minio/cmd/xl-v1-list-objects.go
Harshavardhana fb96779a8a Add large bucket support for erasure coded backend (#5160)
This PR implements an object layer which
combines input erasure sets of XL layers
into a unified namespace.

This object layer extends the existing
erasure coded implementation, it is assumed
in this design that providing > 16 disks is
a static configuration as well i.e if you started
the setup with 32 disks with 4 sets 8 disks per
pack then you would need to provide 4 sets always.

Some design details and restrictions:

- Objects are distributed using consistent ordering
  to a unique erasure coded layer.
- Each pack has its own dsync so locks are synchronized
  properly at pack (erasure layer).
- Each pack still has a maximum of 16 disks
  requirement, you can start with multiple
  such sets statically.
- Static sets set of disks and cannot be
  changed, there is no elastic expansion allowed.
- Static sets set of disks and cannot be
  changed, there is no elastic removal allowed.
- ListObjects() across sets can be noticeably
  slower since List happens on all servers,
  and is merged at this sets layer.

Fixes #5465
Fixes #5464
Fixes #5461
Fixes #5460
Fixes #5459
Fixes #5458
Fixes #5460
Fixes #5488
Fixes #5489
Fixes #5497
Fixes #5496
2018-02-15 17:45:57 -08:00

191 lines
5.8 KiB
Go

/*
* Minio Cloud Storage, (C) 2016 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 (
"sort"
"github.com/minio/minio/pkg/errors"
)
// 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 - used for doing disk.ListDir()
func listDirFactory(isLeaf isLeafFunc, treeWalkIgnoredErrs []error, disks ...StorageAPI) listDirFunc {
// Returns sorted merged entries from all the disks.
listDir := func(bucket, prefixDir, prefixEntry string) (mergedEntries []string, delayIsLeaf bool, err error) {
for _, disk := range disks {
if disk == nil {
continue
}
var entries []string
var newEntries []string
entries, err = disk.ListDir(bucket, prefixDir)
if err != nil {
// For any reason disk was deleted or goes offline, continue
// and list from other disks if possible.
if errors.IsErrIgnored(err, treeWalkIgnoredErrs...) {
continue
}
return nil, false, errors.Trace(err)
}
// Find elements in entries which are not in mergedEntries
for _, entry := range entries {
idx := sort.SearchStrings(mergedEntries, entry)
// if entry is already present in mergedEntries don't add.
if idx < len(mergedEntries) && mergedEntries[idx] == entry {
continue
}
newEntries = append(newEntries, entry)
}
if len(newEntries) > 0 {
// Merge the entries and sort it.
mergedEntries = append(mergedEntries, newEntries...)
sort.Strings(mergedEntries)
}
}
mergedEntries, delayIsLeaf = filterListEntries(bucket, prefixDir, mergedEntries, prefixEntry, isLeaf)
return mergedEntries, delayIsLeaf, nil
}
return listDir
}
// listObjects - wrapper function implemented over file tree walk.
func (xl xlObjects) listObjects(bucket, prefix, marker, delimiter string, maxKeys int) (loi ListObjectsInfo, e error) {
// Default is recursive, if delimiter is set then list non recursive.
recursive := true
if delimiter == slashSeparator {
recursive = false
}
heal := false // true only for xl.ListObjectsHeal
walkResultCh, endWalkCh := xl.listPool.Release(listParams{bucket, recursive, marker, prefix, heal})
if walkResultCh == nil {
endWalkCh = make(chan struct{})
isLeaf := xl.isObject
listDir := listDirFactory(isLeaf, xlTreeWalkIgnoredErrs, xl.getLoadBalancedDisks()...)
walkResultCh = startTreeWalk(bucket, prefix, marker, recursive, listDir, isLeaf, endWalkCh)
}
var objInfos []ObjectInfo
var eof bool
var nextMarker string
for i := 0; i < maxKeys; {
walkResult, ok := <-walkResultCh
if !ok {
// Closed channel.
eof = true
break
}
// For any walk error return right away.
if walkResult.err != nil {
return loi, toObjectErr(walkResult.err, bucket, prefix)
}
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
objInfo, err = xl.getObjectInfo(bucket, entry)
if err != nil {
// Ignore errFileNotFound as the object might have got
// deleted in the interim period of listing and getObjectInfo(),
// ignore quorum error as it might be an entry from an outdated disk.
switch errors.Cause(err) {
case errFileNotFound, errXLReadQuorum:
continue
}
return loi, toObjectErr(err, bucket, prefix)
}
}
nextMarker = objInfo.Name
objInfos = append(objInfos, objInfo)
i++
if walkResult.end {
eof = true
break
}
}
params := listParams{bucket, recursive, nextMarker, prefix, heal}
if !eof {
xl.listPool.Set(params, walkResultCh, endWalkCh)
}
result := ListObjectsInfo{IsTruncated: !eof}
for _, objInfo := range objInfos {
result.NextMarker = objInfo.Name
if objInfo.IsDir {
result.Prefixes = append(result.Prefixes, objInfo.Name)
continue
}
result.Objects = append(result.Objects, objInfo)
}
return result, nil
}
// ListObjects - list all objects at prefix, delimited by '/'.
func (xl xlObjects) ListObjects(bucket, prefix, marker, delimiter string, maxKeys int) (loi ListObjectsInfo, e error) {
if err := checkListObjsArgs(bucket, prefix, marker, delimiter, xl); err != nil {
return loi, err
}
// With max keys of zero we have reached eof, return right here.
if maxKeys == 0 {
return loi, nil
}
// Marker is set validate pre-condition.
if marker != "" {
// Marker not common with prefix is not implemented.Send an empty response
if !hasPrefix(marker, prefix) {
return ListObjectsInfo{}, e
}
}
// For delimiter and prefix as '/' we do not list anything at all
// since according to s3 spec we stop at the 'delimiter' along
// with the prefix. On a flat namespace with 'prefix' as '/'
// we don't have any entries, since all the keys are of form 'keyName/...'
if delimiter == slashSeparator && prefix == slashSeparator {
return loi, nil
}
// Over flowing count - reset to maxObjectList.
if maxKeys < 0 || maxKeys > maxObjectList {
maxKeys = maxObjectList
}
// Initiate a list operation, if successful filter and return quickly.
listObjInfo, err := xl.listObjects(bucket, prefix, marker, delimiter, maxKeys)
if err == nil {
// We got the entries successfully return.
return listObjInfo, nil
}
// Return error at the end.
return loi, toObjectErr(err, bucket, prefix)
}