *Index) ForEachMeasurementName(fn func(name []byte) error) error {
itr, err := i.MeasurementIterator()
if err != nil {
return err
} else if itr == nil {
return nil
}
defer itr.Close()
// Iterate over all measurements.
for {
e, err := itr.Next()
if err != nil {
return err
} else if e == nil {
break
}
if err := fn(e); err != nil {
return err
}
}
return nil
}
// MeasurementExists returns true if a measurement exists.
func (i *Index) MeasurementExists(name []byte) (bool, error) {
n := i.availableThreads()
// Store errors
var found uint32 // Use this to signal we found the measurement.
errC := make(chan error, i.PartitionN)
// Check each partition for the measurement concurrently.
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to check
if idx >= len(i.partitions) {
return // No more work.
}
// Check if the measurement has been found. If it has don't
// need to check this partition and can just move on.
if atomic.LoadUint32(&found) == 1 {
errC <- nil
continue
}
b, err := i.partitions[idx].MeasurementExists(name)
if b {
atomic.StoreUint32(&found, 1)
}
errC <- err
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return false, err
}
}
// Check if we found the measurement.
return atomic.LoadUint32(&found) == 1, nil
}
// MeasurementHasSeries returns true if a measurement has non-tombstoned series.
func (i *Index) MeasurementHasSeries(name []byte) (bool, error) {
for _, p := range i.partitions {
if v, err := p.MeasurementHasSeries(name); err != nil {
return false, err
} else if v {
return true, nil
}
}
return false, nil
}
// fetchByteValues is a helper for gathering values from each partition in the index,
// based on some criteria.
//
// fn is a function that works on partition idx and calls into some method on
// the partition that returns some ordered values.
func (i *Index) fetchByteValues(fn func(idx int) ([][]byte, error)) ([][]byte, error) {
n := i.availableThreads()
// Store results.
names := make([][][]byte, i.PartitionN)
errC := make(chan error, i.PartitionN)
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to work on.
if idx >= len(i.partitions) {
return // No more work.
}
pnames, err := fn(idx)
// This is safe since there are no readers on names until all
// the writers are done.
names[idx] = pnames
errC <- err
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return nil, err
}
}
// It's now safe to read from names.
return slices.MergeSortedBytes(names[:]...), nil
}
// MeasurementIterator returns an iterator over all measurements.
func (i *Index) MeasurementIterator() (tsdb.MeasurementIterator, error) {
itrs := make([]tsdb.MeasurementIterator, 0, len(i.partitions))
for _, p := range i.partitions {
itr, err := p.MeasurementIterator()
if err != nil {
tsdb.MeasurementIterators(itrs).Close()
return nil, err
} else if itr != nil {
itrs = append(itrs, itr)
}
}
return tsdb.MergeMeasurementIterators(itrs...), nil
}
// MeasurementSeriesIDIterator returns an iterator over all series in a measurement.
func (i *Index) MeasurementSeriesIDIterator(name []byte) (tsdb.SeriesIDIterator, error) {
itrs := make([]tsdb.SeriesIDIterator, 0, len(i.partitions))
for _, p := range i.partitions {
itr, err := p.MeasurementSeriesIDIterator(name)
if err != nil {
tsdb.SeriesIDIterators(itrs).Close()
return nil, err
} else if itr != nil {
itrs = append(itrs, itr)
}
}
return tsdb.MergeSeriesIDIterators(itrs...), nil
}
// MeasurementNamesByRegex returns measurement names for the provided regex.
func (i *Index) MeasurementNamesByRegex(re *regexp.Regexp) ([][]byte, error) {
return i.fetchByteValues(func(idx int) ([][]byte, error) {
return i.partitions[idx].MeasurementNamesByRegex(re)
})
}
// DropMeasurement deletes a measurement from the index. It returns the first
// error encountered, if any.
func (i *Index) DropMeasurement(name []byte) error {
n := i.availableThreads()
// Store results.
errC := make(chan error, i.PartitionN)
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to work on.
if idx >= len(i.partitions) {
return // No more work.
}
errC <- i.partitions[idx].DropMeasurement(name)
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return err
}
}
return nil
}
// CreateSeriesListIfNotExists creates a list of series if they doesn't exist in bulk.
func (i *Index) CreateSeriesListIfNotExists(keys [][]byte, names [][]byte, tagsSlice []models.Tags) error {
// All slices must be of equal length.
if len(names) != len(tagsSlice) {
return errors.New("names/tags length mismatch in index")
}
// We need to move different series into collections for each partition
// to process.
pNames := make([][][]byte, i.PartitionN)
pTags := make([][]models.Tags, i.PartitionN)
// Determine partition for series using each series key.
for ki, key := range keys {
pidx := i.partitionIdx(key)
pNames[pidx] = append(pNames[pidx], names[ki])
pTags[pidx] = append(pTags[pidx], tagsSlice[ki])
}
// Process each subset of series on each partition.
n := i.availableThreads()
// Store errors.
errC := make(chan error, i.PartitionN)
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to work on.
if idx >= len(i.partitions) {
return // No more work.
}
errC <- i.partitions[idx].createSeriesListIfNotExists(pNames[idx], pTags[idx])
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return err
}
}
return nil
}
// CreateSeriesIfNotExists creates a series if it doesn't exist or is deleted.
func (i *Index) CreateSeriesIfNotExists(key, name []byte, tags models.Tags) error {
return i.partition(key).createSeriesListIfNotExists([][]byte{name}, []models.Tags{tags})
}
// InitializeSeries is a no-op. This only applies to the in-memory index.
func (i *Index) InitializeSeries(keys, names [][]byte, tags []models.Tags) error {
return nil
}
// DropSeries drops the provided series from the index. If cascade is true
// and this is the last series to the measurement, the measurment will also be dropped.
func (i *Index) DropSeries(seriesID uint64, key []byte, cascade bool) error {
// Remove from partition.
if err := i.partition(key).DropSeries(seriesID); err != nil {
return err
}
if !cascade {
return nil
}
// Extract measurement name.
name, _ := models.ParseKeyBytes(key)
// Check if that was the last series for the measurement in the entire index.
if ok, err := i.MeasurementHasSeries(name); err != nil {
return err
} else if ok {
return nil
}
// If no more series exist in the measurement then delete the measurement.
if err := i.DropMeasurement(name); err != nil {
return err
}
return nil
}
// DropMeasurementIfSeriesNotExist drops a measurement only if there are no more
// series for the measurment.
func (i *Index) DropMeasurementIfSeriesNotExist(name []byte) error {
// Check if that was the last series for the measurement in the entire index.
if ok, err := i.MeasurementHasSeries(name); err != nil {
return err
} else if ok {
return nil
}
// If no more series exist in the measurement then delete the measurement.
return i.DropMeasurement(name)
}
// MeasurementsSketches returns the two sketches for the index by merging all
// instances of the type sketch types in all the partitions.
func (i *Index) MeasurementsSketches() (estimator.Sketch, estimator.Sketch, error) {
s, ts := hll.NewDefaultPlus(), hll.NewDefaultPlus()
for _, p := range i.partitions {
// Get partition's measurement sketches and merge.
ps, pts, err := p.MeasurementsSketches()
if err != nil {
return nil, nil, err
}
if err := s.Merge(ps); err != nil {
return nil, nil, err
} else if err := ts.Merge(pts); err != nil {
return nil, nil, err
}
}
return s, ts, nil
}
// SeriesSketches returns the two sketches for the index by merging all
// instances of the type sketch types in all the partitions.
func (i *Index) SeriesSketches() (estimator.Sketch, estimator.Sketch, error) {
s, ts := hll.NewDefaultPlus(), hll.NewDefaultPlus()
for _, p := range i.partitions {
// Get partition's measurement sketches and merge.
ps, pts, err := p.SeriesSketches()
if err != nil {
return nil, nil, err
}
if err := s.Merge(ps); err != nil {
return nil, nil, err
} else if err := ts.Merge(pts); err != nil {
return nil, nil, err
}
}
return s, ts, nil
}
// Since indexes are not shared across shards, the count returned by SeriesN
// cannot be combined with other shard's results. If you need to count series
// across indexes then use either the database-wide series file, or merge the
// index-level bitsets or sketches.
func (i *Index) SeriesN() int64 {
return int64(i.SeriesIDSet().Cardinality())
}
// HasTagKey returns true if tag key exists. It returns the first error
// encountered if any.
func (i *Index) HasTagKey(name, key []byte) (bool, error) {
n := i.availableThreads()
// Store errors
var found uint32 // Use this to signal we found the tag key.
errC := make(chan error, i.PartitionN)
// Check each partition for the tag key concurrently.
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to check
if idx >= len(i.partitions) {
return // No more work.
}
// Check if the tag key has already been found. If it has, we
// don't need to check this partition and can just move on.
if atomic.LoadUint32(&found) == 1 {
errC <- nil
continue
}
b, err := i.partitions[idx].HasTagKey(name, key)
if b {
atomic.StoreUint32(&found, 1)
}
errC <- err
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return false, err
}
}
// Check if we found the tag key.
return atomic.LoadUint32(&found) == 1, nil
}
// HasTagValue returns true if tag value exists.
func (i *Index) HasTagValue(name, key, value []byte) (bool, error) {
n := i.availableThreads()
// Store errors
var found uint32 // Use this to signal we found the tag key.
errC := make(chan error, i.PartitionN)
// Check each partition for the tag key concurrently.
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to check
if idx >= len(i.partitions) {
return // No more work.
}
// Check if the tag key has already been found. If it has, we
// don't need to check this partition and can just move on.
if atomic.LoadUint32(&found) == 1 {
errC <- nil
continue
}
b, err := i.partitions[idx].HasTagValue(name, key, value)
if b {
atomic.StoreUint32(&found, 1)
}
errC <- err
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return false, err
}
}
// Check if we found the tag key.
return atomic.LoadUint32(&found) == 1, nil
}
// TagKeyIterator returns an iterator for all keys across a single measurement.
func (i *Index) TagKeyIterator(name []byte) (tsdb.TagKeyIterator, error) {
a := make([]tsdb.TagKeyIterator, 0, len(i.partitions))
for _, p := range i.partitions {
itr := p.TagKeyIterator(name)
if itr != nil {
a = append(a, itr)
}
}
return tsdb.MergeTagKeyIterators(a...), nil
}
// TagValueIterator returns an iterator for all values across a single key.
func (i *Index) TagValueIterator(name, key []byte) (tsdb.TagValueIterator, error) {
a := make([]tsdb.TagValueIterator, 0, len(i.partitions))
for _, p := range i.partitions {
itr := p.TagValueIterator(name, key)
if itr != nil {
a = append(a, itr)
}
}
return tsdb.MergeTagValueIterators(a...), nil
}
// TagKeySeriesIDIterator returns a series iterator for all values across a single key.
func (i *Index) TagKeySeriesIDIterator(name, key []byte) (tsdb.SeriesIDIterator, error) {
a := make([]tsdb.SeriesIDIterator, 0, len(i.partitions))
for _, p := range i.partitions {
itr := p.TagKeySeriesIDIterator(name, key)
if itr != nil {
a = append(a, itr)
}
}
return tsdb.MergeSeriesIDIterators(a...), nil
}
// TagValueSeriesIDIterator returns a series iterator for a single tag value.
func (i *Index) TagValueSeriesIDIterator(name, key, value []byte) (tsdb.SeriesIDIterator, error) {
a := make([]tsdb.SeriesIDIterator, 0, len(i.partitions))
for _, p := range i.partitions {
itr := p.TagValueSeriesIDIterator(name, key, value)
if itr != nil {
a = append(a, itr)
}
}
return tsdb.MergeSeriesIDIterators(a...), nil
}
// MeasurementTagKeysByExpr extracts the tag keys wanted by the expression.
func (i *Index) MeasurementTagKeysByExpr(name []byte, expr influxql.Expr) (map[string]struct{}, error) {
n := i.availableThreads()
// Store results.
keys := make([]map[string]struct{}, i.PartitionN)
errC := make(chan error, i.PartitionN)
var pidx uint32 // Index of maximum Partition being worked on.
for k := 0; k < n; k++ {
go func() {
for {
idx := int(atomic.AddUint32(&pidx, 1) - 1) // Get next partition to work on.
if idx >= len(i.partitions) {
return // No more work.
}
// This is safe since there are no readers on keys until all
// the writers are done.
tagKeys, err := i.partitions[idx].MeasurementTagKeysByExpr(name, expr)
keys[idx] = tagKeys
errC <- err
}
}()
}
// Check for error
for i := 0; i < cap(errC); i++ {
if err := <-errC; err != nil {
return nil, err
}
}
// Merge into single map.
result := keys[0]
for k := 1; k < len(i.partitions); k++ {
for k := range keys[k] {
result[k] = struct{}{}
}
}
return result, nil
}
// DiskSizeBytes returns the size of the index on disk.
func (i *Index) DiskSizeBytes() int64 {
fs, err := i.RetainFileSet()
if err != nil {
i.logger.Warn("Index is closing down")
return 0
}
defer fs.Release()
var manifestSize int64
// Get MANIFEST sizes from each partition.
for _, p := range i.partitions {
manifestSize += p.manifestSize
}
return fs.Size() + manifestSize
}
// TagKeyCardinality always returns zero.
// It is not possible to determine cardinality of tags across index files, and
// thus it cannot be done across partitions.
func (i *Index) TagKeyCardinality(name, key []byte) int {
return 0
}
// RetainFileSet returns the set of all files across all partitions.
// This is only needed when all files need to be retained for an operation.
func (i *Index) RetainFileSet() (*FileSet, error) {
i.mu.RLock()
defer i.mu.RUnlock()
fs, _ := NewFileSet(nil, i.sfile, nil)
for _, p := range i.partitions {
pfs, err := p.RetainFileSet()
if err != nil {
fs.Close()
return nil, err
}
fs.files = append(fs.files, pfs.files...)
}
return fs, nil
}
// SetFieldName is a no-op on this index.
func (i *Index) SetFieldName(measurement []byte, name string) {}
// Rebuild rebuilds an index. It's a no-op for this index.
func (i *Index) Rebuild() {}
// IsIndexDir returns true if directory contains at least one partition directory.
func IsIndexDir(path string) (bool, error) {
fis, err := ioutil.ReadDir(path)
if err != nil {
return false, err
}
for _, fi := range fis {
if !fi.IsDir() {
continue
} else if ok, err := IsPartitionDir(filepath.Join(path, fi.Name())); err != nil {
return false, err
} else if ok {
return true, nil
}
}
return false, nil
}