2
2
mirror of https://github.com/octoleo/restic.git synced 2024-12-23 11:28:54 +00:00
restic/internal/index/index.go
greatroar 99755c634b index: Optimize generatePackList
name                 old time/op    new time/op    delta
EncodeIndex/100-8      1.56ms ± 2%    1.48ms ± 3%   -5.37%  (p=0.000 n=10+10)
EncodeIndex/1000-8     14.5ms ± 2%    13.1ms ± 2%   -9.49%  (p=0.000 n=9+10)
EncodeIndex/10000-8     120ms ± 2%     116ms ± 2%   -3.58%  (p=0.000 n=10+10)

name                 old alloc/op   new alloc/op   delta
EncodeIndex/100-8       306kB ± 1%     275kB ± 1%  -10.28%  (p=0.000 n=10+10)
EncodeIndex/1000-8     3.69MB ±11%    2.88MB ± 5%  -22.07%  (p=0.000 n=10+9)
EncodeIndex/10000-8    35.9MB ±11%    31.9MB ±10%  -11.13%  (p=0.005 n=10+10)

name                 old allocs/op  new allocs/op  delta
EncodeIndex/100-8       3.39k ± 0%     2.39k ± 0%  -29.61%  (p=0.000 n=10+10)
EncodeIndex/1000-8      32.6k ± 0%     22.9k ± 0%  -29.63%  (p=0.000 n=10+9)
EncodeIndex/10000-8      326k ± 0%      229k ± 0%  -29.71%  (p=0.000 n=10+10)

The bulk of the allocation rate improvement comes from just removing the
debug.Log calls: every one of those copied a restic.ID to the heap.
2023-01-14 20:41:07 +01:00

580 lines
14 KiB
Go

package index
import (
"context"
"encoding/json"
"io"
"sync"
"time"
"github.com/restic/restic/internal/crypto"
"github.com/restic/restic/internal/errors"
"github.com/restic/restic/internal/restic"
"github.com/restic/restic/internal/debug"
)
// In large repositories, millions of blobs are stored in the repository
// and restic needs to store an index entry for each blob in memory for
// most operations.
// Hence the index data structure defined here is one of the main contributions
// to the total memory requirements of restic.
//
// We store the index entries in indexMaps. In these maps, entries take 56
// bytes each, plus 8/4 = 2 bytes of unused pointers on average, not counting
// malloc and header struct overhead and ignoring duplicates (those are only
// present in edge cases and are also removed by prune runs).
//
// In the index entries, we need to reference the packID. As one pack may
// contain many blobs the packIDs are saved in a separate array and only the index
// within this array is saved in the indexEntry
//
// We assume on average a minimum of 8 blobs per pack; BP=8.
// (Note that for large files there should be 3 blobs per pack as the average chunk
// size is 1.5 MB and the minimum pack size is 4 MB)
//
// We have the following sizes:
// indexEntry: 56 bytes (on amd64)
// each packID: 32 bytes
//
// To save N index entries, we therefore need:
// N * (56 + 2) bytes + N * 32 bytes / BP = N * 62 bytes,
// i.e., fewer than 64 bytes per blob in an index.
// Index holds lookup tables for id -> pack.
type Index struct {
m sync.Mutex
byType [restic.NumBlobTypes]indexMap
packs restic.IDs
final bool // set to true for all indexes read from the backend ("finalized")
ids restic.IDs // set to the IDs of the contained finalized indexes
supersedes restic.IDs
created time.Time
}
// NewIndex returns a new index.
func NewIndex() *Index {
return &Index{
created: time.Now(),
}
}
// addToPacks saves the given pack ID and return the index.
// This procedere allows to use pack IDs which can be easily garbage collected after.
func (idx *Index) addToPacks(id restic.ID) int {
idx.packs = append(idx.packs, id)
return len(idx.packs) - 1
}
const maxuint32 = 1<<32 - 1
func (idx *Index) store(packIndex int, blob restic.Blob) {
// assert that offset and length fit into uint32!
if blob.Offset > maxuint32 || blob.Length > maxuint32 || blob.UncompressedLength > maxuint32 {
panic("offset or length does not fit in uint32. You have packs > 4GB!")
}
m := &idx.byType[blob.Type]
m.add(blob.ID, packIndex, uint32(blob.Offset), uint32(blob.Length), uint32(blob.UncompressedLength))
}
// Final returns true iff the index is already written to the repository, it is
// finalized.
func (idx *Index) Final() bool {
idx.m.Lock()
defer idx.m.Unlock()
return idx.final
}
const (
indexMaxBlobs = 50000
indexMaxBlobsCompressed = 3 * indexMaxBlobs
indexMaxAge = 10 * time.Minute
)
// IndexFull returns true iff the index is "full enough" to be saved as a preliminary index.
var IndexFull = func(idx *Index, compress bool) bool {
idx.m.Lock()
defer idx.m.Unlock()
debug.Log("checking whether index %p is full", idx)
var blobs uint
for typ := range idx.byType {
blobs += idx.byType[typ].len()
}
age := time.Since(idx.created)
var maxBlobs uint
if compress {
maxBlobs = indexMaxBlobsCompressed
} else {
maxBlobs = indexMaxBlobs
}
switch {
case age >= indexMaxAge:
debug.Log("index %p is old enough", idx, age)
return true
case blobs >= maxBlobs:
debug.Log("index %p has %d blobs", idx, blobs)
return true
}
debug.Log("index %p only has %d blobs and is too young (%v)", idx, blobs, age)
return false
}
// StorePack remembers the ids of all blobs of a given pack
// in the index
func (idx *Index) StorePack(id restic.ID, blobs []restic.Blob) {
idx.m.Lock()
defer idx.m.Unlock()
if idx.final {
panic("store new item in finalized index")
}
debug.Log("%v", blobs)
packIndex := idx.addToPacks(id)
for _, blob := range blobs {
idx.store(packIndex, blob)
}
}
func (idx *Index) toPackedBlob(e *indexEntry, t restic.BlobType) restic.PackedBlob {
return restic.PackedBlob{
Blob: restic.Blob{
BlobHandle: restic.BlobHandle{
ID: e.id,
Type: t},
Length: uint(e.length),
Offset: uint(e.offset),
UncompressedLength: uint(e.uncompressedLength),
},
PackID: idx.packs[e.packIndex],
}
}
// Lookup queries the index for the blob ID and returns all entries including
// duplicates. Adds found entries to blobs and returns the result.
func (idx *Index) Lookup(bh restic.BlobHandle, pbs []restic.PackedBlob) []restic.PackedBlob {
idx.m.Lock()
defer idx.m.Unlock()
idx.byType[bh.Type].foreachWithID(bh.ID, func(e *indexEntry) {
pbs = append(pbs, idx.toPackedBlob(e, bh.Type))
})
return pbs
}
// Has returns true iff the id is listed in the index.
func (idx *Index) Has(bh restic.BlobHandle) bool {
idx.m.Lock()
defer idx.m.Unlock()
return idx.byType[bh.Type].get(bh.ID) != nil
}
// LookupSize returns the length of the plaintext content of the blob with the
// given id.
func (idx *Index) LookupSize(bh restic.BlobHandle) (plaintextLength uint, found bool) {
idx.m.Lock()
defer idx.m.Unlock()
e := idx.byType[bh.Type].get(bh.ID)
if e == nil {
return 0, false
}
if e.uncompressedLength != 0 {
return uint(e.uncompressedLength), true
}
return uint(crypto.PlaintextLength(int(e.length))), true
}
// Supersedes returns the list of indexes this index supersedes, if any.
func (idx *Index) Supersedes() restic.IDs {
return idx.supersedes
}
// AddToSupersedes adds the ids to the list of indexes superseded by this
// index. If the index has already been finalized, an error is returned.
func (idx *Index) AddToSupersedes(ids ...restic.ID) error {
idx.m.Lock()
defer idx.m.Unlock()
if idx.final {
return errors.New("index already finalized")
}
idx.supersedes = append(idx.supersedes, ids...)
return nil
}
// Each passes all blobs known to the index to the callback fn. This blocks any
// modification of the index.
func (idx *Index) Each(ctx context.Context, fn func(restic.PackedBlob)) {
idx.m.Lock()
defer idx.m.Unlock()
for typ := range idx.byType {
m := &idx.byType[typ]
m.foreach(func(e *indexEntry) bool {
if ctx.Err() != nil {
return false
}
fn(idx.toPackedBlob(e, restic.BlobType(typ)))
return true
})
}
}
type EachByPackResult struct {
PackID restic.ID
Blobs []restic.Blob
}
// EachByPack returns a channel that yields all blobs known to the index
// grouped by packID but ignoring blobs with a packID in packPlacklist for
// finalized indexes.
// This filtering is used when rebuilding the index where we need to ignore packs
// from the finalized index which have been re-read into a non-finalized index.
// When the context is cancelled, the background goroutine
// terminates. This blocks any modification of the index.
func (idx *Index) EachByPack(ctx context.Context, packBlacklist restic.IDSet) <-chan EachByPackResult {
idx.m.Lock()
ch := make(chan EachByPackResult)
go func() {
defer idx.m.Unlock()
defer close(ch)
byPack := make(map[restic.ID][restic.NumBlobTypes][]*indexEntry)
for typ := range idx.byType {
m := &idx.byType[typ]
m.foreach(func(e *indexEntry) bool {
packID := idx.packs[e.packIndex]
if !idx.final || !packBlacklist.Has(packID) {
v := byPack[packID]
v[typ] = append(v[typ], e)
byPack[packID] = v
}
return true
})
}
for packID, packByType := range byPack {
var result EachByPackResult
result.PackID = packID
for typ, pack := range packByType {
for _, e := range pack {
result.Blobs = append(result.Blobs, idx.toPackedBlob(e, restic.BlobType(typ)).Blob)
}
}
// allow GC once entry is no longer necessary
delete(byPack, packID)
select {
case <-ctx.Done():
return
case ch <- result:
}
}
}()
return ch
}
// Packs returns all packs in this index
func (idx *Index) Packs() restic.IDSet {
idx.m.Lock()
defer idx.m.Unlock()
packs := restic.NewIDSet()
for _, packID := range idx.packs {
packs.Insert(packID)
}
return packs
}
type packJSON struct {
ID restic.ID `json:"id"`
Blobs []blobJSON `json:"blobs"`
}
type blobJSON struct {
ID restic.ID `json:"id"`
Type restic.BlobType `json:"type"`
Offset uint `json:"offset"`
Length uint `json:"length"`
UncompressedLength uint `json:"uncompressed_length,omitempty"`
}
// generatePackList returns a list of packs.
func (idx *Index) generatePackList() ([]packJSON, error) {
list := make([]packJSON, 0, len(idx.packs))
packs := make(map[restic.ID]int, len(list)) // Maps to index in list.
for typ := range idx.byType {
m := &idx.byType[typ]
m.foreach(func(e *indexEntry) bool {
packID := idx.packs[e.packIndex]
if packID.IsNull() {
panic("null pack id")
}
i, ok := packs[packID]
if !ok {
i = len(list)
list = append(list, packJSON{ID: packID})
packs[packID] = i
}
p := &list[i]
// add blob
p.Blobs = append(p.Blobs, blobJSON{
ID: e.id,
Type: restic.BlobType(typ),
Offset: uint(e.offset),
Length: uint(e.length),
UncompressedLength: uint(e.uncompressedLength),
})
return true
})
}
return list, nil
}
type jsonIndex struct {
Supersedes restic.IDs `json:"supersedes,omitempty"`
Packs []packJSON `json:"packs"`
}
// Encode writes the JSON serialization of the index to the writer w.
func (idx *Index) Encode(w io.Writer) error {
debug.Log("encoding index")
idx.m.Lock()
defer idx.m.Unlock()
list, err := idx.generatePackList()
if err != nil {
return err
}
enc := json.NewEncoder(w)
idxJSON := jsonIndex{
Supersedes: idx.supersedes,
Packs: list,
}
return enc.Encode(idxJSON)
}
// Finalize sets the index to final.
func (idx *Index) Finalize() {
debug.Log("finalizing index")
idx.m.Lock()
defer idx.m.Unlock()
idx.final = true
}
// IDs returns the IDs of the index, if available. If the index is not yet
// finalized, an error is returned.
func (idx *Index) IDs() (restic.IDs, error) {
idx.m.Lock()
defer idx.m.Unlock()
if !idx.final {
return nil, errors.New("index not finalized")
}
return idx.ids, nil
}
// SetID sets the ID the index has been written to. This requires that
// Finalize() has been called before, otherwise an error is returned.
func (idx *Index) SetID(id restic.ID) error {
idx.m.Lock()
defer idx.m.Unlock()
if !idx.final {
return errors.New("index is not final")
}
if len(idx.ids) > 0 {
return errors.New("ID already set")
}
debug.Log("ID set to %v", id)
idx.ids = append(idx.ids, id)
return nil
}
// Dump writes the pretty-printed JSON representation of the index to w.
func (idx *Index) Dump(w io.Writer) error {
debug.Log("dumping index")
idx.m.Lock()
defer idx.m.Unlock()
list, err := idx.generatePackList()
if err != nil {
return err
}
outer := jsonIndex{
Supersedes: idx.Supersedes(),
Packs: list,
}
buf, err := json.MarshalIndent(outer, "", " ")
if err != nil {
return err
}
_, err = w.Write(append(buf, '\n'))
if err != nil {
return errors.Wrap(err, "Write")
}
debug.Log("done")
return nil
}
// merge() merges indexes, i.e. idx.merge(idx2) merges the contents of idx2 into idx.
// During merging exact duplicates are removed; idx2 is not changed by this method.
func (idx *Index) merge(idx2 *Index) error {
idx.m.Lock()
defer idx.m.Unlock()
idx2.m.Lock()
defer idx2.m.Unlock()
if !idx2.final {
return errors.New("index to merge is not final")
}
packlen := len(idx.packs)
// first append packs as they might be accessed when looking for duplicates below
idx.packs = append(idx.packs, idx2.packs...)
// copy all index entries of idx2 to idx
for typ := range idx2.byType {
m2 := &idx2.byType[typ]
m := &idx.byType[typ]
// helper func to test if identical entry is contained in idx
hasIdenticalEntry := func(e2 *indexEntry) (found bool) {
m.foreachWithID(e2.id, func(e *indexEntry) {
b := idx.toPackedBlob(e, restic.BlobType(typ))
b2 := idx2.toPackedBlob(e2, restic.BlobType(typ))
if b == b2 {
found = true
}
})
return found
}
m2.foreach(func(e2 *indexEntry) bool {
if !hasIdenticalEntry(e2) {
// packIndex needs to be changed as idx2.pack was appended to idx.pack, see above
m.add(e2.id, e2.packIndex+packlen, e2.offset, e2.length, e2.uncompressedLength)
}
return true
})
}
idx.ids = append(idx.ids, idx2.ids...)
idx.supersedes = append(idx.supersedes, idx2.supersedes...)
return nil
}
// isErrOldIndex returns true if the error may be caused by an old index
// format.
func isErrOldIndex(err error) bool {
e, ok := err.(*json.UnmarshalTypeError)
return ok && e.Value == "array"
}
// DecodeIndex unserializes an index from buf.
func DecodeIndex(buf []byte, id restic.ID) (idx *Index, oldFormat bool, err error) {
debug.Log("Start decoding index")
idxJSON := &jsonIndex{}
err = json.Unmarshal(buf, idxJSON)
if err != nil {
debug.Log("Error %v", err)
if isErrOldIndex(err) {
debug.Log("index is probably old format, trying that")
idx, err = decodeOldIndex(buf)
return idx, err == nil, err
}
return nil, false, errors.Wrap(err, "DecodeIndex")
}
idx = NewIndex()
for _, pack := range idxJSON.Packs {
packID := idx.addToPacks(pack.ID)
for _, blob := range pack.Blobs {
idx.store(packID, restic.Blob{
BlobHandle: restic.BlobHandle{
Type: blob.Type,
ID: blob.ID},
Offset: blob.Offset,
Length: blob.Length,
UncompressedLength: blob.UncompressedLength,
})
}
}
idx.supersedes = idxJSON.Supersedes
idx.ids = append(idx.ids, id)
idx.final = true
debug.Log("done")
return idx, false, nil
}
// DecodeOldIndex loads and unserializes an index in the old format from rd.
func decodeOldIndex(buf []byte) (idx *Index, err error) {
debug.Log("Start decoding old index")
list := []*packJSON{}
err = json.Unmarshal(buf, &list)
if err != nil {
debug.Log("Error %#v", err)
return nil, errors.Wrap(err, "Decode")
}
idx = NewIndex()
for _, pack := range list {
packID := idx.addToPacks(pack.ID)
for _, blob := range pack.Blobs {
idx.store(packID, restic.Blob{
BlobHandle: restic.BlobHandle{
Type: blob.Type,
ID: blob.ID},
Offset: blob.Offset,
Length: blob.Length,
// no compressed length in the old index format
})
}
}
idx.final = true
debug.Log("done")
return idx, nil
}