rclone/backend/oracleobjectstorage/multipart.go
Nick Craig-Wood 2db0e23584 backends: change OpenChunkWriter interface to allow backend concurrency override
Before this change the concurrency used for an upload was rather
inconsistent.

- if size below `--backend-upload-cutoff` (default 200M) do single part upload.

- if size below `--multi-thread-cutoff` (default 256M) or using streaming
  uploads (eg `rclone rcat) do multipart upload using
  `--backend-upload-concurrency` to set the concurrency used by the uploader.

- otherwise do multipart upload using `--multi-thread-streams` to set the
  concurrency.

This change makes the default for the concurrency used be the
`--backend-upload-concurrency`. If `--multi-thread-streams` is set and larger
than the `--backend-upload-concurrency` then that will be used instead.

This means that if the user sets `--backend-upload-concurrency` then it will be
obeyed for all multipart/multi-thread transfers and the user can override them
all with `--multi-thread-streams`.

See: #7056
2023-09-03 11:47:05 +01:00

442 lines
14 KiB
Go

//go:build !plan9 && !solaris && !js
// +build !plan9,!solaris,!js
package oracleobjectstorage
import (
"context"
"crypto/md5"
"encoding/base64"
"encoding/hex"
"fmt"
"io"
"strings"
"sync"
"time"
"github.com/ncw/swift/v2"
"github.com/rclone/rclone/lib/multipart"
"github.com/rclone/rclone/lib/pool"
"golang.org/x/net/http/httpguts"
"github.com/oracle/oci-go-sdk/v65/common"
"github.com/oracle/oci-go-sdk/v65/objectstorage"
"github.com/rclone/rclone/fs"
"github.com/rclone/rclone/fs/chunksize"
"github.com/rclone/rclone/fs/hash"
)
var warnStreamUpload sync.Once
// Info needed for an upload
type uploadInfo struct {
req *objectstorage.PutObjectRequest
md5sumHex string
}
type objectChunkWriter struct {
chunkSize int64
size int64
f *Fs
bucket *string
key *string
uploadID *string
partsToCommit []objectstorage.CommitMultipartUploadPartDetails
partsToCommitMu sync.Mutex
existingParts map[int]objectstorage.MultipartUploadPartSummary
eTag string
md5sMu sync.Mutex
md5s []byte
ui uploadInfo
o *Object
}
func (o *Object) uploadMultipart(ctx context.Context, src fs.ObjectInfo, in io.Reader, options ...fs.OpenOption) error {
_, err := multipart.UploadMultipart(ctx, src, in, multipart.UploadMultipartOptions{
Open: o.fs,
OpenOptions: options,
})
return err
}
// OpenChunkWriter returns the chunk size and a ChunkWriter
//
// Pass in the remote and the src object
// You can also use options to hint at the desired chunk size
func (f *Fs) OpenChunkWriter(
ctx context.Context,
remote string,
src fs.ObjectInfo,
options ...fs.OpenOption) (info fs.ChunkWriterInfo, writer fs.ChunkWriter, err error) {
// Temporary Object under construction
o := &Object{
fs: f,
remote: remote,
}
ui, err := o.prepareUpload(ctx, src, options)
if err != nil {
return info, nil, fmt.Errorf("failed to prepare upload: %w", err)
}
uploadParts := f.opt.MaxUploadParts
if uploadParts < 1 {
uploadParts = 1
} else if uploadParts > maxUploadParts {
uploadParts = maxUploadParts
}
size := src.Size()
// calculate size of parts
chunkSize := f.opt.ChunkSize
// size can be -1 here meaning we don't know the size of the incoming file. We use ChunkSize
// buffers here (default 5 MiB). With a maximum number of parts (10,000) this will be a file of
// 48 GiB which seems like a not too unreasonable limit.
if size == -1 {
warnStreamUpload.Do(func() {
fs.Logf(f, "Streaming uploads using chunk size %v will have maximum file size of %v",
f.opt.ChunkSize, fs.SizeSuffix(int64(chunkSize)*int64(uploadParts)))
})
} else {
chunkSize = chunksize.Calculator(src, size, uploadParts, chunkSize)
}
uploadID, existingParts, err := o.createMultipartUpload(ctx, ui.req)
if err != nil {
return info, nil, fmt.Errorf("create multipart upload request failed: %w", err)
}
bucketName, bucketPath := o.split()
chunkWriter := &objectChunkWriter{
chunkSize: int64(chunkSize),
size: size,
f: f,
bucket: &bucketName,
key: &bucketPath,
uploadID: &uploadID,
existingParts: existingParts,
ui: ui,
o: o,
}
info = fs.ChunkWriterInfo{
ChunkSize: int64(chunkSize),
Concurrency: o.fs.opt.UploadConcurrency,
LeavePartsOnError: o.fs.opt.LeavePartsOnError,
}
fs.Debugf(o, "open chunk writer: started multipart upload: %v", uploadID)
return info, chunkWriter, err
}
// WriteChunk will write chunk number with reader bytes, where chunk number >= 0
func (w *objectChunkWriter) WriteChunk(ctx context.Context, chunkNumber int, reader io.ReadSeeker) (bytesWritten int64, err error) {
if chunkNumber < 0 {
err := fmt.Errorf("invalid chunk number provided: %v", chunkNumber)
return -1, err
}
// Only account after the checksum reads have been done
if do, ok := reader.(pool.DelayAccountinger); ok {
// To figure out this number, do a transfer and if the accounted size is 0 or a
// multiple of what it should be, increase or decrease this number.
do.DelayAccounting(2)
}
m := md5.New()
currentChunkSize, err := io.Copy(m, reader)
if err != nil {
return -1, err
}
// If no data read, don't write the chunk
if currentChunkSize == 0 {
return 0, nil
}
md5sumBinary := m.Sum([]byte{})
w.addMd5(&md5sumBinary, int64(chunkNumber))
md5sum := base64.StdEncoding.EncodeToString(md5sumBinary[:])
// Object storage requires 1 <= PartNumber <= 10000
ossPartNumber := chunkNumber + 1
if existing, ok := w.existingParts[ossPartNumber]; ok {
if md5sum == *existing.Md5 {
fs.Debugf(w.o, "matched uploaded part found, part num %d, skipping part, md5=%v", *existing.PartNumber, md5sum)
w.addCompletedPart(existing.PartNumber, existing.Etag)
return currentChunkSize, nil
}
}
req := objectstorage.UploadPartRequest{
NamespaceName: common.String(w.f.opt.Namespace),
BucketName: w.bucket,
ObjectName: w.key,
UploadId: w.uploadID,
UploadPartNum: common.Int(ossPartNumber),
ContentLength: common.Int64(currentChunkSize),
ContentMD5: common.String(md5sum),
}
w.o.applyPartUploadOptions(w.ui.req, &req)
var resp objectstorage.UploadPartResponse
err = w.f.pacer.Call(func() (bool, error) {
// req.UploadPartBody = io.NopCloser(bytes.NewReader(buf))
// rewind the reader on retry and after reading md5
_, err = reader.Seek(0, io.SeekStart)
if err != nil {
return false, err
}
req.UploadPartBody = io.NopCloser(reader)
resp, err = w.f.srv.UploadPart(ctx, req)
if err != nil {
if ossPartNumber <= 8 {
return shouldRetry(ctx, resp.HTTPResponse(), err)
}
// retry all chunks once have done the first few
return true, err
}
return false, err
})
if err != nil {
fs.Errorf(w.o, "multipart upload failed to upload part:%d err: %v", ossPartNumber, err)
return -1, fmt.Errorf("multipart upload failed to upload part: %w", err)
}
w.addCompletedPart(&ossPartNumber, resp.ETag)
return currentChunkSize, err
}
// add a part number and etag to the completed parts
func (w *objectChunkWriter) addCompletedPart(partNum *int, eTag *string) {
w.partsToCommitMu.Lock()
defer w.partsToCommitMu.Unlock()
w.partsToCommit = append(w.partsToCommit, objectstorage.CommitMultipartUploadPartDetails{
PartNum: partNum,
Etag: eTag,
})
}
func (w *objectChunkWriter) Close(ctx context.Context) (err error) {
req := objectstorage.CommitMultipartUploadRequest{
NamespaceName: common.String(w.f.opt.Namespace),
BucketName: w.bucket,
ObjectName: w.key,
UploadId: w.uploadID,
}
req.PartsToCommit = w.partsToCommit
var resp objectstorage.CommitMultipartUploadResponse
err = w.f.pacer.Call(func() (bool, error) {
resp, err = w.f.srv.CommitMultipartUpload(ctx, req)
// if multipart is corrupted, we will abort the uploadId
if isMultiPartUploadCorrupted(err) {
fs.Debugf(w.o, "multipart uploadId %v is corrupted, aborting...", *w.uploadID)
_ = w.Abort(ctx)
return false, err
}
return shouldRetry(ctx, resp.HTTPResponse(), err)
})
if err != nil {
return err
}
w.eTag = *resp.ETag
hashOfHashes := md5.Sum(w.md5s)
wantMultipartMd5 := fmt.Sprintf("%s-%d", base64.StdEncoding.EncodeToString(hashOfHashes[:]), len(w.partsToCommit))
gotMultipartMd5 := *resp.OpcMultipartMd5
if wantMultipartMd5 != gotMultipartMd5 {
fs.Errorf(w.o, "multipart upload corrupted: multipart md5 differ: expecting %s but got %s", wantMultipartMd5, gotMultipartMd5)
return fmt.Errorf("multipart upload corrupted: md5 differ: expecting %s but got %s", wantMultipartMd5, gotMultipartMd5)
}
fs.Debugf(w.o, "multipart upload %v md5 matched: expecting %s and got %s", *w.uploadID, wantMultipartMd5, gotMultipartMd5)
return nil
}
func isMultiPartUploadCorrupted(err error) bool {
if err == nil {
return false
}
// Check if this oci-err object, and if it is multipart commit error
if ociError, ok := err.(common.ServiceError); ok {
// If it is a timeout then we want to retry that
if ociError.GetCode() == "InvalidUploadPart" {
return true
}
}
return false
}
func (w *objectChunkWriter) Abort(ctx context.Context) error {
fs.Debugf(w.o, "Cancelling multipart upload")
err := w.o.fs.abortMultiPartUpload(
ctx,
w.bucket,
w.key,
w.uploadID)
if err != nil {
fs.Debugf(w.o, "Failed to cancel multipart upload: %v", err)
} else {
fs.Debugf(w.o, "canceled and aborted multipart upload: %v", *w.uploadID)
}
return err
}
// addMd5 adds a binary md5 to the md5 calculated so far
func (w *objectChunkWriter) addMd5(md5binary *[]byte, chunkNumber int64) {
w.md5sMu.Lock()
defer w.md5sMu.Unlock()
start := chunkNumber * md5.Size
end := start + md5.Size
if extend := end - int64(len(w.md5s)); extend > 0 {
w.md5s = append(w.md5s, make([]byte, extend)...)
}
copy(w.md5s[start:end], (*md5binary)[:])
}
func (o *Object) prepareUpload(ctx context.Context, src fs.ObjectInfo, options []fs.OpenOption) (ui uploadInfo, err error) {
bucket, bucketPath := o.split()
ui.req = &objectstorage.PutObjectRequest{
NamespaceName: common.String(o.fs.opt.Namespace),
BucketName: common.String(bucket),
ObjectName: common.String(bucketPath),
}
// Set the mtime in the metadata
modTime := src.ModTime(ctx)
// Fetch metadata if --metadata is in use
meta, err := fs.GetMetadataOptions(ctx, src, options)
if err != nil {
return ui, fmt.Errorf("failed to read metadata from source object: %w", err)
}
ui.req.OpcMeta = make(map[string]string, len(meta)+2)
// merge metadata into request and user metadata
for k, v := range meta {
pv := common.String(v)
k = strings.ToLower(k)
switch k {
case "cache-control":
ui.req.CacheControl = pv
case "content-disposition":
ui.req.ContentDisposition = pv
case "content-encoding":
ui.req.ContentEncoding = pv
case "content-language":
ui.req.ContentLanguage = pv
case "content-type":
ui.req.ContentType = pv
case "tier":
// ignore
case "mtime":
// mtime in meta overrides source ModTime
metaModTime, err := time.Parse(time.RFC3339Nano, v)
if err != nil {
fs.Debugf(o, "failed to parse metadata %s: %q: %v", k, v, err)
} else {
modTime = metaModTime
}
case "btime":
// write as metadata since we can't set it
ui.req.OpcMeta[k] = v
default:
ui.req.OpcMeta[k] = v
}
}
// Set the mtime in the metadata
ui.req.OpcMeta[metaMtime] = swift.TimeToFloatString(modTime)
// read the md5sum if available
// - for non-multipart
// - so we can add a ContentMD5
// - so we can add the md5sum in the metadata as metaMD5Hash if using SSE/SSE-C
// - for multipart provided checksums aren't disabled
// - so we can add the md5sum in the metadata as metaMD5Hash
size := src.Size()
isMultipart := size < 0 || size >= int64(o.fs.opt.UploadCutoff)
var md5sumBase64 string
if !isMultipart || !o.fs.opt.DisableChecksum {
ui.md5sumHex, err = src.Hash(ctx, hash.MD5)
if err == nil && matchMd5.MatchString(ui.md5sumHex) {
hashBytes, err := hex.DecodeString(ui.md5sumHex)
if err == nil {
md5sumBase64 = base64.StdEncoding.EncodeToString(hashBytes)
if isMultipart && !o.fs.opt.DisableChecksum {
// Set the md5sum as metadata on the object if
// - a multipart upload
// - the ETag is not an MD5, e.g. when using SSE/SSE-C
// provided checksums aren't disabled
ui.req.OpcMeta[metaMD5Hash] = md5sumBase64
}
}
}
}
// Set the content type if it isn't set already
if ui.req.ContentType == nil {
ui.req.ContentType = common.String(fs.MimeType(ctx, src))
}
if size >= 0 {
ui.req.ContentLength = common.Int64(size)
}
if md5sumBase64 != "" {
ui.req.ContentMD5 = &md5sumBase64
}
o.applyPutOptions(ui.req, options...)
useBYOKPutObject(o.fs, ui.req)
if o.fs.opt.StorageTier != "" {
storageTier, ok := objectstorage.GetMappingPutObjectStorageTierEnum(o.fs.opt.StorageTier)
if !ok {
return ui, fmt.Errorf("not a valid storage tier: %v", o.fs.opt.StorageTier)
}
ui.req.StorageTier = storageTier
}
// Check metadata keys and values are valid
for key, value := range ui.req.OpcMeta {
if !httpguts.ValidHeaderFieldName(key) {
fs.Errorf(o, "Dropping invalid metadata key %q", key)
delete(ui.req.OpcMeta, key)
} else if value == "" {
fs.Errorf(o, "Dropping nil metadata value for key %q", key)
delete(ui.req.OpcMeta, key)
} else if !httpguts.ValidHeaderFieldValue(value) {
fs.Errorf(o, "Dropping invalid metadata value %q for key %q", value, key)
delete(ui.req.OpcMeta, key)
}
}
return ui, nil
}
func (o *Object) createMultipartUpload(ctx context.Context, putReq *objectstorage.PutObjectRequest) (
uploadID string, existingParts map[int]objectstorage.MultipartUploadPartSummary, err error) {
bucketName, bucketPath := o.split()
f := o.fs
if f.opt.AttemptResumeUpload {
fs.Debugf(o, "attempting to resume upload for %v (if any)", o.remote)
resumeUploads, err := o.fs.findLatestMultipartUpload(ctx, bucketName, bucketPath)
if err == nil && len(resumeUploads) > 0 {
uploadID = *resumeUploads[0].UploadId
existingParts, err = f.listMultipartUploadParts(ctx, bucketName, bucketPath, uploadID)
if err == nil {
fs.Debugf(o, "resuming with existing upload id: %v", uploadID)
return uploadID, existingParts, err
}
}
}
req := objectstorage.CreateMultipartUploadRequest{
NamespaceName: common.String(o.fs.opt.Namespace),
BucketName: common.String(bucketName),
}
req.Object = common.String(bucketPath)
if o.fs.opt.StorageTier != "" {
storageTier, ok := objectstorage.GetMappingStorageTierEnum(o.fs.opt.StorageTier)
if !ok {
return "", nil, fmt.Errorf("not a valid storage tier: %v", o.fs.opt.StorageTier)
}
req.StorageTier = storageTier
}
o.applyMultipartUploadOptions(putReq, &req)
var resp objectstorage.CreateMultipartUploadResponse
err = o.fs.pacer.Call(func() (bool, error) {
resp, err = o.fs.srv.CreateMultipartUpload(ctx, req)
return shouldRetry(ctx, resp.HTTPResponse(), err)
})
if err != nil {
return "", existingParts, err
}
existingParts = make(map[int]objectstorage.MultipartUploadPartSummary)
uploadID = *resp.UploadId
fs.Debugf(o, "created new upload id: %v", uploadID)
return uploadID, existingParts, err
}