rclone/backend/b2/upload.go
Nick Craig-Wood ab803d1278 b2: implement OpenChunkWriter and multi-thread uploads #7056
This implements the OpenChunkWriter interface for b2 which
enables multi-thread uploads.

This makes the memory controls of the s3 backend inoperative; they are
replaced with the global ones.

    --b2-memory-pool-flush-time
    --b2-memory-pool-use-mmap

By using the buffered reader this fixes excessive memory use when
uploading large files as it will share memory pages between all
readers.
2023-08-24 12:39:27 +01:00

472 lines
14 KiB
Go

// Upload large files for b2
//
// Docs - https://www.backblaze.com/b2/docs/large_files.html
package b2
import (
"context"
"crypto/sha1"
"encoding/hex"
"fmt"
gohash "hash"
"io"
"strings"
"sync"
"github.com/rclone/rclone/backend/b2/api"
"github.com/rclone/rclone/fs"
"github.com/rclone/rclone/fs/accounting"
"github.com/rclone/rclone/fs/chunksize"
"github.com/rclone/rclone/fs/hash"
"github.com/rclone/rclone/lib/atexit"
"github.com/rclone/rclone/lib/pool"
"github.com/rclone/rclone/lib/rest"
"golang.org/x/sync/errgroup"
)
type hashAppendingReader struct {
h gohash.Hash
in io.Reader
hexSum string
hexReader io.Reader
}
// Read returns bytes all bytes from the original reader, then the hex sum
// of what was read so far, then EOF.
func (har *hashAppendingReader) Read(b []byte) (int, error) {
if har.hexReader == nil {
n, err := har.in.Read(b)
if err == io.EOF {
har.in = nil // allow GC
err = nil // allow reading hexSum before EOF
har.hexSum = hex.EncodeToString(har.h.Sum(nil))
har.hexReader = strings.NewReader(har.hexSum)
}
return n, err
}
return har.hexReader.Read(b)
}
// AdditionalLength returns how many bytes the appended hex sum will take up.
func (har *hashAppendingReader) AdditionalLength() int {
return hex.EncodedLen(har.h.Size())
}
// HexSum returns the hash sum as hex. It's only available after the original
// reader has EOF'd. It's an empty string before that.
func (har *hashAppendingReader) HexSum() string {
return har.hexSum
}
// newHashAppendingReader takes a Reader and a Hash and will append the hex sum
// after the original reader reaches EOF. The increased size depends on the
// given hash, which may be queried through AdditionalLength()
func newHashAppendingReader(in io.Reader, h gohash.Hash) *hashAppendingReader {
withHash := io.TeeReader(in, h)
return &hashAppendingReader{h: h, in: withHash}
}
// largeUpload is used to control the upload of large files which need chunking
type largeUpload struct {
f *Fs // parent Fs
o *Object // object being uploaded
doCopy bool // doing copy rather than upload
what string // text name of operation for logs
in io.Reader // read the data from here
wrap accounting.WrapFn // account parts being transferred
id string // ID of the file being uploaded
size int64 // total size
parts int // calculated number of parts, if known
sha1smu sync.Mutex // mutex to protect sha1s
sha1s []string // slice of SHA1s for each part
uploadMu sync.Mutex // lock for upload variable
uploads []*api.GetUploadPartURLResponse // result of get upload URL calls
chunkSize int64 // chunk size to use
src *Object // if copying, object we are reading from
}
// newLargeUpload starts an upload of object o from in with metadata in src
//
// If newInfo is set then metadata from that will be used instead of reading it from src
func (f *Fs) newLargeUpload(ctx context.Context, o *Object, in io.Reader, src fs.ObjectInfo, defaultChunkSize fs.SizeSuffix, doCopy bool, newInfo *api.File) (up *largeUpload, err error) {
size := src.Size()
parts := 0
chunkSize := defaultChunkSize
if size == -1 {
fs.Debugf(o, "Streaming upload with --b2-chunk-size %s allows uploads of up to %s and will fail only when that limit is reached.", f.opt.ChunkSize, maxParts*f.opt.ChunkSize)
} else {
chunkSize = chunksize.Calculator(o, size, maxParts, defaultChunkSize)
parts = int(size / int64(chunkSize))
if size%int64(chunkSize) != 0 {
parts++
}
}
opts := rest.Opts{
Method: "POST",
Path: "/b2_start_large_file",
}
bucket, bucketPath := o.split()
bucketID, err := f.getBucketID(ctx, bucket)
if err != nil {
return nil, err
}
var request = api.StartLargeFileRequest{
BucketID: bucketID,
Name: f.opt.Enc.FromStandardPath(bucketPath),
}
if newInfo == nil {
modTime := src.ModTime(ctx)
request.ContentType = fs.MimeType(ctx, src)
request.Info = map[string]string{
timeKey: timeString(modTime),
}
// Set the SHA1 if known
if !o.fs.opt.DisableCheckSum || doCopy {
if calculatedSha1, err := src.Hash(ctx, hash.SHA1); err == nil && calculatedSha1 != "" {
request.Info[sha1Key] = calculatedSha1
}
}
} else {
request.ContentType = newInfo.ContentType
request.Info = newInfo.Info
}
var response api.StartLargeFileResponse
err = f.pacer.Call(func() (bool, error) {
resp, err := f.srv.CallJSON(ctx, &opts, &request, &response)
return f.shouldRetry(ctx, resp, err)
})
if err != nil {
return nil, err
}
up = &largeUpload{
f: f,
o: o,
doCopy: doCopy,
what: "upload",
id: response.ID,
size: size,
parts: parts,
sha1s: make([]string, 0, 16),
chunkSize: int64(chunkSize),
}
// unwrap the accounting from the input, we use wrap to put it
// back on after the buffering
if doCopy {
up.what = "copy"
up.src = src.(*Object)
} else {
up.in, up.wrap = accounting.UnWrap(in)
}
return up, nil
}
// getUploadURL returns the upload info with the UploadURL and the AuthorizationToken
//
// This should be returned with returnUploadURL when finished
func (up *largeUpload) getUploadURL(ctx context.Context) (upload *api.GetUploadPartURLResponse, err error) {
up.uploadMu.Lock()
defer up.uploadMu.Unlock()
if len(up.uploads) == 0 {
opts := rest.Opts{
Method: "POST",
Path: "/b2_get_upload_part_url",
}
var request = api.GetUploadPartURLRequest{
ID: up.id,
}
err := up.f.pacer.Call(func() (bool, error) {
resp, err := up.f.srv.CallJSON(ctx, &opts, &request, &upload)
return up.f.shouldRetry(ctx, resp, err)
})
if err != nil {
return nil, fmt.Errorf("failed to get upload URL: %w", err)
}
} else {
upload, up.uploads = up.uploads[0], up.uploads[1:]
}
return upload, nil
}
// returnUploadURL returns the UploadURL to the cache
func (up *largeUpload) returnUploadURL(upload *api.GetUploadPartURLResponse) {
if upload == nil {
return
}
up.uploadMu.Lock()
up.uploads = append(up.uploads, upload)
up.uploadMu.Unlock()
}
// Add an sha1 to the being built up sha1s
func (up *largeUpload) addSha1(chunkNumber int, sha1 string) {
up.sha1smu.Lock()
defer up.sha1smu.Unlock()
if len(up.sha1s) < chunkNumber+1 {
up.sha1s = append(up.sha1s, make([]string, chunkNumber+1-len(up.sha1s))...)
}
up.sha1s[chunkNumber] = sha1
}
// WriteChunk will write chunk number with reader bytes, where chunk number >= 0
func (up *largeUpload) WriteChunk(ctx context.Context, chunkNumber int, reader io.ReadSeeker) (size int64, err error) {
err = up.f.pacer.Call(func() (bool, error) {
// Discover the size by seeking to the end
size, err = reader.Seek(0, io.SeekEnd)
if err != nil {
return false, err
}
// rewind the reader on retry and after reading size
_, err = reader.Seek(0, io.SeekStart)
if err != nil {
return false, err
}
fs.Debugf(up.o, "Sending chunk %d length %d", chunkNumber, size)
// Get upload URL
upload, err := up.getUploadURL(ctx)
if err != nil {
return false, err
}
in := newHashAppendingReader(reader, sha1.New())
size += int64(in.AdditionalLength())
// Authorization
//
// An upload authorization token, from b2_get_upload_part_url.
//
// X-Bz-Part-Number
//
// A number from 1 to 10000. The parts uploaded for one file
// must have contiguous numbers, starting with 1.
//
// Content-Length
//
// The number of bytes in the file being uploaded. Note that
// this header is required; you cannot leave it out and just
// use chunked encoding. The minimum size of every part but
// the last one is 100 MB (100,000,000 bytes)
//
// X-Bz-Content-Sha1
//
// The SHA1 checksum of the this part of the file. B2 will
// check this when the part is uploaded, to make sure that the
// data arrived correctly. The same SHA1 checksum must be
// passed to b2_finish_large_file.
opts := rest.Opts{
Method: "POST",
RootURL: upload.UploadURL,
Body: up.wrap(in),
ExtraHeaders: map[string]string{
"Authorization": upload.AuthorizationToken,
"X-Bz-Part-Number": fmt.Sprintf("%d", chunkNumber+1),
sha1Header: "hex_digits_at_end",
},
ContentLength: &size,
}
var response api.UploadPartResponse
resp, err := up.f.srv.CallJSON(ctx, &opts, nil, &response)
retry, err := up.f.shouldRetry(ctx, resp, err)
if err != nil {
fs.Debugf(up.o, "Error sending chunk %d (retry=%v): %v: %#v", chunkNumber, retry, err, err)
}
// On retryable error clear PartUploadURL
if retry {
fs.Debugf(up.o, "Clearing part upload URL because of error: %v", err)
upload = nil
}
up.returnUploadURL(upload)
up.addSha1(chunkNumber, in.HexSum())
return retry, err
})
if err != nil {
fs.Debugf(up.o, "Error sending chunk %d: %v", chunkNumber, err)
} else {
fs.Debugf(up.o, "Done sending chunk %d", chunkNumber)
}
return size, err
}
// Copy a chunk
func (up *largeUpload) copyChunk(ctx context.Context, part int, partSize int64) error {
err := up.f.pacer.Call(func() (bool, error) {
fs.Debugf(up.o, "Copying chunk %d length %d", part, partSize)
opts := rest.Opts{
Method: "POST",
Path: "/b2_copy_part",
}
offset := int64(part) * up.chunkSize // where we are in the source file
var request = api.CopyPartRequest{
SourceID: up.src.id,
LargeFileID: up.id,
PartNumber: int64(part + 1),
Range: fmt.Sprintf("bytes=%d-%d", offset, offset+partSize-1),
}
var response api.UploadPartResponse
resp, err := up.f.srv.CallJSON(ctx, &opts, &request, &response)
retry, err := up.f.shouldRetry(ctx, resp, err)
if err != nil {
fs.Debugf(up.o, "Error copying chunk %d (retry=%v): %v: %#v", part, retry, err, err)
}
up.addSha1(part, response.SHA1)
return retry, err
})
if err != nil {
fs.Debugf(up.o, "Error copying chunk %d: %v", part, err)
} else {
fs.Debugf(up.o, "Done copying chunk %d", part)
}
return err
}
// Close closes off the large upload
func (up *largeUpload) Close(ctx context.Context) error {
fs.Debugf(up.o, "Finishing large file %s with %d parts", up.what, up.parts)
opts := rest.Opts{
Method: "POST",
Path: "/b2_finish_large_file",
}
var request = api.FinishLargeFileRequest{
ID: up.id,
SHA1s: up.sha1s,
}
var response api.FileInfo
err := up.f.pacer.Call(func() (bool, error) {
resp, err := up.f.srv.CallJSON(ctx, &opts, &request, &response)
return up.f.shouldRetry(ctx, resp, err)
})
if err != nil {
return err
}
return up.o.decodeMetaDataFileInfo(&response)
}
// Abort aborts the large upload
func (up *largeUpload) Abort(ctx context.Context) error {
fs.Debugf(up.o, "Cancelling large file %s", up.what)
opts := rest.Opts{
Method: "POST",
Path: "/b2_cancel_large_file",
}
var request = api.CancelLargeFileRequest{
ID: up.id,
}
var response api.CancelLargeFileResponse
err := up.f.pacer.Call(func() (bool, error) {
resp, err := up.f.srv.CallJSON(ctx, &opts, &request, &response)
return up.f.shouldRetry(ctx, resp, err)
})
if err != nil {
fs.Errorf(up.o, "Failed to cancel large file %s: %v", up.what, err)
}
return err
}
// Stream uploads the chunks from the input, starting with a required initial
// chunk. Assumes the file size is unknown and will upload until the input
// reaches EOF.
//
// Note that initialUploadBlock must be returned to f.putBuf()
func (up *largeUpload) Stream(ctx context.Context, initialUploadBlock *pool.RW) (err error) {
defer atexit.OnError(&err, func() { _ = up.Abort(ctx) })()
fs.Debugf(up.o, "Starting streaming of large file (id %q)", up.id)
var (
g, gCtx = errgroup.WithContext(ctx)
hasMoreParts = true
)
up.size = initialUploadBlock.Size()
for part := 0; hasMoreParts; part++ {
// Get a block of memory from the pool and token which limits concurrency.
var rw *pool.RW
if part == 1 {
rw = initialUploadBlock
} else {
rw = up.f.getRW(false)
}
// Fail fast, in case an errgroup managed function returns an error
// gCtx is cancelled. There is no point in uploading all the other parts.
if gCtx.Err() != nil {
up.f.putRW(rw)
break
}
// Read the chunk
var n int64
if part == 1 {
n = rw.Size()
} else {
n, err = io.CopyN(rw, up.in, up.chunkSize)
if err == io.EOF {
fs.Debugf(up.o, "Read less than a full chunk, making this the last one.")
hasMoreParts = false
} else if err != nil {
// other kinds of errors indicate failure
up.f.putRW(rw)
return err
}
}
// Keep stats up to date
up.parts = part
up.size += n
if part > maxParts {
up.f.putRW(rw)
return fmt.Errorf("%q too big (%d bytes so far) makes too many parts %d > %d - increase --b2-chunk-size", up.o, up.size, up.parts, maxParts)
}
part := part // for the closure
g.Go(func() (err error) {
defer up.f.putRW(rw)
_, err = up.WriteChunk(gCtx, part, rw)
return err
})
}
err = g.Wait()
if err != nil {
return err
}
return up.Close(ctx)
}
// Copy the chunks from the source to the destination
func (up *largeUpload) Copy(ctx context.Context) (err error) {
defer atexit.OnError(&err, func() { _ = up.Abort(ctx) })()
fs.Debugf(up.o, "Starting %s of large file in %d chunks (id %q)", up.what, up.parts, up.id)
var (
g, gCtx = errgroup.WithContext(ctx)
remaining = up.size
)
g.SetLimit(up.f.opt.UploadConcurrency)
for part := 0; part <= up.parts; part++ {
// Fail fast, in case an errgroup managed function returns an error
// gCtx is cancelled. There is no point in copying all the other parts.
if gCtx.Err() != nil {
break
}
reqSize := remaining
if reqSize >= up.chunkSize {
reqSize = up.chunkSize
}
part := part // for the closure
g.Go(func() (err error) {
return up.copyChunk(gCtx, part, reqSize)
})
remaining -= reqSize
}
err = g.Wait()
if err != nil {
return err
}
return up.Close(ctx)
}