Repository navigation
Expand file tree
/
Copy pathadmin-heal-ops.go
More file actions
977 lines (835 loc) · 26.6 KB
/
Copy pathadmin-heal-ops.go
File metadata and controls
977 lines (835 loc) · 26.6 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
// 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 (
"context"
"encoding/json"
"fmt"
"net/http"
"sort"
"sync"
"time"
"github.com/minio/madmin-go"
"github.com/minio/minio/cmd/logger"
)
// healStatusSummary - overall short summary of a healing sequence
type healStatusSummary string
// healStatusSummary constants
const (
healNotStartedStatus healStatusSummary = "not started"
healRunningStatus = "running"
healStoppedStatus = "stopped"
healFinishedStatus = "finished"
)
const (
// a heal sequence with this many un-consumed heal result
// items blocks until heal-status consumption resumes or is
// aborted due to timeout.
maxUnconsumedHealResultItems = 1000
// if no heal-results are consumed (via the heal-status API)
// for this timeout duration, the heal sequence is aborted.
healUnconsumedTimeout = 24 * time.Hour
// time-duration to keep heal sequence state after it
// completes.
keepHealSeqStateDuration = time.Minute * 10
// nopHeal is a no operating healing action to
// wait for the current healing operation to finish
nopHeal = ""
)
var (
errHealIdleTimeout = fmt.Errorf("healing results were not consumed for too long")
errHealStopSignalled = fmt.Errorf("heal stop signaled")
errFnHealFromAPIErr = func(ctx context.Context, err error) error {
apiErr := toAdminAPIErr(ctx, err)
return fmt.Errorf("Heal internal error: %s: %s",
apiErr.Code, apiErr.Description)
}
)
// healSequenceStatus - accumulated status of the heal sequence
type healSequenceStatus struct {
// summary and detail for failures
Summary healStatusSummary `json:"Summary"`
FailureDetail string `json:"Detail,omitempty"`
StartTime time.Time `json:"StartTime"`
// settings for the heal sequence
HealSettings madmin.HealOpts `json:"Settings"`
// slice of available heal result records
Items []madmin.HealResultItem `json:"Items"`
}
// structure to hold state of all heal sequences in server memory
type allHealState struct {
sync.RWMutex
// map of heal path to heal sequence
healSeqMap map[string]*healSequence // Indexed by endpoint
healLocalDisks map[Endpoint]struct{}
healStatus map[string]healingTracker // Indexed by disk ID
}
// newHealState - initialize global heal state management
func newHealState(cleanup bool) *allHealState {
hstate := &allHealState{
healSeqMap: make(map[string]*healSequence),
healLocalDisks: map[Endpoint]struct{}{},
healStatus: make(map[string]healingTracker),
}
if cleanup {
go hstate.periodicHealSeqsClean(GlobalContext)
}
return hstate
}
func (ahs *allHealState) healDriveCount() int {
ahs.RLock()
defer ahs.RUnlock()
return len(ahs.healLocalDisks)
}
func (ahs *allHealState) popHealLocalDisks(healLocalDisks ...Endpoint) {
ahs.Lock()
defer ahs.Unlock()
for _, ep := range healLocalDisks {
delete(ahs.healLocalDisks, ep)
}
for id, disk := range ahs.healStatus {
for _, ep := range healLocalDisks {
if disk.Endpoint == ep.String() {
delete(ahs.healStatus, id)
}
}
}
}
// updateHealStatus will update the heal status.
func (ahs *allHealState) updateHealStatus(tracker *healingTracker) {
ahs.Lock()
defer ahs.Unlock()
ahs.healStatus[tracker.ID] = *tracker
}
// Sort by zone, set and disk index
func sortDisks(disks []madmin.Disk) {
sort.Slice(disks, func(i, j int) bool {
a, b := &disks[i], &disks[j]
if a.PoolIndex != b.PoolIndex {
return a.PoolIndex < b.PoolIndex
}
if a.SetIndex != b.SetIndex {
return a.SetIndex < b.SetIndex
}
return a.DiskIndex < b.DiskIndex
})
}
// getLocalHealingDisks returns local healing disks indexed by endpoint.
func (ahs *allHealState) getLocalHealingDisks() map[string]madmin.HealingDisk {
ahs.RLock()
defer ahs.RUnlock()
dst := make(map[string]madmin.HealingDisk, len(ahs.healStatus))
for _, v := range ahs.healStatus {
dst[v.Endpoint] = v.toHealingDisk()
}
return dst
}
func (ahs *allHealState) getHealLocalDiskEndpoints() Endpoints {
ahs.RLock()
defer ahs.RUnlock()
var endpoints Endpoints
for ep := range ahs.healLocalDisks {
endpoints = append(endpoints, ep)
}
return endpoints
}
func (ahs *allHealState) pushHealLocalDisks(healLocalDisks ...Endpoint) {
ahs.Lock()
defer ahs.Unlock()
for _, ep := range healLocalDisks {
ahs.healLocalDisks[ep] = struct{}{}
}
}
func (ahs *allHealState) periodicHealSeqsClean(ctx context.Context) {
// Launch clean-up routine to remove this heal sequence (after
// it ends) from the global state after timeout has elapsed.
periodicTimer := time.NewTimer(time.Minute * 5)
defer periodicTimer.Stop()
for {
select {
case <-periodicTimer.C:
periodicTimer.Reset(time.Minute * 5)
now := UTCNow()
ahs.Lock()
for path, h := range ahs.healSeqMap {
if h.hasEnded() && h.endTime.Add(keepHealSeqStateDuration).Before(now) {
delete(ahs.healSeqMap, path)
}
}
ahs.Unlock()
case <-ctx.Done():
// server could be restarting - need
// to exit immediately
return
}
}
}
// getHealSequenceByToken - Retrieve a heal sequence by token. The second
// argument returns if a heal sequence actually exists.
func (ahs *allHealState) getHealSequenceByToken(token string) (h *healSequence, exists bool) {
ahs.Lock()
defer ahs.Unlock()
for _, healSeq := range ahs.healSeqMap {
if healSeq.clientToken == token {
return healSeq, true
}
}
return nil, false
}
// getHealSequence - Retrieve a heal sequence by path. The second
// argument returns if a heal sequence actually exists.
func (ahs *allHealState) getHealSequence(path string) (h *healSequence, exists bool) {
ahs.Lock()
defer ahs.Unlock()
h, exists = ahs.healSeqMap[path]
return h, exists
}
func (ahs *allHealState) stopHealSequence(path string) ([]byte, APIError) {
var hsp madmin.HealStopSuccess
he, exists := ahs.getHealSequence(path)
if !exists {
hsp = madmin.HealStopSuccess{
ClientToken: "unknown",
StartTime: UTCNow(),
}
} else {
clientToken := he.clientToken
if globalIsDistErasure {
clientToken = fmt.Sprintf("%s@%d", he.clientToken, GetProxyEndpointLocalIndex(globalProxyEndpoints))
}
hsp = madmin.HealStopSuccess{
ClientToken: clientToken,
ClientAddress: he.clientAddress,
StartTime: he.startTime,
}
he.stop()
for !he.hasEnded() {
time.Sleep(1 * time.Second)
}
ahs.Lock()
defer ahs.Unlock()
// Heal sequence explicitly stopped, remove it.
delete(ahs.healSeqMap, path)
}
b, err := json.Marshal(&hsp)
return b, toAdminAPIErr(GlobalContext, err)
}
// LaunchNewHealSequence - launches a background routine that performs
// healing according to the healSequence argument. For each heal
// sequence, state is stored in the `globalAllHealState`, which is a
// map of the heal path to `healSequence` which holds state about the
// heal sequence.
//
// Heal results are persisted in server memory for
// `keepHealSeqStateDuration`. This function also launches a
// background routine to clean up heal results after the
// aforementioned duration.
func (ahs *allHealState) LaunchNewHealSequence(h *healSequence, objAPI ObjectLayer) (
respBytes []byte, apiErr APIError, errMsg string) {
if h.forceStarted {
_, apiErr = ahs.stopHealSequence(pathJoin(h.bucket, h.object))
if apiErr.Code != "" {
return respBytes, apiErr, ""
}
} else {
oh, exists := ahs.getHealSequence(pathJoin(h.bucket, h.object))
if exists && !oh.hasEnded() {
errMsg = "Heal is already running on the given path " +
"(use force-start option to stop and start afresh). " +
fmt.Sprintf("The heal was started by IP %s at %s, token is %s",
h.clientAddress, h.startTime.Format(http.TimeFormat), h.clientToken)
return nil, errorCodes.ToAPIErr(ErrHealAlreadyRunning), errMsg
}
}
ahs.Lock()
defer ahs.Unlock()
// Check if new heal sequence to be started overlaps with any
// existing, running sequence
hpath := pathJoin(h.bucket, h.object)
for k, hSeq := range ahs.healSeqMap {
if !hSeq.hasEnded() && (HasPrefix(k, hpath) || HasPrefix(hpath, k)) {
errMsg = "The provided heal sequence path overlaps with an existing " +
fmt.Sprintf("heal path: %s", k)
return nil, errorCodes.ToAPIErr(ErrHealOverlappingPaths), errMsg
}
}
// Add heal state and start sequence
ahs.healSeqMap[hpath] = h
// Launch top-level background heal go-routine
go h.healSequenceStart(objAPI)
clientToken := h.clientToken
if globalIsDistErasure {
clientToken = fmt.Sprintf("%s@%d", h.clientToken, GetProxyEndpointLocalIndex(globalProxyEndpoints))
}
b, err := json.Marshal(madmin.HealStartSuccess{
ClientToken: clientToken,
ClientAddress: h.clientAddress,
StartTime: h.startTime,
})
if err != nil {
logger.LogIf(h.ctx, err)
return nil, toAdminAPIErr(h.ctx, err), ""
}
return b, noError, ""
}
// PopHealStatusJSON - Called by heal-status API. It fetches the heal
// status results from global state and returns its JSON
// representation. The clientToken helps ensure there aren't
// conflicting clients fetching status.
func (ahs *allHealState) PopHealStatusJSON(hpath string,
clientToken string) ([]byte, APIErrorCode) {
// fetch heal state for given path
h, exists := ahs.getHealSequence(hpath)
if !exists {
// heal sequence doesn't exist, must have finished.
jbytes, err := json.Marshal(healSequenceStatus{
Summary: healFinishedStatus,
})
return jbytes, toAdminAPIErrCode(GlobalContext, err)
}
// Check if client-token is valid
if clientToken != h.clientToken {
return nil, ErrHealInvalidClientToken
}
// Take lock to access and update the heal-sequence
h.mutex.Lock()
defer h.mutex.Unlock()
numItems := len(h.currentStatus.Items)
// calculate index of most recently available heal result
// record.
lastResultIndex := h.lastSentResultIndex
if numItems > 0 {
lastResultIndex = h.currentStatus.Items[numItems-1].ResultIndex
}
h.lastSentResultIndex = lastResultIndex
jbytes, err := json.Marshal(h.currentStatus)
if err != nil {
h.currentStatus.Items = nil
logger.LogIf(h.ctx, err)
return nil, ErrInternalError
}
h.currentStatus.Items = nil
return jbytes, ErrNone
}
// healSource denotes single entity and heal option.
type healSource struct {
bucket string
object string
versionID string
opts *madmin.HealOpts // optional heal option overrides default setting
}
// healSequence - state for each heal sequence initiated on the
// server.
type healSequence struct {
// bucket, and object on which heal seq. was initiated
bucket, object string
// A channel of entities (format, buckets, objects) to heal
sourceCh chan healSource
// A channel of entities with heal result
respCh chan healResult
// Report healing progress
reportProgress bool
// time at which heal sequence was started
startTime time.Time
// time at which heal sequence has ended
endTime time.Time
// Heal client info
clientToken, clientAddress string
// was this heal sequence force started?
forceStarted bool
// heal settings applied to this heal sequence
settings madmin.HealOpts
// current accumulated status of the heal sequence
currentStatus healSequenceStatus
// channel signaled by background routine when traversal has
// completed
traverseAndHealDoneCh chan error
// canceler to cancel heal sequence.
cancelCtx context.CancelFunc
// the last result index sent to client
lastSentResultIndex int64
// Number of total items scanned against item type
scannedItemsMap map[madmin.HealItemType]int64
// Number of total items healed against item type
healedItemsMap map[madmin.HealItemType]int64
// Number of total items where healing failed against endpoint and drive state
healFailedItemsMap map[string]int64
// The time of the last scan/heal activity
lastHealActivity time.Time
// Holds the request-info for logging
ctx context.Context
// used to lock this structure as it is concurrently accessed
mutex sync.RWMutex
}
// NewHealSequence - creates healSettings, assumes bucket and
// objPrefix are already validated.
func newHealSequence(ctx context.Context, bucket, objPrefix, clientAddr string,
hs madmin.HealOpts, forceStart bool) *healSequence {
reqInfo := &logger.ReqInfo{RemoteHost: clientAddr, API: "Heal", BucketName: bucket}
reqInfo.AppendTags("prefix", objPrefix)
ctx, cancel := context.WithCancel(logger.SetReqInfo(ctx, reqInfo))
clientToken := mustGetUUID()
return &healSequence{
respCh: make(chan healResult),
bucket: bucket,
object: objPrefix,
reportProgress: true,
startTime: UTCNow(),
clientToken: clientToken,
clientAddress: clientAddr,
forceStarted: forceStart,
settings: hs,
currentStatus: healSequenceStatus{
Summary: healNotStartedStatus,
HealSettings: hs,
},
traverseAndHealDoneCh: make(chan error),
cancelCtx: cancel,
ctx: ctx,
scannedItemsMap: make(map[madmin.HealItemType]int64),
healedItemsMap: make(map[madmin.HealItemType]int64),
healFailedItemsMap: make(map[string]int64),
}
}
// getScannedItemsCount - returns a count of all scanned items
func (h *healSequence) getScannedItemsCount() int64 {
var count int64
h.mutex.RLock()
defer h.mutex.RUnlock()
for _, v := range h.scannedItemsMap {
count = count + v
}
return count
}
// getScannedItemsMap - returns map of all scanned items against type
func (h *healSequence) getScannedItemsMap() map[madmin.HealItemType]int64 {
h.mutex.RLock()
defer h.mutex.RUnlock()
// Make a copy before returning the value
retMap := make(map[madmin.HealItemType]int64, len(h.scannedItemsMap))
for k, v := range h.scannedItemsMap {
retMap[k] = v
}
return retMap
}
// getHealedItemsMap - returns the map of all healed items against type
func (h *healSequence) getHealedItemsMap() map[madmin.HealItemType]int64 {
h.mutex.RLock()
defer h.mutex.RUnlock()
// Make a copy before returning the value
retMap := make(map[madmin.HealItemType]int64, len(h.healedItemsMap))
for k, v := range h.healedItemsMap {
retMap[k] = v
}
return retMap
}
// gethealFailedItemsMap - returns map of all items where heal failed against
// drive endpoint and status
func (h *healSequence) gethealFailedItemsMap() map[string]int64 {
h.mutex.RLock()
defer h.mutex.RUnlock()
// Make a copy before returning the value
retMap := make(map[string]int64, len(h.healFailedItemsMap))
for k, v := range h.healFailedItemsMap {
retMap[k] = v
}
return retMap
}
// isQuitting - determines if the heal sequence is quitting (due to an
// external signal)
func (h *healSequence) isQuitting() bool {
select {
case <-h.ctx.Done():
return true
default:
return false
}
}
// check if the heal sequence has ended
func (h *healSequence) hasEnded() bool {
h.mutex.RLock()
defer h.mutex.RUnlock()
// background heal never ends
if h.clientToken == bgHealingUUID {
return false
}
return !h.endTime.IsZero()
}
// stops the heal sequence - safe to call multiple times.
func (h *healSequence) stop() {
h.cancelCtx()
}
// pushHealResultItem - pushes a heal result item for consumption in
// the heal-status API. It blocks if there are
// maxUnconsumedHealResultItems. When it blocks, the heal sequence
// routine is effectively paused - this happens when the server has
// accumulated the maximum number of heal records per heal
// sequence. When the client consumes further records, the heal
// sequence automatically resumes. The return value indicates if the
// operation succeeded.
func (h *healSequence) pushHealResultItem(r madmin.HealResultItem) error {
// start a timer to keep an upper time limit to find an empty
// slot to add the given heal result - if no slot is found it
// means that the server is holding the maximum amount of
// heal-results in memory and the client has not consumed it
// for too long.
unconsumedTimer := time.NewTimer(healUnconsumedTimeout)
defer func() {
// stop the timeout timer so it is garbage collected.
if !unconsumedTimer.Stop() {
<-unconsumedTimer.C
}
}()
var itemsLen int
for {
h.mutex.Lock()
itemsLen = len(h.currentStatus.Items)
if itemsLen == maxUnconsumedHealResultItems {
// wait for a second, or quit if an external
// stop signal is received or the
// unconsumedTimer fires.
select {
// Check after a second
case <-time.After(time.Second):
h.mutex.Unlock()
continue
case <-h.ctx.Done():
h.mutex.Unlock()
// discard result and return.
return errHealStopSignalled
// Timeout if no results consumed for too long.
case <-unconsumedTimer.C:
h.mutex.Unlock()
return errHealIdleTimeout
}
}
break
}
// Set the correct result index for the new result item
if itemsLen > 0 {
r.ResultIndex = 1 + h.currentStatus.Items[itemsLen-1].ResultIndex
} else {
r.ResultIndex = 1 + h.lastSentResultIndex
}
// append to results
h.currentStatus.Items = append(h.currentStatus.Items, r)
// release lock
h.mutex.Unlock()
return nil
}
// healSequenceStart - this is the top-level background heal
// routine. It launches another go-routine that actually traverses
// on-disk data, checks and heals according to the selected
// settings. This go-routine itself, (1) monitors the traversal
// routine for completion, and (2) listens for external stop
// signals. When either event happens, it sets the finish status for
// the heal-sequence.
func (h *healSequence) healSequenceStart(objAPI ObjectLayer) {
// Set status as running
h.mutex.Lock()
h.currentStatus.Summary = healRunningStatus
h.currentStatus.StartTime = UTCNow()
h.mutex.Unlock()
if h.sourceCh == nil {
go h.traverseAndHeal(objAPI)
} else {
go h.healFromSourceCh()
}
select {
case err, ok := <-h.traverseAndHealDoneCh:
if !ok {
return
}
h.mutex.Lock()
h.endTime = UTCNow()
// Heal traversal is complete.
if err == nil {
// heal traversal succeeded.
h.currentStatus.Summary = healFinishedStatus
} else {
// heal traversal had an error.
h.currentStatus.Summary = healStoppedStatus
h.currentStatus.FailureDetail = err.Error()
}
h.mutex.Unlock()
case <-h.ctx.Done():
h.mutex.Lock()
h.endTime = UTCNow()
h.currentStatus.Summary = healFinishedStatus
h.mutex.Unlock()
// drain traverse channel so the traversal
// go-routine does not leak.
go func() {
// Eventually the traversal go-routine closes
// the channel and returns, so this go-routine
// itself will not leak.
<-h.traverseAndHealDoneCh
}()
}
}
func (h *healSequence) logHeal(healType madmin.HealItemType) {
h.mutex.Lock()
h.scannedItemsMap[healType]++
h.lastHealActivity = UTCNow()
h.mutex.Unlock()
}
func (h *healSequence) queueHealTask(source healSource, healType madmin.HealItemType) error {
globalHealConfigMu.Lock()
opts := globalHealConfig
globalHealConfigMu.Unlock()
// Send heal request
task := healTask{
bucket: source.bucket,
object: source.object,
versionID: source.versionID,
opts: h.settings,
responseCh: h.respCh,
}
if source.opts != nil {
task.opts = *source.opts
}
if opts.Bitrot {
task.opts.ScanMode = madmin.HealDeepScan
}
// Wait and proceed if there are active requests
waitForLowHTTPReq(opts.IOCount, opts.Sleep)
h.mutex.Lock()
h.scannedItemsMap[healType]++
h.lastHealActivity = UTCNow()
h.mutex.Unlock()
globalBackgroundHealRoutine.queueHealTask(task)
select {
case res := <-h.respCh:
if !h.reportProgress {
// Object might have been deleted, by the time heal
// was attempted, we should ignore this object and
// return the error and not calculate this object
// as part of the metrics.
if isErrObjectNotFound(res.err) || isErrVersionNotFound(res.err) {
// Return the error so that caller can handle it.
return res.err
}
h.mutex.Lock()
defer h.mutex.Unlock()
// Progress is not reported in case of background heal processing.
// Instead we increment relevant counter based on the heal result
// for prometheus reporting.
if res.err != nil {
for _, d := range res.result.After.Drives {
// For failed items we report the endpoint and drive state
// This will help users take corrective actions for drives
h.healFailedItemsMap[d.Endpoint+","+d.State]++
}
} else {
// Only object type reported for successful healing
h.healedItemsMap[res.result.Type]++
}
// Report caller of any failure
return res.err
}
res.result.Type = healType
if res.err != nil {
// Object might have been deleted, by the time heal
// was attempted, we should ignore this object and return success.
if isErrObjectNotFound(res.err) || isErrVersionNotFound(res.err) {
return nil
}
// Only report object error
if healType != madmin.HealItemObject {
return res.err
}
res.result.Detail = res.err.Error()
}
return h.pushHealResultItem(res.result)
case <-h.ctx.Done():
return nil
}
}
func (h *healSequence) healItemsFromSourceCh() error {
for {
select {
case source, ok := <-h.sourceCh:
if !ok {
return nil
}
var itemType madmin.HealItemType
switch source.bucket {
case nopHeal:
continue
case SlashSeparator:
itemType = madmin.HealItemMetadata
default:
if source.object == "" {
itemType = madmin.HealItemBucket
} else {
itemType = madmin.HealItemObject
}
}
if err := h.queueHealTask(source, itemType); err != nil {
switch err.(type) {
case ObjectExistsAsDirectory:
case ObjectNotFound:
case VersionNotFound:
default:
logger.LogIf(h.ctx, fmt.Errorf("Heal attempt failed for %s: %w",
pathJoin(source.bucket, source.object), err))
}
}
case <-h.ctx.Done():
return nil
}
}
}
func (h *healSequence) healFromSourceCh() {
h.healItemsFromSourceCh()
}
func (h *healSequence) healDiskMeta(objAPI ObjectLayer) error {
// Try to pro-actively heal backend-encrypted file.
if err := h.queueHealTask(healSource{
bucket: minioMetaBucket,
object: backendEncryptedFile,
}, madmin.HealItemBucketMetadata); err != nil {
if !isErrObjectNotFound(err) && !isErrVersionNotFound(err) {
return err
}
}
// Start healing the config prefix.
return h.healMinioSysMeta(objAPI, minioConfigPrefix)()
}
func (h *healSequence) healItems(objAPI ObjectLayer, bucketsOnly bool) error {
if err := h.healDiskMeta(objAPI); err != nil {
return err
}
// Heal buckets and objects
return h.healBuckets(objAPI, bucketsOnly)
}
// traverseAndHeal - traverses on-disk data and performs healing
// according to settings. At each "safe" point it also checks if an
// external quit signal has been received and quits if so. Since the
// healing traversal may be mutating on-disk data when an external
// quit signal is received, this routine cannot quit immediately and
// has to wait until a safe point is reached, such as between scanning
// two objects.
func (h *healSequence) traverseAndHeal(objAPI ObjectLayer) {
bucketsOnly := false // Heals buckets and objects also.
h.traverseAndHealDoneCh <- h.healItems(objAPI, bucketsOnly)
close(h.traverseAndHealDoneCh)
}
// healMinioSysMeta - heals all files under a given meta prefix, returns a function
// which in-turn heals the respective meta directory path and any files in int.
func (h *healSequence) healMinioSysMeta(objAPI ObjectLayer, metaPrefix string) func() error {
return func() error {
// NOTE: Healing on meta is run regardless
// of any bucket being selected, this is to ensure that
// meta are always upto date and correct.
return objAPI.HealObjects(h.ctx, minioMetaBucket, metaPrefix, h.settings, func(bucket, object, versionID string) error {
if h.isQuitting() {
return errHealStopSignalled
}
err := h.queueHealTask(healSource{
bucket: bucket,
object: object,
versionID: versionID,
}, madmin.HealItemBucketMetadata)
// Object might have been deleted, by the time heal
// was attempted we ignore this object an move on.
if isErrObjectNotFound(err) || isErrVersionNotFound(err) {
return nil
}
return err
})
}
}
// healDiskFormat - heals format.json, return value indicates if a
// failure error occurred.
func (h *healSequence) healDiskFormat() error {
if h.isQuitting() {
return errHealStopSignalled
}
return h.queueHealTask(healSource{bucket: SlashSeparator}, madmin.HealItemMetadata)
}
// healBuckets - check for all buckets heal or just particular bucket.
func (h *healSequence) healBuckets(objAPI ObjectLayer, bucketsOnly bool) error {
if h.isQuitting() {
return errHealStopSignalled
}
// 1. If a bucket was specified, heal only the bucket.
if h.bucket != "" {
return h.healBucket(objAPI, h.bucket, bucketsOnly)
}
buckets, err := objAPI.ListBuckets(h.ctx)
if err != nil {
return errFnHealFromAPIErr(h.ctx, err)
}
// Heal latest buckets first.
sort.Slice(buckets, func(i, j int) bool {
return buckets[i].Created.After(buckets[j].Created)
})
for _, bucket := range buckets {
if err = h.healBucket(objAPI, bucket.Name, bucketsOnly); err != nil {
return err
}
}
return nil
}
// healBucket - traverses and heals given bucket
func (h *healSequence) healBucket(objAPI ObjectLayer, bucket string, bucketsOnly bool) error {
if err := h.queueHealTask(healSource{bucket: bucket}, madmin.HealItemBucket); err != nil {
if !isErrObjectNotFound(err) && !isErrVersionNotFound(err) {
return err
}
}
if bucketsOnly {
return nil
}
if !h.settings.Recursive {
if h.object != "" {
// Check if an object named as the objPrefix exists,
// and if so heal it.
oi, err := objAPI.GetObjectInfo(h.ctx, bucket, h.object, ObjectOptions{})
if err == nil {
if err = h.healObject(bucket, h.object, oi.VersionID); err != nil {
if isErrObjectNotFound(err) || isErrVersionNotFound(err) {
return nil
}
return err
}
}
}
return nil
}
if err := objAPI.HealObjects(h.ctx, bucket, h.object, h.settings, h.healObject); err != nil {
// Object might have been deleted, by the time heal
// was attempted we ignore this object an move on.
if !isErrObjectNotFound(err) && !isErrVersionNotFound(err) {
return errFnHealFromAPIErr(h.ctx, err)
}
}
return nil
}
// healObject - heal the given object and record result
func (h *healSequence) healObject(bucket, object, versionID string) error {
if h.isQuitting() {
return errHealStopSi
3643
gnalled
}
err := h.queueHealTask(healSource{
bucket: bucket,
object: object,
versionID: versionID,
}, madmin.HealItemObject)
return err
}