1 // Copyright (C) The Arvados Authors. All rights reserved.
3 // SPDX-License-Identifier: AGPL-3.0
24 "git.curoverse.com/arvados.git/sdk/go/arvados"
25 "github.com/AdRoll/goamz/aws"
26 "github.com/AdRoll/goamz/s3"
30 s3DefaultReadTimeout = arvados.Duration(10 * time.Minute)
31 s3DefaultConnectTimeout = arvados.Duration(time.Minute)
35 // ErrS3TrashDisabled is returned by Trash if that operation
36 // is impossible with the current config.
37 ErrS3TrashDisabled = fmt.Errorf("trash function is disabled because -trash-lifetime=0 and -s3-unsafe-delete=false")
39 s3AccessKeyFile string
40 s3SecretKeyFile string
45 s3RaceWindow time.Duration
53 maxClockSkew = 600 * time.Second
54 nearlyRFC1123 = "Mon, 2 Jan 2006 15:04:05 GMT"
57 type s3VolumeAdder struct {
61 // String implements flag.Value
62 func (s *s3VolumeAdder) String() string {
66 func (s *s3VolumeAdder) Set(bucketName string) error {
68 return fmt.Errorf("no container name given")
70 if s3AccessKeyFile == "" || s3SecretKeyFile == "" {
71 return fmt.Errorf("-s3-access-key-file and -s3-secret-key-file arguments must given before -s3-bucket-volume")
73 if deprecated.flagSerializeIO {
74 log.Print("Notice: -serialize is not supported by s3-bucket volumes.")
76 s.Config.Volumes = append(s.Config.Volumes, &S3Volume{
78 AccessKeyFile: s3AccessKeyFile,
79 SecretKeyFile: s3SecretKeyFile,
82 RaceWindow: arvados.Duration(s3RaceWindow),
83 S3Replication: s3Replication,
84 UnsafeDelete: s3UnsafeDelete,
85 ReadOnly: deprecated.flagReadonly,
91 func s3regions() (okList []string) {
92 for r := range aws.Regions {
93 okList = append(okList, r)
99 VolumeTypes = append(VolumeTypes, func() VolumeWithExamples { return &S3Volume{} })
101 flag.Var(&s3VolumeAdder{theConfig},
103 "Use the given bucket as a storage volume. Can be given multiple times.")
108 fmt.Sprintf("AWS region used for subsequent -s3-bucket-volume arguments. Allowed values are %+q.", s3regions()))
113 "Endpoint URL used for subsequent -s3-bucket-volume arguments. If blank, use the AWS endpoint corresponding to the -s3-region argument. For Google Storage, use \"https://storage.googleapis.com\".")
116 "s3-access-key-file",
118 "`File` containing the access key used for subsequent -s3-bucket-volume arguments.")
121 "s3-secret-key-file",
123 "`File` containing the secret key used for subsequent -s3-bucket-volume arguments.")
128 "Maximum eventual consistency latency for subsequent -s3-bucket-volume arguments.")
133 "Replication level reported to clients for subsequent -s3-bucket-volume arguments.")
138 "EXPERIMENTAL. Enable deletion (garbage collection) even when trash lifetime is zero, even though there are known race conditions that can cause data loss.")
141 // S3Volume implements Volume using an S3 bucket.
142 type S3Volume struct {
148 LocationConstraint bool
151 ConnectTimeout arvados.Duration
152 ReadTimeout arvados.Duration
153 RaceWindow arvados.Duration
156 StorageClasses []string
163 // Examples implements VolumeWithExamples.
164 func (*S3Volume) Examples() []Volume {
167 AccessKeyFile: "/etc/aws_s3_access_key.txt",
168 SecretKeyFile: "/etc/aws_s3_secret_key.txt",
171 Bucket: "example-bucket-name",
174 RaceWindow: arvados.Duration(24 * time.Hour),
175 ConnectTimeout: arvados.Duration(time.Minute),
176 ReadTimeout: arvados.Duration(5 * time.Minute),
179 AccessKeyFile: "/etc/gce_s3_access_key.txt",
180 SecretKeyFile: "/etc/gce_s3_secret_key.txt",
181 Endpoint: "https://storage.googleapis.com",
183 Bucket: "example-bucket-name",
186 RaceWindow: arvados.Duration(24 * time.Hour),
187 ConnectTimeout: arvados.Duration(time.Minute),
188 ReadTimeout: arvados.Duration(5 * time.Minute),
193 // Type implements Volume.
194 func (*S3Volume) Type() string {
198 // Start populates private fields and verifies the configuration is
200 func (v *S3Volume) Start() error {
201 region, ok := aws.Regions[v.Region]
202 if v.Endpoint == "" {
204 return fmt.Errorf("unrecognized region %+q; try specifying -s3-endpoint instead", v.Region)
207 return fmt.Errorf("refusing to use AWS region name %+q with endpoint %+q; "+
208 "specify empty endpoint (\"-s3-endpoint=\") or use a different region name", v.Region, v.Endpoint)
212 S3Endpoint: v.Endpoint,
213 S3LocationConstraint: v.LocationConstraint,
219 auth.AccessKey, err = readKeyFromFile(v.AccessKeyFile)
223 auth.SecretKey, err = readKeyFromFile(v.SecretKeyFile)
228 // Zero timeouts mean "wait forever", which is a bad
229 // default. Default to long timeouts instead.
230 if v.ConnectTimeout == 0 {
231 v.ConnectTimeout = s3DefaultConnectTimeout
233 if v.ReadTimeout == 0 {
234 v.ReadTimeout = s3DefaultReadTimeout
237 client := s3.New(auth, region)
238 client.ConnectTimeout = time.Duration(v.ConnectTimeout)
239 client.ReadTimeout = time.Duration(v.ReadTimeout)
240 v.bucket = &s3bucket{
249 // DeviceID returns a globally unique ID for the storage bucket.
250 func (v *S3Volume) DeviceID() string {
251 return "s3://" + v.Endpoint + "/" + v.Bucket
254 func (v *S3Volume) getReaderWithContext(ctx context.Context, loc string) (rdr io.ReadCloser, err error) {
255 ready := make(chan bool)
257 rdr, err = v.getReader(loc)
264 theConfig.debugLogf("s3: abandoning getReader(): %s", ctx.Err())
271 return nil, ctx.Err()
275 // getReader wraps (Bucket)GetReader.
277 // In situations where (Bucket)GetReader would fail because the block
278 // disappeared in a Trash race, getReader calls fixRace to recover the
279 // data, and tries again.
280 func (v *S3Volume) getReader(loc string) (rdr io.ReadCloser, err error) {
281 rdr, err = v.bucket.GetReader(loc)
282 err = v.translateError(err)
283 if err == nil || !os.IsNotExist(err) {
287 _, err = v.bucket.Head("recent/"+loc, nil)
288 err = v.translateError(err)
290 // If we can't read recent/X, there's no point in
291 // trying fixRace. Give up.
299 rdr, err = v.bucket.GetReader(loc)
301 log.Printf("warning: reading %s after successful fixRace: %s", loc, err)
302 err = v.translateError(err)
307 // Get a block: copy the block data into buf, and return the number of
309 func (v *S3Volume) Get(ctx context.Context, loc string, buf []byte) (int, error) {
310 rdr, err := v.getReaderWithContext(ctx, loc)
316 ready := make(chan bool)
321 n, err = io.ReadFull(rdr, buf)
324 case nil, io.EOF, io.ErrUnexpectedEOF:
327 err = v.translateError(err)
332 theConfig.debugLogf("s3: interrupting ReadFull() with Close() because %s", ctx.Err())
334 // Must wait for ReadFull to return, to ensure it
335 // doesn't write to buf after we return.
336 theConfig.debugLogf("s3: waiting for ReadFull() to fail")
344 // Compare the given data with the stored data.
345 func (v *S3Volume) Compare(ctx context.Context, loc string, expect []byte) error {
346 errChan := make(chan error, 1)
348 _, err := v.bucket.Head("recent/"+loc, nil)
355 case err = <-errChan:
358 // Checking for "loc" itself here would interfere with
359 // future GET requests.
361 // On AWS, if X doesn't exist, a HEAD or GET request
362 // for X causes X's non-existence to be cached. Thus,
363 // if we test for X, then create X and return a
364 // signature to our client, the client might still get
365 // 404 from all keepstores when trying to read it.
367 // To avoid this, we avoid doing HEAD X or GET X until
368 // we know X has been written.
370 // Note that X might exist even though recent/X
371 // doesn't: for example, the response to HEAD recent/X
372 // might itself come from a stale cache. In such
373 // cases, we will return a false negative and
374 // PutHandler might needlessly create another replica
375 // on a different volume. That's not ideal, but it's
376 // better than passing the eventually-consistent
377 // problem on to our clients.
378 return v.translateError(err)
380 rdr, err := v.getReaderWithContext(ctx, loc)
385 return v.translateError(compareReaderWithBuf(ctx, rdr, expect, loc[:32]))
388 // Put writes a block.
389 func (v *S3Volume) Put(ctx context.Context, loc string, block []byte) error {
391 return MethodDisabledError
396 md5, err := hex.DecodeString(loc)
400 opts.ContentMD5 = base64.StdEncoding.EncodeToString(md5)
403 // Send the block data through a pipe, so that (if we need to)
404 // we can close the pipe early and abandon our PutReader()
405 // goroutine, without worrying about PutReader() accessing our
406 // block buffer after we release it.
407 bufr, bufw := io.Pipe()
409 io.Copy(bufw, bytes.NewReader(block))
414 ready := make(chan bool)
417 if ctx.Err() != nil {
418 theConfig.debugLogf("%s: abandoned PutReader goroutine finished with err: %s", v, err)
422 err = v.bucket.PutReader(loc, bufr, int64(size), "application/octet-stream", s3ACL, opts)
426 err = v.bucket.PutReader("recent/"+loc, nil, 0, "application/octet-stream", s3ACL, s3.Options{})
430 theConfig.debugLogf("%s: taking PutReader's input away: %s", v, ctx.Err())
431 // Our pipe might be stuck in Write(), waiting for
432 // PutReader() to read. If so, un-stick it. This means
433 // PutReader will get corrupt data, but that's OK: the
434 // size and MD5 won't match, so the write will fail.
435 go io.Copy(ioutil.Discard, bufr)
436 // CloseWithError() will return once pending I/O is done.
437 bufw.CloseWithError(ctx.Err())
438 theConfig.debugLogf("%s: abandoning PutReader goroutine", v)
441 // Unblock pipe in case PutReader did not consume it.
442 io.Copy(ioutil.Discard, bufr)
443 return v.translateError(err)
447 // Touch sets the timestamp for the given locator to the current time.
448 func (v *S3Volume) Touch(loc string) error {
450 return MethodDisabledError
452 _, err := v.bucket.Head(loc, nil)
453 err = v.translateError(err)
454 if os.IsNotExist(err) && v.fixRace(loc) {
455 // The data object got trashed in a race, but fixRace
457 } else if err != nil {
460 err = v.bucket.PutReader("recent/"+loc, nil, 0, "application/octet-stream", s3ACL, s3.Options{})
461 return v.translateError(err)
464 // Mtime returns the stored timestamp for the given locator.
465 func (v *S3Volume) Mtime(loc string) (time.Time, error) {
466 _, err := v.bucket.Head(loc, nil)
468 return zeroTime, v.translateError(err)
470 resp, err := v.bucket.Head("recent/"+loc, nil)
471 err = v.translateError(err)
472 if os.IsNotExist(err) {
473 // The data object X exists, but recent/X is missing.
474 err = v.bucket.PutReader("recent/"+loc, nil, 0, "application/octet-stream", s3ACL, s3.Options{})
476 log.Printf("error: creating %q: %s", "recent/"+loc, err)
477 return zeroTime, v.translateError(err)
479 log.Printf("info: created %q to migrate existing block to new storage scheme", "recent/"+loc)
480 resp, err = v.bucket.Head("recent/"+loc, nil)
482 log.Printf("error: created %q but HEAD failed: %s", "recent/"+loc, err)
483 return zeroTime, v.translateError(err)
485 } else if err != nil {
486 // HEAD recent/X failed for some other reason.
489 return v.lastModified(resp)
492 // IndexTo writes a complete list of locators with the given prefix
493 // for which Get() can retrieve data.
494 func (v *S3Volume) IndexTo(prefix string, writer io.Writer) error {
495 // Use a merge sort to find matching sets of X and recent/X.
497 Bucket: v.bucket.Bucket,
499 PageSize: v.IndexPageSize,
502 Bucket: v.bucket.Bucket,
503 Prefix: "recent/" + prefix,
504 PageSize: v.IndexPageSize,
506 v.bucket.stats.Tick(&v.bucket.stats.Ops, &v.bucket.stats.ListOps)
507 v.bucket.stats.Tick(&v.bucket.stats.Ops, &v.bucket.stats.ListOps)
508 for data, recent := dataL.First(), recentL.First(); data != nil; data = dataL.Next() {
509 v.bucket.stats.Tick(&v.bucket.stats.Ops, &v.bucket.stats.ListOps)
511 // Conveniently, "recent/*" and "trash/*" are
512 // lexically greater than all hex-encoded data
513 // hashes, so stopping here avoids iterating
514 // over all of them needlessly with dataL.
517 if !v.isKeepBlock(data.Key) {
521 // stamp is the list entry we should use to report the
522 // last-modified time for this data block: it will be
523 // the recent/X entry if one exists, otherwise the
524 // entry for the data block itself.
527 // Advance to the corresponding recent/X marker, if any
529 if cmp := strings.Compare(recent.Key[7:], data.Key); cmp < 0 {
530 recent = recentL.Next()
531 v.bucket.stats.Tick(&v.bucket.stats.Ops, &v.bucket.stats.ListOps)
535 recent = recentL.Next()
536 v.bucket.stats.Tick(&v.bucket.stats.Ops, &v.bucket.stats.ListOps)
539 // recent/X marker is missing: we'll
540 // use the timestamp on the data
545 t, err := time.Parse(time.RFC3339, stamp.LastModified)
549 fmt.Fprintf(writer, "%s+%d %d\n", data.Key, data.Size, t.UnixNano())
554 // Trash a Keep block.
555 func (v *S3Volume) Trash(loc string) error {
557 return MethodDisabledError
559 if t, err := v.Mtime(loc); err != nil {
561 } else if time.Since(t) < theConfig.BlobSignatureTTL.Duration() {
564 if theConfig.TrashLifetime == 0 {
566 return ErrS3TrashDisabled
568 return v.translateError(v.bucket.Del(loc))
570 err := v.checkRaceWindow(loc)
574 err = v.safeCopy("trash/"+loc, loc)
578 return v.translateError(v.bucket.Del(loc))
581 // checkRaceWindow returns a non-nil error if trash/loc is, or might
582 // be, in the race window (i.e., it's not safe to trash loc).
583 func (v *S3Volume) checkRaceWindow(loc string) error {
584 resp, err := v.bucket.Head("trash/"+loc, nil)
585 err = v.translateError(err)
586 if os.IsNotExist(err) {
587 // OK, trash/X doesn't exist so we're not in the race
590 } else if err != nil {
591 // Error looking up trash/X. We don't know whether
592 // we're in the race window
595 t, err := v.lastModified(resp)
597 // Can't parse timestamp
600 safeWindow := t.Add(theConfig.TrashLifetime.Duration()).Sub(time.Now().Add(time.Duration(v.RaceWindow)))
602 // We can't count on "touch trash/X" to prolong
603 // trash/X's lifetime. The new timestamp might not
604 // become visible until now+raceWindow, and EmptyTrash
605 // is allowed to delete trash/X before then.
606 return fmt.Errorf("same block is already in trash, and safe window ended %s ago", -safeWindow)
608 // trash/X exists, but it won't be eligible for deletion until
609 // after now+raceWindow, so it's safe to overwrite it.
613 // safeCopy calls PutCopy, and checks the response to make sure the
614 // copy succeeded and updated the timestamp on the destination object
615 // (PutCopy returns 200 OK if the request was received, even if the
617 func (v *S3Volume) safeCopy(dst, src string) error {
618 resp, err := v.bucket.PutCopy(dst, s3ACL, s3.CopyOptions{
619 ContentType: "application/octet-stream",
620 MetadataDirective: "REPLACE",
621 }, v.bucket.Name+"/"+src)
622 err = v.translateError(err)
626 if t, err := time.Parse(time.RFC3339Nano, resp.LastModified); err != nil {
627 return fmt.Errorf("PutCopy succeeded but did not return a timestamp: %q: %s", resp.LastModified, err)
628 } else if time.Now().Sub(t) > maxClockSkew {
629 return fmt.Errorf("PutCopy succeeded but returned an old timestamp: %q: %s", resp.LastModified, t)
634 // Get the LastModified header from resp, and parse it as RFC1123 or
635 // -- if it isn't valid RFC1123 -- as Amazon's variant of RFC1123.
636 func (v *S3Volume) lastModified(resp *http.Response) (t time.Time, err error) {
637 s := resp.Header.Get("Last-Modified")
638 t, err = time.Parse(time.RFC1123, s)
639 if err != nil && s != "" {
640 // AWS example is "Sun, 1 Jan 2006 12:00:00 GMT",
641 // which isn't quite "Sun, 01 Jan 2006 12:00:00 GMT"
642 // as required by HTTP spec. If it's not a valid HTTP
643 // header value, it's probably AWS (or s3test) giving
644 // us a nearly-RFC1123 timestamp.
645 t, err = time.Parse(nearlyRFC1123, s)
650 // Untrash moves block from trash back into store
651 func (v *S3Volume) Untrash(loc string) error {
652 err := v.safeCopy(loc, "trash/"+loc)
656 err = v.bucket.PutReader("recent/"+loc, nil, 0, "application/octet-stream", s3ACL, s3.Options{})
657 return v.translateError(err)
660 // Status returns a *VolumeStatus representing the current in-use
661 // storage capacity and a fake available capacity that doesn't make
662 // the volume seem full or nearly-full.
663 func (v *S3Volume) Status() *VolumeStatus {
664 return &VolumeStatus{
666 BytesFree: BlockSize * 1000,
671 // InternalStats returns bucket I/O and API call counters.
672 func (v *S3Volume) InternalStats() interface{} {
673 return &v.bucket.stats
676 // String implements fmt.Stringer.
677 func (v *S3Volume) String() string {
678 return fmt.Sprintf("s3-bucket:%+q", v.Bucket)
681 // Writable returns false if all future Put, Mtime, and Delete calls
682 // are expected to fail.
683 func (v *S3Volume) Writable() bool {
687 // Replication returns the storage redundancy of the underlying
688 // device. Configured via command line flag.
689 func (v *S3Volume) Replication() int {
690 return v.S3Replication
693 // GetStorageClasses implements Volume
694 func (v *S3Volume) GetStorageClasses() []string {
695 return v.StorageClasses
698 var s3KeepBlockRegexp = regexp.MustCompile(`^[0-9a-f]{32}$`)
700 func (v *S3Volume) isKeepBlock(s string) bool {
701 return s3KeepBlockRegexp.MatchString(s)
704 // fixRace(X) is called when "recent/X" exists but "X" doesn't
705 // exist. If the timestamps on "recent/"+loc and "trash/"+loc indicate
706 // there was a race between Put and Trash, fixRace recovers from the
707 // race by Untrashing the block.
708 func (v *S3Volume) fixRace(loc string) bool {
709 trash, err := v.bucket.Head("trash/"+loc, nil)
711 if !os.IsNotExist(v.translateError(err)) {
712 log.Printf("error: fixRace: HEAD %q: %s", "trash/"+loc, err)
716 trashTime, err := v.lastModified(trash)
718 log.Printf("error: fixRace: parse %q: %s", trash.Header.Get("Last-Modified"), err)
722 recent, err := v.bucket.Head("recent/"+loc, nil)
724 log.Printf("error: fixRace: HEAD %q: %s", "recent/"+loc, err)
727 recentTime, err := v.lastModified(recent)
729 log.Printf("error: fixRace: parse %q: %s", recent.Header.Get("Last-Modified"), err)
733 ageWhenTrashed := trashTime.Sub(recentTime)
734 if ageWhenTrashed >= theConfig.BlobSignatureTTL.Duration() {
735 // No evidence of a race: block hasn't been written
736 // since it became eligible for Trash. No fix needed.
740 log.Printf("notice: fixRace: %q: trashed at %s but touched at %s (age when trashed = %s < %s)", loc, trashTime, recentTime, ageWhenTrashed, theConfig.BlobSignatureTTL)
741 log.Printf("notice: fixRace: copying %q to %q to recover from race between Put/Touch and Trash", "recent/"+loc, loc)
742 err = v.safeCopy(loc, "trash/"+loc)
744 log.Printf("error: fixRace: %s", err)
750 func (v *S3Volume) translateError(err error) error {
751 switch err := err.(type) {
753 if (err.StatusCode == http.StatusNotFound && err.Code == "NoSuchKey") ||
754 strings.Contains(err.Error(), "Not Found") {
755 return os.ErrNotExist
757 // Other 404 errors like NoSuchVersion and
758 // NoSuchBucket are different problems which should
759 // get called out downstream, so we don't convert them
760 // to os.ErrNotExist.
765 // EmptyTrash looks for trashed blocks that exceeded TrashLifetime
766 // and deletes them from the volume.
767 func (v *S3Volume) EmptyTrash() {
768 var bytesInTrash, blocksInTrash, bytesDeleted, blocksDeleted int64
770 // Define "ready to delete" as "...when EmptyTrash started".
773 emptyOneKey := func(trash *s3.Key) {
775 if !v.isKeepBlock(loc) {
778 atomic.AddInt64(&bytesInTrash, trash.Size)
779 atomic.AddInt64(&blocksInTrash, 1)
781 trashT, err := time.Parse(time.RFC3339, trash.LastModified)
783 log.Printf("warning: %s: EmptyTrash: %q: parse %q: %s", v, trash.Key, trash.LastModified, err)
786 recent, err := v.bucket.Head("recent/"+loc, nil)
787 if err != nil && os.IsNotExist(v.translateError(err)) {
788 log.Printf("warning: %s: EmptyTrash: found trash marker %q but no %q (%s); calling Untrash", v, trash.Key, "recent/"+loc, err)
791 log.Printf("error: %s: EmptyTrash: Untrash(%q): %s", v, loc, err)
794 } else if err != nil {
795 log.Printf("warning: %s: EmptyTrash: HEAD %q: %s", v, "recent/"+loc, err)
798 recentT, err := v.lastModified(recent)
800 log.Printf("warning: %s: EmptyTrash: %q: parse %q: %s", v, "recent/"+loc, recent.Header.Get("Last-Modified"), err)
803 if trashT.Sub(recentT) < theConfig.BlobSignatureTTL.Duration() {
804 if age := startT.Sub(recentT); age >= theConfig.BlobSignatureTTL.Duration()-time.Duration(v.RaceWindow) {
805 // recent/loc is too old to protect
806 // loc from being Trashed again during
807 // the raceWindow that starts if we
808 // delete trash/X now.
810 // Note this means (TrashCheckInterval
811 // < BlobSignatureTTL - raceWindow) is
812 // necessary to avoid starvation.
813 log.Printf("notice: %s: EmptyTrash: detected old race for %q, calling fixRace + Touch", v, loc)
818 _, err := v.bucket.Head(loc, nil)
819 if os.IsNotExist(err) {
820 log.Printf("notice: %s: EmptyTrash: detected recent race for %q, calling fixRace", v, loc)
823 } else if err != nil {
824 log.Printf("warning: %s: EmptyTrash: HEAD %q: %s", v, loc, err)
828 if startT.Sub(trashT) < theConfig.TrashLifetime.Duration() {
831 err = v.bucket.Del(trash.Key)
833 log.Printf("warning: %s: EmptyTrash: deleting %q: %s", v, trash.Key, err)
836 atomic.AddInt64(&bytesDeleted, trash.Size)
837 atomic.AddInt64(&blocksDeleted, 1)
839 _, err = v.bucket.Head(loc, nil)
841 log.Printf("warning: %s: EmptyTrash: HEAD %q succeeded immediately after deleting %q", v, loc, loc)
844 if !os.IsNotExist(v.translateError(err)) {
845 log.Printf("warning: %s: EmptyTrash: HEAD %q: %s", v, loc, err)
848 err = v.bucket.Del("recent/" + loc)
850 log.Printf("warning: %s: EmptyTrash: deleting %q: %s", v, "recent/"+loc, err)
854 var wg sync.WaitGroup
855 todo := make(chan *s3.Key, theConfig.EmptyTrashWorkers)
856 for i := 0; i < 1 || i < theConfig.EmptyTrashWorkers; i++ {
860 for key := range todo {
867 Bucket: v.bucket.Bucket,
869 PageSize: v.IndexPageSize,
871 for trash := trashL.First(); trash != nil; trash = trashL.Next() {
877 if err := trashL.Error(); err != nil {
878 log.Printf("error: %s: EmptyTrash: lister: %s", v, err)
880 log.Printf("EmptyTrash stats for %v: Deleted %v bytes in %v blocks. Remaining in trash: %v bytes in %v blocks.", v.String(), bytesDeleted, blocksDeleted, bytesInTrash-bytesDeleted, blocksInTrash-blocksDeleted)
883 type s3Lister struct {
892 // First fetches the first page and returns the first item. It returns
893 // nil if the response is the empty set or an error occurs.
894 func (lister *s3Lister) First() *s3.Key {
899 // Next returns the next item, fetching the next page if necessary. It
900 // returns nil if the last available item has already been fetched, or
902 func (lister *s3Lister) Next() *s3.Key {
903 if len(lister.buf) == 0 && lister.nextMarker != "" {
909 // Return the most recent error encountered by First or Next.
910 func (lister *s3Lister) Error() error {
914 func (lister *s3Lister) getPage() {
915 resp, err := lister.Bucket.List(lister.Prefix, "", lister.nextMarker, lister.PageSize)
916 lister.nextMarker = ""
921 if resp.IsTruncated {
922 lister.nextMarker = resp.NextMarker
924 lister.buf = make([]s3.Key, 0, len(resp.Contents))
925 for _, key := range resp.Contents {
926 if !strings.HasPrefix(key.Key, lister.Prefix) {
927 log.Printf("warning: s3Lister: S3 Bucket.List(prefix=%q) returned key %q", lister.Prefix, key.Key)
930 lister.buf = append(lister.buf, key)
934 func (lister *s3Lister) pop() (k *s3.Key) {
935 if len(lister.buf) > 0 {
937 lister.buf = lister.buf[1:]
942 // s3bucket wraps s3.bucket and counts I/O and API usage stats.
943 type s3bucket struct {
948 func (b *s3bucket) GetReader(path string) (io.ReadCloser, error) {
949 rdr, err := b.Bucket.GetReader(path)
950 b.stats.Tick(&b.stats.Ops, &b.stats.GetOps)
952 return NewCountingReader(rdr, b.stats.TickInBytes), err
955 func (b *s3bucket) Head(path string, headers map[string][]string) (*http.Response, error) {
956 resp, err := b.Bucket.Head(path, headers)
957 b.stats.Tick(&b.stats.Ops, &b.stats.HeadOps)
962 func (b *s3bucket) PutReader(path string, r io.Reader, length int64, contType string, perm s3.ACL, options s3.Options) error {
964 // goamz will only send Content-Length: 0 when reader
965 // is nil due to net.http.Request.ContentLength
966 // behavior. Otherwise, Content-Length header is
967 // omitted which will cause some S3 services
968 // (including AWS and Ceph RadosGW) to fail to create
972 r = NewCountingReader(r, b.stats.TickOutBytes)
974 err := b.Bucket.PutReader(path, r, length, contType, perm, options)
975 b.stats.Tick(&b.stats.Ops, &b.stats.PutOps)
980 func (b *s3bucket) Del(path string) error {
981 err := b.Bucket.Del(path)
982 b.stats.Tick(&b.stats.Ops, &b.stats.DelOps)
987 type s3bucketStats struct {
997 func (s *s3bucketStats) TickErr(err error) {
1001 errType := fmt.Sprintf("%T", err)
1002 if err, ok := err.(*s3.Error); ok {
1003 errType = errType + fmt.Sprintf(" %d %s", err.StatusCode, err.Code)
1005 s.statsTicker.TickErr(err, errType)