Repository navigation
Expand file tree
/
Copy pathdata-scanner.go
More file actions
1401 lines (1267 loc) · 43 KB
/
Copy pathdata-scanner.go
File metadata and controls
1401 lines (1267 loc) · 43 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Copyright (c) 2015-2021 MinIO, Inc.
//
// This file is part of MinIO Object Storage stack
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU Affero General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Affero General Public License for more details.
//
// You should have received a copy of the GNU Affero General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
package cmd
import (
"bytes"
"context"
"encoding/binary"
"errors"
"math"
"math/rand"
"net/http"
"os"
"path"
"strings"
"sync"
"time"
"github.com/bits-and-blooms/bloom/v3"
"github.com/minio/madmin-go"
"github.com/minio/minio/internal/bucket/lifecycle"
"github.com/minio/minio/internal/bucket/replication"
"github.com/minio/minio/internal/color"
"github.com/minio/minio/internal/config/heal"
"github.com/minio/minio/internal/event"
"github.com/minio/minio/internal/hash"
"github.com/minio/minio/internal/logger"
"github.com/minio/pkg/console"
)
const (
dataScannerSleepPerFolder = time.Millisecond // Time to wait between folders.
dataUsageUpdateDirCycles = 16 // Visit all folders every n cycles.
dataScannerCompactLeastObject = 500 // Compact when there is less than this many objects in a branch.
dataScannerCompactAtChildren = 10000 // Compact when there are this many children in a branch.
dataScannerCompactAtFolders = dataScannerCompactAtChildren / 4 // Compact when this many subfolders in a single folder.
dataScannerStartDelay = 1 * time.Minute // Time to wait on startup and between cycles.
healDeleteDangling = true
healFolderIncludeProb = 32 // Include a clean folder one in n cycles.
healObjectSelectProb = 512 // Overall probability of a file being scanned; one in n.
)
var (
globalHealConfig heal.Config
globalHealConfigMu sync.Mutex
dataScann
8000
erLeaderLockTimeout = newDynamicTimeout(30*time.Second, 10*time.Second)
// Sleeper values are updated when config is loaded.
scannerSleeper = newDynamicSleeper(10, 10*time.Second)
scannerCycle = &safeDuration{
t: dataScannerStartDelay,
}
)
// initDataScanner will start the scanner in the background.
func initDataScanner(ctx context.Context, objAPI ObjectLayer) {
go runDataScanner(ctx, objAPI)
}
type safeDuration struct {
sync.Mutex
t time.Duration
}
func (s *safeDuration) Update(t time.Duration) {
s.Lock()
defer s.Unlock()
s.t = t
}
func (s *safeDuration) Get() time.Duration {
s.Lock()
defer s.Unlock()
return s.t
}
// runDataScanner will start a data scanner.
// The function will block until the context is canceled.
// There should only ever be one scanner running per cluster.
func runDataScanner(pctx context.Context, objAPI ObjectLayer) {
// Make sure only 1 scanner is running on the cluster.
locker := objAPI.NewNSLock(minioMetaBucket, "runDataScanner.lock")
var ctx context.Context
r := rand.New(rand.NewSource(time.Now().UnixNano()))
for {
lkctx, err := locker.GetLock(pctx, dataScannerLeaderLockTimeout)
if err != nil {
time.Sleep(time.Duration(r.Float64() * float64(scannerCycle.Get())))
continue
}
ctx = lkctx.Context()
defer lkctx.Cancel()
break
// No unlock for "leader" lock.
}
// Load current bloom cycle
nextBloomCycle := intDataUpdateTracker.current() + 1
br, err := objAPI.GetObjectNInfo(ctx, dataUsageBucket, dataUsageBloomName, nil, http.Header{}, readLock, ObjectOptions{})
if err != nil {
if !isErrObjectNotFound(err) && !isErrBucketNotFound(err) {
logger.LogIf(ctx, err)
}
} else {
if br.ObjInfo.Size == 8 {
if err = binary.Read(br, binary.LittleEndian, &nextBloomCycle); err != nil {
logger.LogIf(ctx, err)
}
}
br.Close()
}
scannerTimer := time.NewTimer(scannerCycle.Get())
defer scannerTimer.Stop()
for {
select {
case <-ctx.Done():
return
case <-scannerTimer.C:
// Reset the timer for next cycle.
scannerTimer.Reset(scannerCycle.Get())
if intDataUpdateTracker.debug {
console.Debugln("starting scanner cycle")
}
// Wait before starting next cycle and wait on startup.
results := make(chan madmin.DataUsageInfo, 1)
go storeDataUsageInBackend(ctx, objAPI, results)
bf, err := globalNotificationSys.updateBloomFilter(ctx, nextBloomCycle)
logger.LogIf(ctx, err)
err = objAPI.NSScanner(ctx, bf, results)
logger.LogIf(ctx, err)
if err == nil {
// Store new cycle...
nextBloomCycle++
var tmp [8]byte
binary.LittleEndian.PutUint64(tmp[:], nextBloomCycle)
r, err := hash.NewReader(bytes.NewReader(tmp[:]), int64(len(tmp)), "", "", int64(len(tmp)))
if err != nil {
logger.LogIf(ctx, err)
continue
}
_, err = objAPI.PutObject(ctx, dataUsageBucket, dataUsageBloomName, NewPutObjReader(r), ObjectOptions{})
if !isErrBucketNotFound(err) {
logger.LogIf(ctx, err)
}
}
}
}
}
type cachedFolder struct {
name string
parent *dataUsageHash
objectHealProbDiv uint32
}
type folderScanner struct {
root string
getSize getSizeFn
oldCache dataUsageCache
newCache dataUsageCache
updateCache dataUsageCache
withFilter *bloomFilter
dataUsageScannerDebug bool
healFolderInclude uint32 // Include a clean folder one in n cycles.
healObjectSelect uint32 // Do a heal check on an object once every n cycles. Must divide into healFolderInclude
disks []StorageAPI
// If set updates will be sent regularly to this channel.
// Will not be closed when returned.
updates chan<- dataUsageEntry
8BD0
lastUpdate time.Time
}
// Cache structure and compaction:
//
// A cache structure will be kept with a tree of usages.
// The cache is a tree structure where each keeps track of its children.
//
// An uncompacted branch contains a count of the files only directly at the
// branch level, and contains link to children branches or leaves.
//
// The leaves are "compacted" based on a number of properties.
// A compacted leaf contains the totals of all files beneath it.
//
// A leaf is only scanned once every dataUsageUpdateDirCycles,
// rarer if the bloom filter for the path is clean and no lifecycles are applied.
// Skipped leaves have their totals transferred from the previous cycle.
//
// A clean leaf will be included once every healFolderIncludeProb for partial heal scans.
// When selected there is a one in healObjectSelectProb that any object will be chosen for heal scan.
//
// Compaction happens when either:
//
// 1) The folder (and subfolders) contains less than dataScannerCompactLeastObject objects.
// 2) The folder itself contains more than dataScannerCompactAtFolders folders.
// 3) The folder only contains objects and no subfolders.
//
// A bucket root will never be compacted.
//
// Furthermore if a has more than dataScannerCompactAtChildren recursive children (uncompacted folders)
// the tree will be recursively scanned and the branches with the least number of objects will be
// compacted until the limit is reached.
//
// This ensures that any branch will never contain an unreasonable amount of other branches,
// and also that small branches with few objects don't take up unreasonable amounts of space.
// This keeps the cache size at a reasonable size for all buckets.
//
// Whenever a branch is scanned, it is assumed that it will be un-compacted
// before it hits any of the above limits.
// This will make the branch rebalance itself when scanned if the distribution of objects has changed.
// scanDataFolder will scanner the basepath+cache.Info.Name and return an updated cache.
// The returned cache will always be valid, but may not be updated from the existing.
// Before each operation sleepDuration is called which can be used to temporarily halt the scanner.
// If the supplied context is canceled the function will return at the first chance.
func scanDataFolder(ctx context.Context, basePath string, cache dataUsageCache, getSize getSizeFn) (dataUsageCache, error) {
t := UTCNow()
logPrefix := color.Green("data-usage: ")
logSuffix := color.Blue("- %v + %v", basePath, cache.Info.Name)
if intDataUpdateTracker.debug {
defer func() {
console.Debugf(logPrefix+" Scanner time: %v %s\n", time.Since(t), logSuffix)
}()
}
switch cache.Info.Name {
case "", dataUsageRoot:
return cache, errors.New("internal error: root scan attempted")
}
s := folderScanner{
root: basePath,
getSize: getSize,
oldCache: cache,
newCache: dataUsageCache{Info: cache.Info},
updateCache: dataUsageCache{Info: cache.Info},
dataUsageScannerDebug: intDataUpdateTracker.debug,
healFolderInclude: 0,
healObjectSelect: 0,
updates: cache.Info.updates,
}
// Add disks for set healing.
if len(cache.Disks) > 0 {
objAPI, ok := newObjectLayerFn().(*erasureServerPools)
if ok {
s.disks = objAPI.GetDisksID(cache.Disks...)
if len(s.disks) != len(cache.Disks) {
console.Debugf(logPrefix+"Missing disks, want %d, found %d. Cannot heal. %s\n", len(cache.Disks), len(s.disks), logSuffix)
s.disks = s.disks[:0]
}
}
}
// Enable healing in XL mode.
if globalIsErasure && !cache.Info.SkipHealing {
// Include a clean folder one in n cycles.
s.healFolderInclude = healFolderIncludeProb
// Do a heal check on an object once every n cycles. Must divide into healFolderInclude
s.healObjectSelect = healObjectSelectProb
}
if len(cache.Info.BloomFilter) > 0 {
s.withFilter = &bloomFilter{BloomFilter: &bloom.BloomFilter{}}
_, err := s.withFilter.ReadFrom(bytes.NewReader(cache.Info.BloomFilter))
if err != nil {
logger.LogIf(ctx, err, logPrefix+"Error reading bloom filter")
s.withFilter = nil
}
}
if s.dataUsageScannerDebug {
console.Debugf(logPrefix+"Start scanning. Bloom filter: %v %s\n", s.withFilter != nil, logSuffix)
}
done := ctx.Done()
if s.dataUsageScannerDebug {
console.Debugf(logPrefix+"Cycle: %v, Entries: %v %s\n", cache.Info.NextCycle, len(cache.Cache), logSuffix)
}
// Read top level in bucket.
select {
case <-done:
return cache, ctx.Err()
default:
}
root := dataUsageEntry{}
folder := cachedFolder{name: cache.Info.Name, objectHealProbDiv: 1}
err := s.scanFolder(ctx, folder, &root)
if err != nil {
// No useful information...
return cache, err
}
if s.dataUsageScannerDebug {
console.Debugf(logPrefix+"Finished scanner, %v entries (%+v) %s \n", len(s.newCache.Cache), *s.newCache.sizeRecursive(s.newCache.Info.Name), logSuffix)
}
s.newCache.Info.LastUpdate = UTCNow()
s.newCache.Info.NextCycle++
return s.newCache, nil
}
// sendUpdate() should be called on a regular basis when the newCache contains more recent total than previously.
// May or may not send an update upstream.
func (f *folderScanner) sendUpdate() {
// Send at most an update every minute.
if f.updates == nil || time.Since(f.lastUpdate) < time.Minute {
return
}
if flat := f.updateCache.sizeRecursive(f.newCache.Info.Name); flat != nil {
select {
case f.updates <- *flat:
default:
}
f.lastUpdate = time.Now()
}
}
// scanFolder will scan the provided folder.
// Files found in the folders will be added to f.newCache.
// If final is provided folders will be put into f.newFolders or f.existingFolders.
// If final is not provided the folders found are returned from the function.
func (f *folderScanner) scanFolder(ctx context.Context, folder cachedFolder, into *dataUsageEntry) error {
done := ctx.Done()
scannerLogPrefix := color.Green("folder-scanner:")
thisHash := hashPath(folder.name)
// Store initial compaction state.
wasCompacted := into.Compacted
for {
select {
case <-done:
return ctx.Err()
default:
}
existing, ok := f.oldCache.Cache[thisHash.Key()]
var abandonedChildren dataUsageHashMap
if !into.Compacted {
abandonedChildren = f.oldCache.findChildrenCopy(thisHash)
}
// If there are lifecycle rules for the prefix, remove the filter.
filter := f.withFilter
_, prefix := path2BucketObjectWithBasePath(f.root, folder.name)
var activeLifeCycle *lifecycle.Lifecycle
if f.oldCache.Info.lifeCycle != nil && f.oldCache.Info.lifeCycle.HasActiveRules(prefix, true) {
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" Prefix %q has active rules\n", prefix)
}
activeLifeCycle = f.oldCache.Info.lifeCycle
filter = nil
}
// If there are replication rules for the prefix, remove the filter.
var replicationCfg replicationConfig
if !f.oldCache.Info.replication.Empty() && f.oldCache.Info.replication.Config.HasActiveRules(prefix, true) {
replicationCfg = f.oldCache.Info.replication
filter = nil
}
// Check if we can skip it due to bloom filter...
if filter != nil && ok && existing.Compacted {
// If folder isn't in filter and we have data, skip it completely.
if folder.name != dataUsageRoot && !filter.containsDir(folder.name) {
if f.healObjectSelect == 0 || !thisHash.mod(f.oldCache.Info.NextCycle, f.healFolderInclude/folder.objectHealProbDiv) {
f.newCache.copyWithChildren(&f.oldCache, thisHash, folder.parent)
f.updateCache.copyWithChildren(&f.oldCache, thisHash, folder.parent)
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" Skipping non-updated folder: %v\n", folder.name)
}
return nil
}
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" Adding non-updated folder to heal check: %v\n", folder.name)
}
// If probability was already scannerHealFolderInclude, keep it.
folder.objectHealProbDiv = f.healFolderInclude
}
}
scannerSleeper.Sleep(ctx, dataScannerSleepPerFolder)
var existingFolders, newFolders []cachedFolder
var foundObjects bool
err := readDirFn(path.Join(f.root, folder.name), func(entName string, typ os.FileMode) error {
// Parse
entName = pathClean(path.Join(folder.name, entName))
if entName == "" {
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" no bucket (%s,%s)\n", f.root, entName)
}
return errDoneForNow
}
bucket, prefix := path2BucketObjectWithBasePath(f.root, entName)
if bucket == "" {
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" no bucket (%s,%s)\n", f.root, entName)
}
return errDoneForNow
}
if isReservedOrInvalidBucket(bucket, false) {
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" invalid bucket: %v, entry: %v\n", bucket, entName)
}
return errDoneForNow
}
select {
case <-done:
return errDoneForNow
default:
}
if typ&os.ModeDir != 0 {
h := hashPath(entName)
_, exists := f.oldCache.Cache[h.Key()]
this := cachedFolder{name: entName, parent: &thisHash, objectHealProbDiv: folder.objectHealProbDiv}
delete(abandonedChildren, h.Key()) // h.Key() already accounted for.
if exists {
existingFolders = append(existingFolders, this)
f.updateCache.copyWithChildren(&f.oldCache, h, &thisHash)
} else {
newFolders = append(newFolders, this)
}
return nil
}
// Dynamic time delay.
wait := scannerSleeper.Timer(ctx)
// Get file size, ignore errors.
item := scannerItem{
Path: path.Join(f.root, entName),
Typ: typ,
bucket: bucket,
prefix: path.Dir(prefix),
objectName: path.Base(entName),
debug: f.dataUsageScannerDebug,
lifeCycle: activeLifeCycle,
replication: replicationCfg,
heal: thisHash.mod(f.oldCache.Info.NextCycle, f.healObjectSelect/folder.objectHealProbDiv) && globalIsErasure,
}
// if the drive belongs to an erasure set
// that is already being healed, skip the
// healing attempt on this drive.
item.heal = item.heal && f.healObjectSelect > 0
sz, err := f.getSize(item)
if err == errSkipFile {
wait() // wait to proceed to next entry.
return nil
}
// successfully read means we have a valid object.
foundObjects = true
// Remove filename i.e is the meta file to construct object name
item.transformMetaDir()
// Object already accounted for, remove from heal map,
// simply because getSize() function already heals the
// object.
delete(abandonedChildren, path.Join(item.bucket, item.objectPath()))
into.addSizes(sz)
into.Objects++
wait() // wait to proceed to next entry.
return nil
})
if err != nil {
return err
}
if foundObjects && globalIsErasure {
// If we found an object in erasure mode, we skip subdirs (only datadirs)...
break
}
// If we have many subfolders, compact ourself.
if !into.Compacted &&
f.newCache.Info.Name != folder.name &&
len(existingFolders)+len(newFolders) >= dataScannerCompactAtFolders {
into.Compacted = true
newFolders = append(newFolders, existingFolders...)
existingFolders = nil
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" Preemptively compacting: %v, entries: %v\n", folder.name, len(existingFolders)+len(newFolders))
}
}
scanFolder := func(folder cachedFolder) {
if contextCanceled(ctx) {
return
}
dst := into
if !into.Compacted {
dst = &dataUsageEntry{Compacted: false}
}
if err := f.scanFolder(ctx, folder, dst); err != nil {
logger.LogIf(ctx, err)
return
}
if !into.Compacted {
into.addChild(dataUsageHash(folder.name))
}
// We scanned a folder, optionally send update.
f.sendUpdate()
}
// Scan new...
for _, folder := range newFolders {
h := hashPath(folder.name)
// Add new folders to the update tree so totals update for these.
if !into.Compacted {
var foundAny bool
parent := thisHash
for parent != hashPath(f.updateCache.Info.Name) {
e := f.updateCache.find(parent.Key())
if e == nil || e.Compacted {
foundAny = true
break
}
if next := f.updateCache.searchParent(parent); next == nil {
foundAny = true
break
} else {
parent = *next
}
}
if !foundAny {
// Add non-compacted empty entry.
f.updateCache.replaceHashed(h, &thisHash, dataUsageEntry{})
}
}
scanFolder(folder)
// Add new folders if this is new and we don't have existing.
if !into.Compacted {
parent := f.updateCache.find(thisHash.Key())
if parent != nil && !parent.Compacted {
f.updateCache.deleteRecursive(h)
f.updateCache.copyWithChildren(&f.newCache, h, &thisHash)
}
}
}
// Scan existing...
for _, folder := range existingFolders {
h := hashPath(folder.name)
// Check if we should skip scanning folder...
// We can only skip if we are not indexing into a compacted destination
// and the entry itself is compacted.
if !into.Compacted && f.oldCache.isCompacted(h) {
if !h.mod(f.oldCache.Info.NextCycle, dataUsageUpdateDirCycles) {
if f.healObjectSelect == 0 || !h.mod(f.oldCache.Info.NextCycle, f.healFolderInclude/folder.objectHealProbDiv) {
// Transfer and add as child...
f.newCache.copyWithChildren(&f.oldCache, h, folder.parent)
into.addChild(h)
continue
}
folder.objectHealProbDiv = dataUsageUpdateDirCycles
}
}
scanFolder(folder)
}
// Scan for healing
if f.healObjectSelect == 0 || len(abandonedChildren) == 0 {
// If we are not heal scanning, return now.
break
}
objAPI, ok := newObjectLayerFn().(*erasureServerPools)
if !ok || len(f.disks) == 0 {
break
}
bgSeq, found := globalBackgroundHealState.getHealSequenceByToken(bgHealingUUID)
if !found {
break
}
// Whatever remains in 'abandonedChildren' are folders at this level
// that existed in the previous run but wasn't found now.
//
// This may be because of 2 reasons:
//
// 1) The folder/object was deleted.
// 2) We come from another disk and this disk missed the write.
//
// We therefore perform a heal check.
// If that doesn't bring it back we remove the folder and assume it was deleted.
// This means that the next run will not look for it.
// How to resolve results.
resolver := metadataResolutionParams{
dirQuorum: getReadQuorum(len(f.disks)),
objQuorum: getReadQuorum(len(f.disks)),
bucket: "",
}
healObjectsPrefix := color.Green("healObjects:")
for k := range abandonedChildren {
bucket, prefix := path2BucketObject(k)
if f.dataUsageScannerDebug {
console.Debugf(scannerLogPrefix+" checking disappeared folder: %v/%v\n", bucket, prefix)
}
resolver.bucket = bucket
foundObjs := false
dangling := false
ctx, cancel := context.WithCancel(ctx)
err := listPathRaw(ctx, listPathRawOptions{
disks: f.disks,
bucket: bucket,
path: prefix,
recursive: true,
reportNotFound: true,
minDisks: len(f.disks), // We want full consistency.
// Weird, maybe transient error.
agreed: func(entry metaCacheEntry) {
if f.dataUsageScannerDebug {
console.Debugf(healObjectsPrefix+" got agreement: %v\n", entry.name)
}
},
// Some disks have data for this.
partial: func(entries metaCacheEntries, nAgreed int, errs []error) {
if f.dataUsageScannerDebug {
console.Debugf(healObjectsPrefix+" got partial, %d agreed, errs: %v\n", nAgreed, errs)
}
// agreed value less than expected quorum
dangling = nAgreed < resolver.objQuorum || nAgreed < resolver.dirQuorum
entry, ok := entries.resolve(&resolver)
if !ok {
for _, err := range errs {
if err != nil {
return
}
}
// If no errors, queue it for healing.
entry, _ = entries.firstFound()
}
if f.dataUsageScannerDebug {
console.Debugf(healObjectsPrefix+" resolved to: %v, dir: %v\n", entry.name, entry.isDir())
}
if entry.isDir() {
return
}
// wait on timer per object.
wait := scannerSleeper.Timer(ctx)
// We got an entry which we should be able to heal.
fiv, err := entry.fileInfoVersions(bucket)
if err != nil {
wait()
err := bgSeq.queueHealTask(healSource{
bucket: bucket,
object: entry.name,
versionID: "",
}, madmin.HealItemObject)
if !isErrObjectNotFound(err) && !isErrVersionNotFound(err) {
logger.LogIf(ctx, err)
}
foundObjs = foundObjs || err == nil
return
}
for _, ver := range fiv.Versions {
// Sleep and reset.
wait()
wait = scannerSleeper.Timer(ctx)
err := bgSeq.queueHealTask(healSource{
bucket: bucket,
object: fiv.Name,
versionID: ver.VersionID,
}, madmin.HealItemObject)
if !isErrObjectNotFound(err) && !isErrVersionNotFound(err) {
logger.LogIf(ctx, err)
}
foundObjs = foundObjs || err == nil
}
},
// Too many disks failed.
finished: func(errs []error) {
if f.dataUsageScannerDebug {
console.Debugf(healObjectsPrefix+" too many errors: %v\n", errs)
}
cancel()
},
})
if f.dataUsageScannerDebug && err != nil && err != errFileNotFound {
console.Debugf(healObjectsPrefix+" checking returned value %v (%T)\n", err, err)
}
// If we found one or more disks with this folder, delete it.
if err == nil && dangling {
if f.dataUsageScannerDebug {
console.Debugf(healObjectsPrefix+" deleting dangling directory %s\n", prefix)
}
// wait on timer per object.
wait := scannerSleeper.Timer(ctx)
objAPI.HealObjects(ctx, bucket, prefix, madmin.HealOpts{
Recursive: true,
Remove: healDeleteDangling,
}, func(bucket, object, versionID string) error {
// Wait for each heal as per scanner frequency.
wait()
wait = scannerSleeper.Timer(ctx)
return bgSeq.queueHealTask(healSource{
bucket: bucket,
object: object,
versionID: versionID,
}, madmin.HealItemObject)
})
}
// Add unless healing returned an error.
if foundObjs {
this := cachedFolder{name: k, parent: &thisHash, objectHealProbDiv: 1}
scanFolder(this)
}
}
break
}
if !wasCompacted {
f.newCache.replaceHashed(thisHash, folder.parent, *into)
}
if !into.Compacted && f.newCache.Info.Name != folder.name {
flat := f.newCache.sizeRecursive(thisHash.Key())
flat.Compacted = true
var compact bool
if flat.Objects < dataScannerCompactLeastObject {
if f.dataUsageScannerDebug && flat.Objects > 1 {
// Disabled, rather chatty:
//console.Debugf(scannerLogPrefix+" Only %d objects, compacting %s -> %+v\n", flat.Objects, folder.name, flat)
}
compact = true
} else {
// Compact if we only have objects as children...
compact = true
for k := range into.Children {
if v, ok := f.newCache.Cache[k]; ok {
if len(v.Children) > 0 || v.Objects > 1 {
compact = false
break
}
}
}
if f.dataUsageScannerDebug && compact {
// Disabled, rather chatty:
//console.Debugf(scannerLogPrefix+" Only objects (%d), compacting %s -> %+v\n", flat.Objects, folder.name, flat)
}
}
if compact {
f.newCache.deleteRecursive(thisHash)
f.newCache.replaceHashed(thisHash, folder.parent, *flat)
}
}
// Compact if too many children...
if !into.Compacted {
f.newCache.reduceChildrenOf(thisHash, dataScannerCompactAtChildren, f.newCache.Info.Name != folder.name)
}
if _, ok := f.updateCache.Cache[thisHash.Key()]; !wasCompacted && ok {
// Replace if existed before.
if flat := f.newCache.sizeRecursive(thisHash.Key()); flat != nil {
f.updateCache.deleteRecursive(thisHash)
f.updateCache.replaceHashed(thisHash, folder.parent, *flat)
}
}
return nil
}
// scannerItem represents each file while walking.
type scannerItem struct {
Path string
Typ os.FileMode
bucket string // Bucket.
prefix string // Only the prefix if any, does not have final object name.
objectName string // Only the object name without prefixes.
lifeCycle *lifecycle.Lifecycle
replication replicationConfig
heal bool // Has the object been selected for heal check?
debug bool
}
type sizeSummary struct {
totalSize int64
versions uint64
replicatedSize int64
pendingSize int64
failedSize int64
replicaSize int64
pendingCount uint64
failedCount uint64
}
type getSizeFn func(item scannerItem) (sizeSummary, error)
// transformMetaDir will transform a directory to prefix/file.ext
func (i *scannerItem) transformMetaDir() {
split := strings.Split(i.prefix, SlashSeparator)
if len(split) > 1 {
i.prefix = path.Join(split[:len(split)-1]...)
} else {
i.prefix = ""
}
// Object name is last element
i.objectName = split[len(split)-1]
}
// actionMeta contains information used to apply actions.
type actionMeta struct {
oi ObjectInfo
bitRotScan bool // indicates if bitrot check was requested.
}
var applyActionsLogPrefix = color.Green("applyActions:")
func (i *scannerItem) applyHealing(ctx context.Context, o ObjectLayer, meta actionMeta) (size int64) {
if i.debug {
if meta.oi.VersionID != "" {
console.Debugf(applyActionsLogPrefix+" heal checking: %v/%v v(%s)\n", i.bucket, i.objectPath(), meta.oi.VersionID)
} else {
console.Debugf(applyActionsLogPrefix+" heal checking: %v/%v\n", i.bucket, i.objectPath())
}
}
healOpts := madmin.HealOpts{Remove: healDeleteDangling}
if meta.bitRotScan {
healOpts.ScanMode = madmin.HealDeepScan
}
res, err := o.HealObject(ctx, i.bucket, i.objectPath(), meta.oi.VersionID, healOpts)
if isErrObjectNotFound(err) || isErrVersionNotFound(err) {
return 0
}
if err != nil && !errors.Is(err, NotImplemented{}) {
logger.LogIf(ctx, err)
return 0
}
return res.ObjectSize
}
func (i *scannerItem) applyLifecycle(ctx context.Context, o ObjectLayer, meta actionMeta) (applied bool, size int64) {
size, err := meta.oi.GetActualSize()
if i.debug {
logger.LogIf(ctx, err)
}
if i.lifeCycle == nil {
if i.debug {
// disabled, very chatty:
// console.Debugf(applyActionsLogPrefix+" no lifecycle rules to apply: %q\n", i.objectPath())
}
return false, size
}
versionID := meta.oi.VersionID
action := i.lifeCycle.ComputeAction(
lifecycle.ObjectOpts{
Name: i.objectPath(),
UserTags: meta.oi.UserTags,
ModTime: meta.oi.ModTime,
VersionID: meta.oi.VersionID,
DeleteMarker: meta.oi.DeleteMarker,
IsLatest: meta.oi.IsLatest,
NumVersions: meta.oi.NumVersions,
SuccessorModTime: meta.oi.SuccessorModTime,
RestoreOngoing: meta.oi.RestoreOngoing,
RestoreExpires: meta.oi.RestoreExpires,
TransitionStatus: meta.oi.TransitionStatus,
RemoteTiersImmediately: globalDebugRemoteTiersImmediately,
})
if i.debug {
if versionID != "" {
console.Debugf(applyActionsLogPrefix+" lifecycle: %q (version-id=%s), Initial scan: %v\n", i.objectPath(), versionID, action)
} else {
console.Debugf(applyActionsLogPrefix+" lifecycle: %q Initial scan: %v\n", i.objectPath(), action)
}
}
switch action {
case lifecycle.DeleteAction, lifecycle.DeleteVersionAction:
case lifecycle.TransitionAction, lifecycle.TransitionVersionAction:
case lifecycle.DeleteRestoredAction, lifecycle.DeleteRestoredVersionAction:
default:
// No action.
if i.debug {
console.Debugf(applyActionsLogPrefix+" object not expirable: %q\n", i.objectPath())
}
return false, size
}
obj, err := o.GetObjectInfo(ctx, i.bucket, i.objectPath(), ObjectOptions{
VersionID: versionID,
})
if err != nil {
switch err.(type) {
case MethodNotAllowed: // This happens usually for a delete marker
if !obj.DeleteMarker { // if this is not a delete marker log and return
// Do nothing - heal in the future.
logger.LogIf(ctx, err)
return false, size
}
case ObjectNotFound, VersionNotFound:
// object not found or version not found return 0
return false, 0
default:
// All other errors proceed.
logger.LogIf(ctx, err)
return false, size
}
}
action = evalActionFromLifecycle(ctx, *i.lifeCycle, obj, i.debug)
if action != lifecycle.NoneAction {
applied = applyLifecycleAction(ctx, action, o, obj)
}
if applied {
switch action {
case lifecycle.TransitionAction, lifecycle.TransitionVersionAction:
return true, size
}
// For all other lifecycle actions that remove data
return true, 0
}
return false, size
}
// applyTierObjSweep removes remote object pending deletion and the free-version
// tracking this information.
func (i *scannerItem) applyTierObjSweep(ctx context.Context, o ObjectLayer, meta actionMeta) {
if !meta.oi.tierFreeVersion {
// nothing to be done
return
}
ignoreNotFoundErr := func(err error) error {
switch {
case isErrVersionNotFound(err), isErrObjectNotFound(err):
return nil
}
return err
}
// Remove the remote object
err := deleteObjectFromRemoteTier(ctx, meta.oi.transitionedObjName, meta.oi.transitionVersionID, meta.oi.TransitionTier)
if ignoreNotFoundErr(err) != nil {
logger.LogIf(ctx, err)
return
}
// Remove this free version
opts := ObjectOptions{}
opts.VersionID = meta.oi.VersionID
_, err = o.DeleteObject(ctx, meta.oi.Bucket, meta.oi.Name, opts)
if err == nil {
auditLogLifecycle(ctx, meta.oi, ILMFreeVersionDelete)
}
if ignoreNotFoundErr(err) != nil {
logger.LogIf(ctx, err)
}
}