Repository navigation
Expand file tree
/
Copy pathdisk-cache.go
More file actions
1135 lines (1032 loc) · 38.3 KB
/
Copy pathdisk-cache.go
File metadata and controls
1135 lines (1032 loc) · 38.3 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
/*
* MinIO Cloud St
10BC8
orage, (C) 2018 MinIO, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package cmd
import (
"context"
"errors"
"fmt"
"io"
"io/ioutil"
"net/http"
"os"
"sort"
"strconv"
"strings"
"time"
"github.com/djherbis/atime"
"github.com/minio/minio/cmd/crypto"
"github.com/minio/minio/cmd/logger"
"github.com/minio/minio/pkg/hash"
"github.com/minio/minio/pkg/wildcard"
)
const (
// disk cache needs to have object size space free for a cache entry to be created.
cacheTrashDir = "trash"
cacheCleanupInterval = 10 // in minutes
)
// abstract slice of cache drives backed by FS.
type diskCache struct {
cfs []*cacheFSObjects
}
// Abstracts disk caching - used by the S3 layer
type cacheObjects struct {
// pointer to disk cache
cache *diskCache
// ListObjects pool management.
listPool *TreeWalkPool
// file path patterns to exclude from cache
exclude []string
// Object functions pointing to the corresponding functions of backend implementation.
GetObjectNInfoFn func(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, lockType LockType, opts ObjectOptions) (gr *GetObjectReader, err error)
GetObjectFn func(ctx context.Context, bucket, object string, startOffset int64, length int64, writer io.Writer, etag string, opts ObjectOptions) (err error)
GetObjectInfoFn func(ctx context.Context, bucket, object string, opts ObjectOptions) (objInfo ObjectInfo, err error)
PutObjectFn func(ctx context.Context, bucket, object string, data *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error)
DeleteObjectFn func(ctx context.Context, bucket, object string) error
DeleteObjectsFn func(ctx context.Context, bucket string, objects []string) ([]error, error)
ListObjectsFn func(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (result ListObjectsInfo, err error)
ListObjectsV2Fn func(ctx context.Context, bucket, prefix, continuationToken, delimiter string, maxKeys int, fetchOwner bool, startAfter string) (result ListObjectsV2Info, err error)
ListBucketsFn func(ctx context.Context) (buckets []BucketInfo, err error)
GetBucketInfoFn func(ctx context.Context, bucket string) (bucketInfo BucketInfo, err error)
NewMultipartUploadFn func(ctx context.Context, bucket, object string, opts ObjectOptions) (uploadID string, err error)
PutObjectPartFn func(ctx context.Context, bucket, object, uploadID string, partID int, data *PutObjReader, opts ObjectOptions) (info PartInfo, err error)
AbortMultipartUploadFn func(ctx context.Context, bucket, object, uploadID string) error
CompleteMultipartUploadFn func(ctx context.Context, bucket, object, uploadID string, uploadedParts []CompletePart, opts ObjectOptions) (objInfo ObjectInfo, err error)
DeleteBucketFn func(ctx context.Context, bucket string) error
}
// CacheStorageInfo - represents total, free capacity of
// underlying cache storage.
type CacheStorageInfo struct {
Total uint64 // Total cache disk space.
Free uint64 // Free cache available space.
}
// CacheObjectLayer implements primitives for cache object API layer.
type CacheObjectLayer interface {
// Bucket operations.
ListObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (result ListObjectsInfo, err error)
ListObjectsV2(ctx context.Context, bucket, prefix, continuationToken, delimiter string, maxKeys int, fetchOwner bool, startAfter string) (result ListObjectsV2Info, err error)
GetBucketInfo(ctx context.Context, bucket string) (bucketInfo BucketInfo, err error)
ListBuckets(ctx context.Context) (buckets []BucketInfo, err error)
DeleteBucket(ctx context.Context, bucket string) error
// Object operations.
GetObjectNInfo(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, lockType LockType, opts ObjectOptions) (gr *GetObjectReader, err error)
GetObject(ctx context.Context, bucket, object string, startOffset int64, length int64, writer io.Writer, etag string, opts ObjectOptions) (err error)
GetObjectInfo(ctx context.Context, bucket, object string, opts ObjectOptions) (objInfo ObjectInfo, err error)
PutObject(ctx context.Context, bucket, object string, data *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error)
DeleteObject(ctx context.Context, bucket, object string) error
DeleteObjects(ctx context.Context, bucket string, objects []string) ([]error, error)
// Multipart operations.
NewMultipartUpload(ctx context.Context, bucket, object string, opts ObjectOptions) (uploadID string, err error)
PutObjectPart(ctx context.Context, bucket, object, uploadID string, partID int, data *PutObjReader, opts ObjectOptions) (info PartInfo, err error)
AbortMultipartUpload(ctx context.Context, bucket, object, uploadID string) error
CompleteMultipartUpload(ctx context.Context, bucket, object, uploadID string, uploadedParts []CompletePart, opts ObjectOptions) (objInfo ObjectInfo, err error)
// Storage operations.
StorageInfo(ctx context.Context) CacheStorageInfo
}
// IsCacheable returns if the object should be saved in the cache.
func (o ObjectInfo) IsCacheable() bool {
return !crypto.IsEncrypted(o.UserDefined)
}
// backendDownError returns true if err is due to backend failure or faulty disk if in server mode
func backendDownError(err error) bool {
_, backendDown := err.(BackendDown)
return backendDown || IsErr(err, baseErrs...)
}
// get cache disk where object is currently cached for a GET operation. If object does not exist at that location,
// treat the list of cache drives as a circular buffer and walk through them starting at hash index
// until an online drive is found.If object is not found, fall back to the first online cache drive
// closest to the hash index, so that object can be recached.
func (c diskCache) getCachedFSLoc(ctx context.Context, bucket, object string) (*cacheFSObjects, error) {
index := c.hashIndex(bucket, object)
numDisks := len(c.cfs)
// save first online cache disk closest to the hint index
var firstOnlineDisk *cacheFSObjects
for k := 0; k < numDisks; k++ {
i := (index + k) % numDisks
if c.cfs[i] == nil {
continue
}
if c.cfs[i].IsOnline() {
if firstOnlineDisk == nil {
firstOnlineDisk = c.cfs[i]
}
if c.cfs[i].Exists(ctx, bucket, object) {
return c.cfs[i], nil
}
}
}
if firstOnlineDisk != nil {
return firstOnlineDisk, nil
}
return nil, errDiskNotFound
}
// choose a cache deterministically based on hash of bucket,object. The hash index is treated as
// a hint. In the event that the cache drive at hash index is offline, treat the list of cache drives
// as a circular buffer and walk through them starting at hash index until an online drive is found.
func (c diskCache) getCacheFS(ctx context.Context, bucket, object string) (*cacheFSObjects, error) {
index := c.hashIndex(bucket, object)
numDisks := len(c.cfs)
for k := 0; k < numDisks; k++ {
i := (index + k) % numDisks
if c.cfs[i] == nil {
continue
}
if c.cfs[i].IsOnline() {
return c.cfs[i], nil
}
}
return nil, errDiskNotFound
}
// Compute a unique hash sum for bucket and object
func (c diskCache) hashIndex(bucket, object string) int {
return crcHashMod(pathJoin(bucket, object), len(c.cfs))
}
// construct a metadata k-v map
func (c cacheObjects) getMetadata(objInfo ObjectInfo) map[string]string {
metadata := make(map[string]string)
metadata["etag"] = objInfo.ETag
metadata["content-type"] = objInfo.ContentType
metadata["content-encoding"] = objInfo.ContentEncoding
for key, val := range objInfo.UserDefined {
metadata[key] = val
}
return metadata
}
func (c cacheObjects) GetObjectNInfo(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, lockType LockType, opts ObjectOptions) (gr *GetObjectReader, err error) {
if c.isCacheExclude(bucket, object) {
return c.GetObjectNInfoFn(ctx, bucket, object, rs, h, lockType, opts)
}
// fetch cacheFSObjects if object is currently cached or nearest available cache drive
dcache, err := c.cache.getCachedFSLoc(ctx, bucket, object)
if err != nil {
return c.GetObjectNInfoFn(ctx, bucket, object, rs, h, lockType, opts)
}
cacheReader, cacheErr := dcache.GetObjectNInfo(ctx, bucket, object, rs, h, lockType, opts)
objInfo, err := c.GetObjectInfoFn(ctx, bucket, object, opts)
if backendDownError(err) && cacheErr == nil {
return cacheReader, nil
} else if err != nil {
if _, ok := err.(ObjectNotFound); ok {
if cacheErr == nil {
cacheReader.Close()
// Delete cached entry if backend object
// was deleted.
dcache.Delete(ctx, bucket, object)
}
}
return nil, err
}
if !objInfo.IsCacheable() || filterFromCache(objInfo.UserDefined) {
return c.GetObjectNInfoFn(ctx, bucket, object, rs, h, lockType, opts)
}
if cacheErr == nil {
if cacheReader.ObjInfo.ETag == objInfo.ETag && !isStaleCache(objInfo) {
// Object is not stale, so serve from cache
return cacheReader, nil
}
cacheReader.Close()
// Object is stale, so delete from cache
dcache.Delete(ctx, bucket, object)
}
// Since we got here, we are serving the request from backend,
// and also adding the object to the cache.
if rs != nil {
// We don't cache partial objects.
return c.GetObjectNInfoFn(ctx, bucket, object, rs, h, lockType, opts)
}
if !dcache.diskAvailable(objInfo.Size) {
return c.GetObjectNInfoFn(ctx, bucket, object, rs, h, lockType, opts)
}
bkReader, bkErr := c.GetObjectNInfoFn(ctx, bucket, object, rs, h, lockType, opts)
if bkErr != nil {
return nil, bkErr
}
// Initialize pipe.
pipeReader, pipeWriter := io.Pipe()
teeReader := io.TeeReader(bkReader, pipeWriter)
hashReader, herr := hash.NewReader(pipeReader, bkReader.ObjInfo.Size, "", "", bkReader.ObjInfo.Size, globalCLIContext.StrictS3Compat)
if herr != nil {
bkReader.Close()
return nil, herr
}
go func() {
putErr := dcache.Put(ctx, bucket, object, NewPutObjReader(hashReader, nil, nil), ObjectOptions{UserDefined: c.getMetadata(bkReader.ObjInfo)})
// close the write end of the pipe, so the error gets
// propagated to getObjReader
pipeWriter.CloseWithError(putErr)
}()
cleanupBackend := func() { bkReader.Close() }
cleanupPipe := func() { pipeReader.Close() }
return NewGetObjectReaderFromReader(teeReader, bkReader.ObjInfo, opts.CheckCopyPrecondFn, cleanupBackend, cleanupPipe)
}
// Uses cached-object to serve the request. If object is not cached it serves the request from the backend and also
// stores it in the cache for serving subsequent requests.
func (c cacheObjects) GetObject(ctx context.Context, bucket, object string, startOffset int64, length int64, writer io.Writer, etag string, opts ObjectOptions) (err error) {
GetObjectFn := c.GetObjectFn
GetObjectInfoFn := c.GetObjectInfoFn
if c.isCacheExclude(bucket, object) {
return GetObjectFn(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
// fetch cacheFSObjects if object is currently cached or nearest available cache drive
dcache, err := c.cache.getCachedFSLoc(ctx, bucket, object)
if err != nil {
return GetObjectFn(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
// stat object on backend
objInfo, err := GetObjectInfoFn(ctx, bucket, object, opts)
backendDown := backendDownError(err)
if err != nil && !backendDown {
if _, ok := err.(ObjectNotFound); ok {
// Delete the cached entry if backend object was deleted.
dcache.Delete(ctx, bucket, object)
}
return err
}
if !backendDown && !objInfo.IsCacheable() {
return GetObjectFn(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
if !backendDown && filterFromCache(objInfo.UserDefined) {
return GetObjectFn(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
cachedObjInfo, err := dcache.GetObjectInfo(ctx, bucket, object, opts)
if err == nil {
if backendDown {
// If the backend is down, serve the request from cache.
return dcache.Get(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
if cachedObjInfo.ETag == objInfo.ETag && !isStaleCache(objInfo) {
return dcache.Get(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
dcache.Delete(ctx, bucket, object)
}
if startOffset != 0 || (length > 0 && length != objInfo.Size) {
// We don't cache partial objects.
return GetObjectFn(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
if !dcache.diskAvailable(objInfo.Size) {
return GetObjectFn(ctx, bucket, object, startOffset, length, writer, etag, opts)
}
// Initialize pipe.
pipeReader, pipeWriter := io.Pipe()
hashReader, err := hash.NewReader(pipeReader, objInfo.Size, "", "", objInfo.Size, globalCLIContext.StrictS3Compat)
if err != nil {
return err
}
go func() {
gerr := GetObjectFn(ctx, bucket, object, 0, objInfo.Size, io.MultiWriter(writer, pipeWriter), etag, opts)
pipeWriter.CloseWithError(gerr) // Close writer explicitly signaling we wrote all data.
}()
opts.UserDefined = c.getMetadata(objInfo)
err = dcache.Put(ctx, bucket, object, NewPutObjReader(hashReader, nil, nil), opts)
if err != nil {
return err
}
pipeReader.Close()
return
}
// Returns ObjectInfo from cache if available.
func (c cacheObjects) GetObjectInfo(ctx context.Context, bucket, object string, opts ObjectOptions) (ObjectInfo, error) {
getObjectInfoFn := c.GetObjectInfoFn
if c.isCacheExclude(bucket, object) {
return getObjectInfoFn(ctx, bucket, object, opts)
}
// fetch cacheFSObjects if object is currently cached or nearest available cache drive
dcache, err := c.cache.getCachedFSLoc(ctx, bucket, object)
if err != nil {
return getObjectInfoFn(ctx, bucket, object, opts)
}
objInfo, err := getObjectInfoFn(ctx, bucket, object, opts)
if err != nil {
if _, ok := err.(ObjectNotFound); ok {
// Delete the cached entry if backend object was deleted.
dcache.Delete(ctx, bucket, object)
return ObjectInfo{}, err
}
if !backendDownError(err) {
return ObjectInfo{}, err
}
// when backend is down, serve from cache.
cachedObjInfo, cerr := dcache.GetObjectInfo(ctx, bucket, object, opts)
if cerr == nil {
return cachedObjInfo, nil
}
return ObjectInfo{}, BackendDown{}
}
// when backend is up, do a sanity check on cached object
cachedObjInfo, err := dcache.GetObjectInfo(ctx, bucket, object, opts)
if err != nil {
return objInfo, nil
}
if cachedObjInfo.ETag != objInfo.ETag {
// Delete the cached entry if the backend object was replaced.
dcache.Delete(ctx, bucket, object)
}
return objInfo, nil
}
// Returns function "listDir" of the type listDirFunc.
// isLeaf - is used by listDir function to check if an entry is a leaf or non-leaf entry.
// disks - list of fsObjects
func listDirCacheFactory(isLeaf func(string, string) bool, disks []*cacheFSObjects) ListDirFunc {
listCacheDirs := func(bucket, prefixDir, prefixEntry string) (dirs []string) {
var entries []string
for _, disk := range disks {
// ignore disk-caches that might be missing/offline
if disk == nil {
continue
}
fs := disk.FSObjects
var err error
entries, err = readDir(pathJoin(fs.fsPath, bucket, prefixDir))
if err != nil {
continue
}
for i := range entries {
if isLeaf(bucket, entries[i]) {
entries[i] = strings.TrimSuffix(entries[i], slashSeparator)
}
}
// Filter entries that have the prefix prefixEntry.
entries = filterMatchingPrefix(entries, prefixEntry)
dirs = append(dirs, entries...)
}
return dirs
}
// listDir - lists all the entries at a given prefix and given entry in the prefix.
listDir := func(bucket, prefixDir, prefixEntry string) (mergedEntries []string) {
cacheEntries := listCacheDirs(bucket, prefixDir, prefixEntry)
for _, entry := range cacheEntries {
// Find elements in entries which are not in mergedEntries
idx := sort.SearchStrings(mergedEntries, entry)
// if entry is already present in mergedEntries don't add.
if idx < len(mergedEntries) && mergedEntries[idx] == entry {
continue
}
mergedEntries = append(mergedEntries, entry)
sort.Strings(mergedEntries)
}
return mergedEntries
}
return listDir
}
// List all objects at prefix upto maxKeys, optionally delimited by '/' from the cache. Maintains the list pool
// state for future re-entrant list requests.
func (c cacheObjects) listCacheObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (result ListObjectsInfo, err error) {
var objInfos []ObjectInfo
var eof bool
var nextMarker string
recursive := true
if delimiter == slashSeparator {
recursive = false
}
walkResultCh, endWalkCh := c.listPool.Release(listParams{bucket, recursive, marker, prefix, false})
if walkResultCh == nil {
endWalkCh = make(chan struct{})
listDir := listDirCacheFactory(func(bucket, object string) bool {
fs, err := c.cache.getCacheFS(ctx, bucket, object)
if err != nil {
return false
}
_, err = fs.getObjectInfo(ctx, bucket, object)
return err == nil
}, c.cache.cfs)
walkResultCh = startTreeWalk(ctx, bucket, prefix, marker, recursive, listDir, endWalkCh)
}
for i := 0; i < maxKeys; {
walkResult, ok := <-walkResultCh
if !ok {
// Closed channel.
eof = true
break
}
entry := walkResult.entry
var objInfo ObjectInfo
if hasSuffix(entry, slashSeparator) {
// Object name needs to be full path.
objInfo.Bucket = bucket
objInfo.Name = entry
objInfo.IsDir = true
} else {
// Set the Mode to a "regular" file.
var err error
fs, err := c.cache.getCacheFS(ctx, bucket, entry)
if err != nil {
// Ignore errDiskNotFound
if err == errDiskNotFound {
continue
}
return result, toObjectErr(err, bucket, prefix)
}
objInfo, err = fs.getObjectInfo(ctx, bucket, entry)
if err != nil {
// Ignore ObjectNotFound error
if _, ok := err.(ObjectNotFound); ok {
continue
}
return result, toObjectErr(err, bucket, prefix)
}
}
nextMarker = objInfo.Name
objInfos = append(objInfos, objInfo)
i++
if walkResult.end {
eof = true
break
}
}
params := listParams{bucket, recursive, nextMarker, prefix, false}
if !eof {
c.listPool.Set(params, walkResultCh, endWalkCh)
}
result = ListObjectsInfo{IsTruncated: !eof}
for _, objInfo := range objInfos {
result.NextMarker = objInfo.Name
if objInfo.IsDir && delimiter == slashSeparator {
result.Prefixes = append(result.Prefixes, objInfo.Name)
continue
}
result.Objects = append(result.Objects, objInfo)
}
return result, nil
}
// listCacheV2Objects lists all blobs in bucket filtered by prefix from the cache
func (c cacheObjects) listCacheV2Objects(ctx context.Context, bucket, prefix, continuationToken, delimiter string, maxKeys int, fetchOwner bool, startAfter string) (result ListObjectsV2Info, err error) {
loi, err := c.listCacheObjects(ctx, bucket, prefix, continuationToken, delimiter, maxKeys)
if err != nil {
return result, err
}
listObjectsV2Info := ListObjectsV2Info{
IsTruncated: loi.IsTruncated,
ContinuationToken: continuationToken,
NextContinuationToken: loi.NextMarker,
Objects: loi.Objects,
Prefixes: loi.Prefixes,
}
return listObjectsV2Info, err
}
// List all objects at prefix upto maxKeys., optionally delimited by '/'. Maintains the list pool
// state for future re-entrant list requests. Retrieve from cache if backend is down
func (c cacheObjects) ListObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (result ListObjectsInfo, err error) {
listObjectsFn := c.ListObjectsFn
result, err = listObjectsFn(ctx, bucket, prefix, marker, delimiter, maxKeys)
if err != nil {
if backendDownError(err) {
return c.listCacheObjects(ctx, bucket, prefix, marker, delimiter, maxKeys)
}
return
}
return
}
// ListObjectsV2 lists all blobs in bucket filtered by prefix
func (c cacheObjects) ListObjectsV2(ctx context.Context, bucket, prefix, continuationToken, delimiter string, maxKeys int, fetchOwner bool, startAfter string) (result ListObjectsV2Info, err error) {
listObjectsV2Fn := c.ListObjectsV2Fn
result, err = listObjectsV2Fn(ctx, bucket, prefix, continuationToken, delimiter, maxKeys, fetchOwner, startAfter)
if err != nil {
if backendDownError(err) {
return c.listCacheV2Objects(ctx, bucket, prefix, continuationToken, delimiter, maxKeys, fetchOwner, startAfter)
}
return
}
return
}
// Lists all the buckets in the cache
func (c cacheObjects) listBuckets(ctx context.Context) (buckets []BucketInfo, err error) {
m := make(map[string]string)
for _, cache := range c.cache.cfs {
// ignore disk-caches that might be missing/offline
if cache == nil {
continue
}
entries, err := cache.ListBuckets(ctx)
if err != nil {
return nil, err
}
for _, entry := range entries {
_, ok := m[entry.Name]
if !ok {
m[entry.Name] = entry.Name
buckets = append(buckets, entry)
}
}
}
// Sort bucket infos by bucket name.
sort.Sort(byBucketName(buckets))
return
}
// Returns list of buckets from cache or the backend. If the backend is down, buckets
// available on cache are served.
func (c cacheObjects) ListBuckets(ctx context.Context) (buckets []BucketInfo, err error) {
listBucketsFn := c.ListBucketsFn
buckets, err = listBucketsFn(ctx)
if err != nil {
if backendDownError(err) {
return c.listBuckets(ctx)
}
return []BucketInfo{}, err
}
return
}
// Returns bucket info from cache if backend is down.
func (c cacheObjects) GetBucketInfo(ctx context.Context, bucket string) (bucketInfo BucketInfo, err error) {
getBucketInfoFn := c.GetBucketInfoFn
bucketInfo, err = getBucketInfoFn(ctx, bucket)
if backendDownError(err) {
for _, cache := range c.cache.cfs {
// ignore disk-caches that might be missing/offline
if cache == nil {
continue
}
if bucketInfo, err = cache.GetBucketInfo(ctx, bucket); err == nil {
return
}
}
}
return
}
// Delete Object deletes from cache as well if backend operation succeeds
func (c cacheObjects) DeleteObject(ctx context.Context, bucket, object string) (err error) {
if err = c.DeleteObjectFn(ctx, bucket, object); err != nil {
return
}
if c.isCacheExclude(bucket, object) {
return
}
dcache, cerr := c.cache.getCachedFSLoc(ctx, bucket, object)
if cerr == nil {
_ = dcache.DeleteObject(ctx, bucket, object)
}
return
}
func (c cacheObjects) DeleteObjects(ctx context.Context, bucket string, objects []string) ([]error, error) {
errs := make([]error, len(objects))
for idx, object := range objects {
errs[idx] = c.DeleteObject(ctx, bucket, object)
}
return errs, nil
}
// Returns true if object should be excluded from cache
func (c cacheObjects) isCacheExclude(bucket, object string) bool {
for _, pattern := range c.exclude {
matchStr := fmt.Sprintf("%s/%s", bucket, object)
if ok := wildcard.MatchSimple(pattern, matchStr); ok {
return true
}
}
return false
}
// PutObject - caches the uploaded object for single Put operations
func (c cacheObjects) PutObject(ctx context.Context, bucket, object string, r *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error) {
putObjectFn := c.PutObjectFn
data := r.Reader
dcache, err := c.cache.getCacheFS(ctx, bucket, object)
if err != nil {
// disk cache could not be located,execute backend call.
return putObjectFn(ctx, bucket, object, r, opts)
}
size := r.Size()
// fetch from backend if there is no space on cache drive
if !dcache.diskAvailable(size) {
return putObjectFn(ctx, bucket, object, r, opts)
}
// fetch from backend if cache exclude pattern or cache-control
// directive set to exclude
if c.isCacheExclude(bucket, object) || filterFromCache(opts.UserDefined) {
dcache.Delete(ctx, bucket, object)
return putObjectFn(ctx, bucket, object, r, opts)
}
objInfo = ObjectInfo{}
// Initialize pipe to stream data to backend
pipeReader, pipeWriter := io.Pipe()
hashReader, err := hash.NewReader(pipeReader, size, data.MD5HexString(), data.SHA256HexString(), data.ActualSize(), globalCLIContext.StrictS3Compat)
if err != nil {
return ObjectInfo{}, err
}
// Initialize pipe to stream data to cache
rPipe, wPipe := io.Pipe()
cHashReader, err := hash.NewReader(rPipe, size, data.MD5HexString(), data.SHA256HexString(), data.ActualSize(), globalCLIContext.StrictS3Compat)
if err != nil {
return ObjectInfo{}, err
}
oinfoCh := make(chan ObjectInfo)
errCh := make(chan error)
go func() {
oinfo, perr := putObjectFn(ctx, bucket, object, NewPutObjReader(hashReader, nil, nil), opts)
if perr != nil {
pipeWriter.CloseWithError(perr)
wPipe.CloseWithError(perr)
close(oinfoCh)
errCh <- perr
return
}
close(errCh)
oinfoCh <- oinfo
}()
go func() {
if err = dcache.Put(ctx, bucket, object, NewPutObjReader(cHashReader, nil, nil), opts); err != nil {
wPipe.CloseWithError(err)
return
}
}()
mwriter := io.MultiWriter(pipeWriter, wPipe)
_, err = io.Copy(mwriter, data)
if err != nil {
err = <-errCh
return objInfo, err
}
pipeWriter.Close()
wPipe.Close()
objInfo = <-oinfoCh
return objInfo, err
}
// NewMultipartUpload - Starts a new multipart upload operation to backend and cache.
func (c cacheObjects) NewMultipartUpload(ctx context.Context, bucket, object string, opts ObjectOptions) (uploadID string, err error) {
newMultipartUploadFn := c.NewMultipartUploadFn
if c.isCacheExclude(bucket, object) || filterFromCache(opts.UserDefined) {
return newMultipartUploadFn(ctx, bucket, object, opts)
}
dcache, err := c.cache.getCacheFS(ctx, bucket, object)
if err != nil {
// disk cache could not be located,execute backend call.
return newMultipartUploadFn(ctx, bucket, object, opts)
}
uploadID, err = newMultipartUploadFn(ctx, bucket, object, opts)
if err != nil {
return
}
// create new multipart upload in cache with same uploadID
dcache.NewMultipartUpload(ctx, bucket, object, uploadID, opts)
return uploadID, err
}
// PutObjectPart - uploads part to backend and cache simultaneously.
func (c cacheObjects) PutObjectPart(ctx context.Context, bucket, object, uploadID string, partID int, r *PutObjReader, opts ObjectOptions) (info PartInfo, err error) {
data := r.Reader
putObjectPartFn := c.PutObjectPartFn
dcache, err := c.cache.getCacheFS(ctx, bucket, object)
if err != nil {
// disk cache could not be located,execute backend call.
return putObjectPartFn(ctx, bucket, object, uploadID, partID, r, opts)
}
if c.isCacheExclude(bucket, object) {
return putObjectPartFn(ctx, bucket, object, uploadID, partID, r, opts)
}
// make sure cache has at least size space available
size := data.Size()
if !dcache.diskAvailable(size) {
select {
case dcache.purgeChan <- struct{}{}:
default:
}
return putObjectPartFn(ctx, bucket, object, uploadID, partID, r, opts)
}
info = PartInfo{}
// Initialize pipe to stream data to backend
pipeReader, pipeWriter := io.Pipe()
hashReader, err := hash.NewReader(pipeReader, size, data.MD5HexString(), data.SHA256HexString(), data.ActualSize(), globalCLIContext.StrictS3Compat)
if err != nil {
return
}
// Initialize pipe to stream data to cache
rPipe, wPipe := io.Pipe()
cHashReader, err := hash.NewReader(rPipe, size, data.MD5HexString(), data.SHA256HexString(), data.ActualSize(), globalCLIContext.StrictS3Compat)
if err != nil {
return
}
pinfoCh := make(chan PartInfo)
errorCh := make(chan error)
go func() {
info, err = putObjectPartFn(ctx, bucket, object, uploadID, partID, NewPutObjReader(hashReader, nil, nil), opts)
if err != nil {
close(pinfoCh)
pipeWriter.CloseWithError(err)
wPipe.CloseWithError(err)
errorCh <- err
return
}
close(errorCh)
pinfoCh <- info
}()
go func() {
if _, perr := dcache.PutObjectPart(ctx, bucket, object, uploadID, partID, NewPutObjReader(cHashReader, nil, nil), opts); perr != nil {
wPipe.CloseWithError(perr)
return
}
}()
mwriter := io.MultiWriter(pipeWriter, wPipe)
_, err = io.Copy(mwriter, data)
if err != nil {
err = <-errorCh
return PartInfo{}, err
}
pipeWriter.Close()
wPipe.Close()
info = <-pinfoCh
return info, err
}
// AbortMultipartUpload - aborts multipart upload on backend and cache.
func (c cacheObjects) AbortMultipartUpload(ctx context.Context, bucket, object, uploadID string) error {
abortMultipartUploadFn := c.AbortMultipartUploadFn
if c.isCacheExclude(bucket, object) {
return abortMultipartUploadFn(ctx, bucket, object, uploadID)
}
dcache, err := c.cache.getCacheFS(ctx, bucket, object)
if err != nil {
// disk cache could not be located,execute backend call.
return abortMultipartUploadFn(ctx, bucket, object, uploadID)
}
// execute backend operation
err = abortMultipartUploadFn(ctx, bucket, object, uploadID)
if err != nil {
return err
}
// abort multipart upload on cache
dcache.AbortMultipartUpload(ctx, bucket, object, uploadID)
return nil
}
// CompleteMultipartUpload - completes multipart upload operation on backend and cache.
func (c cacheObjects) CompleteMultipartUpload(ctx context.Context, bucket, object, uploadID string, uploadedParts []CompletePart, opts ObjectOptions) (objInfo ObjectInfo, err error) {
completeMultipartUploadFn := c.CompleteMultipartUploadFn
if c.isCacheExclude(bucket, object) {
return completeMultipartUploadFn(ctx, bucket, object, uploadID, uploadedParts, opts)
}
dcache, err := c.cache.getCacheFS(ctx, bucket, object)
if err != nil {
// disk cache could not be located,execute backend call.
return completeMultipartUploadFn(ctx, bucket, object, uploadID, uploadedParts, opts)
}
// perform backend operation
objInfo, err = completeMultipartUploadFn(ctx, bucket, object, uploadID, uploadedParts, opts)
if err != nil {
return
}
// create new multipart upload in cache with same uploadID
dcache.CompleteMultipartUpload(ctx, bucket, object, uploadID, uploadedParts, opts)
return
}
// StorageInfo - returns underlying storage statistics.
func (c cacheObjects) StorageInfo(ctx context.Context) (cInfo CacheStorageInfo) {
var total, free uint64
for _, cfs := range c.cache.cfs {
if cfs == nil {
continue
}
info, err := getDiskInfo(cfs.fsPath)
logger.GetReqInfo(ctx).AppendTags("cachePath", cfs.fsPath)
logger.LogIf(ctx, err)
total += info.Total
free += info.Free
}
return CacheStorageInfo{
Total: total,
Free: free,
}
}
// DeleteBucket - marks bucket to be deleted from cache if bucket is deleted from backend.
func (c cacheObjects) DeleteBucket(ctx context.Context, bucket string) (err error) {
deleteBucketFn := c.DeleteBucketFn
var toDel []*cacheFSObjects
for _, cfs := range c.cache.cfs {
// ignore disk-caches that might be missing/offline
if cfs == nil {
continue
}
if _, cerr := cfs.GetBucketInfo(ctx, bucket); cerr == nil {
toDel = append(toDel, cfs)
}
}
// perform backend operation
err = deleteBucketFn(ctx, bucket)
if err != nil {
return
}
// move bucket metadata and content to cache's trash dir
for _, d := range toDel {
d.moveBucketToTrash(ctx, bucket)
}
return
}
// newCache initializes the cacheFSObjects for the "drives" specified in config.json
// or the global env overrides.
func newCache(config CacheConfig) (*diskCache, error) {
var cfsObjects []*cacheFSObjects
ctx := logger.SetReqInfo(context.Background(), &logger.ReqInfo{})
formats, err := loadAndValidateCacheFormat(ctx, config.Drives)
if err != nil {
return nil, err
}
for i, dir := range config.Drives {
// skip cacheFSObjects creation for cache drives missing a format.json
if formats[i] == nil {
cfsObjects = append(cfsObjects, nil)
continue
}
if err := checkAtimeSupport(dir); err != nil {
return nil, errors.New("Atime support required for disk caching")
}
cache, err := newCacheFSObjects(dir, config.Expiry, config.MaxUse)
if err != nil {
return nil, err
}
// Start the purging go-routine for entries that have expired
go cache.purge()
// Start trash purge routine for deleted buckets.
go cache.purgeTrash()
cfsObjects = append(cfsObjects, cache)
}
return &diskCache{cfs: cfsObjects}, nil
}
// Return error if Atime is disabled on the O/S
func checkAtimeSupport(dir string) (err error) {
file, err := ioutil.TempFile(dir, "prefix")
if err != nil {
return
}
defer os.Remove(file.Name())
finfo1, err := os.Stat(file.Name())
if err != nil {
return
}
// add a sleep to ensure atime change is detected
time.Sleep(10 * time.Millisecond)
if _, err = io.Copy(ioutil.Discard, file); err != nil {
return
}
finfo2, err := os.Stat(file.Name())
if atime.Get(finfo2).Equal(atime.Get(finfo1)) {
return errors.New("Atime not supported")
}
return
}
// Returns cacheObjects for use by Server.
func newServerCacheObjects(config CacheConfig) (CacheObjectLayer, error) {
// list of disk caches for cache "drives" specified in config.json or MINIO_CACHE_DRIVES env var.
dcache, err := newCache(config)
if err != nil {
return nil, err
}
return &cacheObjects{
cache: dcache,
exclude: config.Exclude,
listPool: NewTreeWalkPool(globalLookupTimeout),
GetObjectFn: func(ctx context.Context, bucket, object string, startOffset int64, length int64, writer io.Writer, etag string, opts ObjectOptions) error {
return newObjectLayerFn().GetObject(ctx, bucket, object, startOffset, length, writer, etag, opts)
},
GetObjectInfoFn: func(ctx context.Context, bucket, object string, opts ObjectOptions) (ObjectInfo, error) {
return newObjectLayerFn().GetObjectInfo(ctx, bucket, object, opts)
},
GetObjectNInfoFn: func(ctx context.Context, bucket, object string, rs *HTTPRangeSpec, h http.Header, lockType LockType, opts ObjectOptions) (gr *GetObjectReader, err error) {
return newObjectLayerFn().GetObjectNInfo(ctx, bucket, object, rs, h, lockType, opts)
},
PutObjectFn: func(ctx context.Context, bucket, object string, data *PutObjReader, opts ObjectOptions) (objInfo ObjectInfo, err error) {
return newObjectLayerFn().PutObject(ctx, bucket, object, data, opts)
},
DeleteObjectFn: func(ctx context.Context, bucket, object string) error {
return newObjectLayerFn().DeleteObject(ctx, bucket, object)
},
DeleteObjectsFn: func(ctx context.Context, bucket string, objects []string) ([]error, error) {
errs := make([]error, len(objects))
for idx, object := range objects {
errs[idx] = newObjectLayerFn().DeleteObject(ctx, bucket, object)
}
return errs, nil
},
ListObjectsFn: func(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (result ListObjectsInfo, err error) {
return newObjectLayerFn().ListObjects(ctx, bucket, prefix, marker, delimiter, maxKeys)
},