optimize listObjects to list only from 3 random disks (#9184)

master
Harshavardhana 5 years ago committed by GitHub
parent cfc9cfd84a
commit da04cb91ce
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
  1. 66
      cmd/xl-sets.go
  2. 62
      cmd/xl-zones.go

@ -887,8 +887,7 @@ func leastEntry(entryChs []FileInfoCh, entries []FileInfo, entriesValid []bool)
}
// mergeEntriesCh - merges FileInfo channel to entries upto maxKeys.
// If partialQuorumOnly is set only objects that does not have full quorum is returned.
func mergeEntriesCh(entryChs []FileInfoCh, maxKeys int, totalDrives int, partialQuorumOnly bool) (entries FilesInfo) {
func mergeEntriesCh(entryChs []FileInfoCh, maxKeys int, ndisks int) (entries FilesInfo) {
var i = 0
entriesInfos := make([]FileInfo, len(entryChs))
entriesValid := make([]bool, len(entryChs))
@ -899,27 +898,11 @@ func mergeEntriesCh(entryChs []FileInfoCh, maxKeys int, totalDrives int, partial
break
}
rquorum := fi.Quorum
// Quorum is zero for all directories.
if rquorum == 0 {
// Choose N/2 quoroum for directory entries.
rquorum = totalDrives / 2
}
if partialQuorumOnly {
// When healing is enabled, we should
// list only objects which need healing.
if quorumCount == totalDrives {
// Skip good entries.
if quorumCount < ndisks-1 {
// Skip entries which are not found on upto ndisks.
continue
}
} else {
// Regular listing, we skip entries not in quorum.
if quorumCount < rquorum {
// Skip entries which do not have quorum.
continue
}
}
entries.Files = append(entries.Files, fi)
i++
if i == maxKeys {
@ -951,11 +934,19 @@ func isTruncated(entryChs []FileInfoCh, entries []FileInfo, entriesValid []bool)
return isTruncated
}
// Starts a walk channel across all disks and returns a slice.
func (s *xlSets) startMergeWalks(ctx context.Context, bucket, prefix, marker string, recursive bool, endWalkCh <-chan struct{}) []FileInfoCh {
return s.startMergeWalksN(ctx, bucket, prefix, marker, recursive, endWalkCh, -1)
}
// Starts a walk channel across all disks and returns a slice of
// FileInfo channels which can be read from.
func (s *xlSets) startMergeWalksN(ctx context.Context, bucket, prefix, marker string, recursive bool, endWalkCh <-chan struct{}, ndisks int) []FileInfoCh {
var entryChs []FileInfoCh
var success int
for _, set := range s.sets {
for _, disk := range set.getDisks() {
// Reset for the next erasure set.
success = ndisks
for _, disk := range set.getLoadBalancedDisks() {
if disk == nil {
// Disk can be offline
continue
@ -968,6 +959,10 @@ func (s *xlSets) startMergeWalks(ctx context.Context, bucket, prefix, marker str
entryChs = append(entryChs, FileInfoCh{
Ch: entryCh,
})
success--
if success == 0 {
break
}
}
}
return entryChs
@ -977,7 +972,8 @@ func (s *xlSets) listObjectsNonSlash(ctx context.Context, bucket, prefix, marker
endWalkCh := make(chan struct{})
defer close(endWalkCh)
entryChs := s.startMergeWalks(context.Background(), bucket, prefix, "", true, endWalkCh)
const ndisks = 3
entryChs := s.startMergeWalksN(context.Background(), bucket, prefix, "", true, endWalkCh, ndisks)
var objInfos []ObjectInfo
var eof bool
@ -989,18 +985,15 @@ func (s *xlSets) listObjectsNonSlash(ctx context.Context, bucket, prefix, marker
if len(objInfos) == maxKeys {
break
}
result, quorumCount, ok := leastEntry(entryChs, entries, entriesValid)
if !ok {
eof = true
break
}
rquorum := result.Quorum
// Quorum is zero for all directories.
if rquorum == 0 {
// Choose N/2 quorum for directory entries.
rquorum = s.drivesPerSet / 2
}
if quorumCount < rquorum {
if quorumCount < ndisks-1 {
// Skip entries which are not found on upto ndisks.
continue
}
@ -1080,7 +1073,7 @@ func (s *xlSets) listObjectsNonSlash(ctx context.Context, bucket, prefix, marker
// walked and merged at this layer. Resulting value through the merge process sends
// the data in lexically sorted order.
// If partialQuorumOnly is set only objects that does not have full quorum is returned.
func (s *xlSets) listObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int, partialQuorumOnly bool) (loi ListObjectsInfo, err error) {
func (s *xlSets) listObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (loi ListObjectsInfo, err error) {
if err = checkListObjsArgs(ctx, bucket, prefix, marker, s); err != nil {
return loi, err
}
@ -1123,13 +1116,16 @@ func (s *xlSets) listObjects(ctx context.Context, bucket, prefix, marker, delimi
recursive = false
}
const ndisks = 3
entryChs, endWalkCh := s.pool.Release(listParams{bucket: bucket, recursive: recursive, marker: marker, prefix: prefix})
if entryChs == nil {
endWalkCh = make(chan struct{})
entryChs = s.startMergeWalks(context.Background(), bucket, prefix, marker, recursive, endWalkCh)
// start file tree walk across at most randomly 3 disks in a set.
entryChs = s.startMergeWalksN(context.Background(), bucket, prefix, marker, recursive, endWalkCh, ndisks)
}
entries := mergeEntriesCh(entryChs, maxKeys, s.drivesPerSet, partialQuorumOnly)
entries := mergeEntriesCh(entryChs, maxKeys, ndisks)
if len(entries.Files) == 0 {
return loi, nil
}
@ -1157,7 +1153,7 @@ func (s *xlSets) listObjects(ctx context.Context, bucket, prefix, marker, delimi
// walked and merged at this layer. Resulting value through the merge process sends
// the data in lexically sorted order.
func (s *xlSets) ListObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (loi ListObjectsInfo, err error) {
return s.listObjects(ctx, bucket, prefix, marker, delimiter, maxKeys, false)
return s.listObjects(ctx, bucket, prefix, marker, delimiter, maxKeys)
}
func (s *xlSets) ListMultipartUploads(ctx context.Context, bucket, prefix, keyMarker, uploadIDMarker, delimiter string, maxUploads int) (result ListMultipartsInfo, err error) {

@ -592,20 +592,16 @@ func (z *xlZones) listObjectsNonSlash(ctx context.Context, bucket, prefix, marke
endWalkCh := make(chan struct{})
defer close(endWalkCh)
const ndisks = 3
for _, zone := range z.zones {
zonesEntryChs = append(zonesEntryChs,
zone.startMergeWalks(ctx, bucket, prefix, "", true, endWalkCh))
zone.startMergeWalksN(ctx, bucket, prefix, "", true, endWalkCh, ndisks))
}
var objInfos []ObjectInfo
var eof bool
var prevPrefix string
var zoneDrivesPerSet []int
for _, zone := range z.zones {
zoneDrivesPerSet = append(zoneDrivesPerSet, zone.drivesPerSet)
}
var zonesEntriesInfos [][]FileInfo
var zonesEntriesValid [][]bool
for _, entryChs := range zonesEntryChs {
@ -617,18 +613,15 @@ func (z *xlZones) listObjectsNonSlash(ctx context.Context, bucket, prefix, marke
if len(objInfos) == maxKeys {
break
}
result, quorumCount, zoneIndex, ok := leastEntryZone(zonesEntryChs, zonesEntriesInfos, zonesEntriesValid)
result, quorumCount, _, ok := leastEntryZone(zonesEntryChs, zonesEntriesInfos, zonesEntriesValid)
if !ok {
eof = true
break
}
rquorum := result.Quorum
// Quorum is zero for all directories.
if rquorum == 0 {
// Choose N/2 quorum for directory entries.
rquorum = zoneDrivesPerSet[zoneIndex] / 2
}
if quorumCount < rquorum {
if quorumCount < ndisks-1 {
// Skip entries which are not found on upto ndisks.
continue
}
@ -704,7 +697,7 @@ func (z *xlZones) listObjectsNonSlash(ctx context.Context, bucket, prefix, marke
return result, nil
}
func (z *xlZones) listObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int, heal bool) (ListObjectsInfo, error) {
func (z *xlZones) listObjects(ctx context.Context, bucket, prefix, marker, delimiter string, maxKeys int) (ListObjectsInfo, error) {
loi := ListObjectsInfo{}
if err := checkListObjsArgs(ctx, bucket, prefix, marker, z); err != nil {
@ -752,22 +745,18 @@ func (z *xlZones) listObjects(ctx context.Context, bucket, prefix, marker, delim
var zonesEntryChs [][]FileInfoCh
var zonesEndWalkCh []chan struct{}
const ndisks = 3
for _, zone := range z.zones {
entryChs, endWalkCh := zone.pool.Release(listParams{bucket, recursive, marker, prefix})
if entryChs == nil {
endWalkCh = make(chan struct{})
entryChs = zone.startMergeWalks(ctx, bucket, prefix, marker, recursive, endWalkCh)
entryChs = zone.startMergeWalksN(ctx, bucket, prefix, marker, recursive, endWalkCh, ndisks)
}
zonesEntryChs = append(zonesEntryChs, entryChs)
zonesEndWalkCh = append(zonesEndWalkCh, endWalkCh)
}
var zoneDrivesPerSet []int
for _, zone := range z.zones {
zoneDrivesPerSet = append(zoneDrivesPerSet, zone.drivesPerSet)
}
entries := mergeZonesEntriesCh(zonesEntryChs, maxKeys, zoneDrivesPerSet, heal)
entries := mergeZonesEntriesCh(zonesEntryChs, maxKeys, ndisks)
if len(entries.Files) == 0 {
return loi, nil
}
@ -873,7 +862,7 @@ func leastEntryZone(zoneEntryChs [][]FileInfoCh, zoneEntries [][]FileInfo, zoneE
}
// mergeZonesEntriesCh - merges FileInfo channel to entries upto maxKeys.
func mergeZonesEntriesCh(zonesEntryChs [][]FileInfoCh, maxKeys int, zoneDrives []int, heal bool) (entries FilesInfo) {
func mergeZonesEntriesCh(zonesEntryChs [][]FileInfoCh, maxKeys int, ndisks int) (entries FilesInfo) {
var i = 0
var zonesEntriesInfos [][]FileInfo
var zonesEntriesValid [][]bool
@ -882,32 +871,17 @@ func mergeZonesEntriesCh(zonesEntryChs [][]FileInfoCh, maxKeys int, zoneDrives [
zonesEntriesValid = append(zonesEntriesValid, make([]bool, len(entryChs)))
}
for {
fi, quorumCount, zoneIndex, valid := leastEntryZone(zonesEntryChs, zonesEntriesInfos, zonesEntriesValid)
if !valid {
fi, quorumCount, _, ok := leastEntryZone(zonesEntryChs, zonesEntriesInfos, zonesEntriesValid)
if !ok {
// We have reached EOF across all entryChs, break the loop.
break
}
rquorum := fi.Quorum
// Quorum is zero for all directories.
if rquorum == 0 {
// Choose N/2 quoroum for directory entries.
rquorum = zoneDrives[zoneIndex] / 2
}
if heal {
// When healing is enabled, we should
// list only objects which need healing.
if quorumCount == zoneDrives[zoneIndex] {
// Skip good entries.
if quorumCount < ndisks-1 {
// Skip entries which are not found on upto ndisks.
continue
}
} else {
// Regular listing, we skip entries not in quorum.
if quorumCount < rquorum {
// Skip entries which do not have quorum.
continue
}
}
entries.Files = append(entries.Files, fi)
i++
if i == maxKeys {
@ -953,7 +927,7 @@ func (z *xlZones) ListObjects(ctx context.Context, bucket, prefix, marker, delim
return z.zones[0].ListObjects(ctx, bucket, prefix, marker, delimiter, maxKeys)
}
return z.listObjects(ctx, bucket, prefix, marker, delimiter, maxKeys, false)
return z.listObjects(ctx, bucket, prefix, marker, delimiter, maxKeys)
}
func (z *xlZones) ListMultipartUploads(ctx context.Context, bucket, prefix, keyMarker, uploadIDMarker, delimiter string, maxUploads int) (ListMultipartsInfo, error) {

Loading…
Cancel
Save